Thoughts on Investing and Starting Up

Forget the Personalities. Do the Math.

The metrics we inherited from SaaS were built on two assumptions that AI doesn't have.

This is the strangest investing environment I’ve worked in.

Pre-seed rounds priced at $50M on a couple resumes, but $5M of revenue getting nothing. Funds that had never touched Israel writing checks here with complete disregard for valuation. Companies raising at a billion eight months after a round nobody could justify. Capital pouring in at a pace that makes the 2021 cycle look measured.

A lot of that is personality driven. It’s momentum, it’s fear of missing the one that matters, it’s a handful of loud people setting the price for everyone else. That’s a real story, but it’s not today’s.

Today I want to ignore the personalities entirely and look at the math, because the math is telling us what isn’t getting priced in… yet.

Back in June I wrote that I was hunting for the function underneath AI business models. The SaaS era had one. Calculus: ARR is the derivative of recognized revenue, revenue is the accumulation, and once you see that, the entire business model unlocks. I said I’d go back to my engineering textbooks looking for the equivalent, and that I’d share it when it was clear.

It’s pretty clear. I recorded a full episode on it, and this is the short version.

Start with what SaaS actually had. Two things:

• The first is that revenue accumulates and stays. Contractual, annual, paid up front. Your customers are effectively lending you money. Once the stream turns on it stays on.

• The second is that gross margins are consistent and high. Not as high as the marketing decks claimed, since a lot of what should be COGS got buried in R&D, but high enough that the variable cost of serving one more customer was close to a rounding error.

Put those together and something remarkable happens. Growth cures every problem. A company could go public at negative 150% operating margins and nobody blinked, because everyone understood the accumulation would catch up. I looked at plenty of these businesses, sliced the cohorts, found real inefficiencies, and was wrong to worry. They grew out of it.

Every metric we trust was built on top of those two assumptions. Net dollar retention, SaaS magic number, LTV:CAC. All of them are hacks that only work because accumulation and high margins were the foundation underneath.

AI has neither.

Switching costs are gone, and so is guaranteed accumulation. A customer can move every workload from one model to another in a matter of minutes, and the routers now do it automatically on price and performance. Calling that recurring revenue is borderline egregious.

And gross margin is no longer an assumption you can wave past. It’s variable, it’s not high, and it moves under your feet every time a new frontier model ships.

Take those two assumptions away and the inherited metrics aren’t leading indicators anymore. They’re a house of cards.

When margins are high and revenue accumulates, growth cures problems. When margins are variable and revenue isn’t contractually locked, growth reveals them. Same chart pointing up and to the right. Multiply it by a negative gross margin number and the faster it grows, the more you lose.

This isn’t the first time that a company mistakenly used SaaS metrics, only to come tumbling down when the lack of contractually recurring revenue and high gross margins revealed themselves. Remember when WeWork was masqueraded as a recurring revenue business? Sure, rent recurs, but margins are not high and the liabilities are very real. This creates an entirely different set of sensitivities and weak points that the SaaS metrics miss. AI is experiencing a similar phenomenon. People are grabbing for what worked for them in the past without evaluating the fundamental underpinnings of why it worked.

So what do you measure instead?

The answer came from a mechanical engineering class I took seventeen years ago, ME161. It’s basically about what happens when a car hits a speed bump. A company absorbing a new model release is the same sort of dynamic system. There’s a ratio that describes how well a system handles a shock like that, and engineers have been using it for a century. It’s called Zeta (ζ), the damping ratio.

• Push everything to the newest model the week it ships and your margin craters and rings (underdamped)

• Ignore the new models entirely and your margin holds while your product gets eaten alive (overdamped)

Neither is the answer. The answer is the middle, and unlike the SaaS benchmarks we all memorized, the target isn’t something somebody observed across a comp set. It’s mathematically derived. Critically damped is the mathematically correct place to be. It’s the absolute goal for every company.

This changes the diligence questions.

For years the early-stage version of this was some flavor of “show me the efficiency metrics.” That doesn’t work when a company is nine months old and has absorbed two model releases - regardless of the revenue scale. There’s no trace to read yet. This is not dissimilar to the irrelevance of SaaS Magic number at $50k of ARR.

So the question I’m asking founders now is not which model they’re running on. That answer expires in six weeks. It’s what their method is for deciding. How fast do they figure out what a new model is actually good at, what moves down to cheaper inference, and what gets built on the frontier that wasn’t possible last quarter. The method is the durable asset. The model choice is a snapshot.

The companies that have a real answer to that are demonstrating something. The ones who tell you they’re on the newest model because it’s the best one are telling you they’re underdamped and don’t know it.

A company is an organization who takes advantage of the best new technologies to efficiently deliver value to their market. There’s a lot of ingredients buried in there, but we all intuitively understand that the most valuable companies in the world are the ones who have figured out how to do this consistently across different waves of technology and market environments.

I go through the whole thing on the new episode: why SaaS worked, why AI breaks it, and how to actually calculate zeta from data you already have. Two streams, revenue and cost, plotted against every model release.

Stay tuned for more and subscribe to for more in-depth looks at business models.

Portfolio Highlights

TerrainPlot is founded by our very own former principal, Vance Farouk Grissom, along with Brandon Hills and Antonio Ferris. TerrainPlot turns survey records into CAD geometry for land surveyors. Boundary work still starts with someone redrawing a pile of documents by hand before the real work begins. TerrainPlot takes the whole record, a single deed or a 200 page title package, scanned or cursive or typed, finds every legal description and plat inside it, and draws each as real geometry rather than a traced image. Aliquot parts get georeferenced against the actual PLSS section grid instead of floating until someone places them. Every line links back to its source, the software flags where it's uncertain and shows closure, and when a shape doesn't close it hunts for flipped bearings and transposed digits. Surveyors at firms including Bohler Engineering and D2 Surveying are using it, exporting DXF to their own CAD standards.

We're excited to have invested in Josh Greene and Tim Moreton in Otis, which builds product intelligence for AI-native teams. Analytics tell you what ran and evals tell you the model passed a test, but none of it tells you whether the user actually got value. Otis listens to the conversations users have with a product, combines them with real behavior, and clusters the result into things a team can act on: the same correction users make on drafted output over and over, the feature dozens of accounts ask the copilot for and never file a ticket about, the moment users quietly stop trusting an agent and take the work back by hand. It installs through a coding agent, runs async so there's no added latency, and redacts PII before anything hits disk. The signal that never looked like signal, in the one place nobody was reading.

Very True Podcast

We’re open to any questions to cover in future episodes, just fill out the Q&A form.

Remember to subscribe via Spotify or Apple Podcasts.

Who we are

Verissimo Ventures is an early-stage Venture Fund based in Israel and the US. We take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our third fund.

Thoughts on Investing and Starting Up

The World Is Flat... Again

And just like last time, it’s not so simple.

Twenty years ago, Thomas Friedman told us the world was flat. He was early. The internet flattened communication - information could move anywhere, instantly - but it never flattened the ability to build. So the playing field leveled, and then, paradoxically, the winners concentrated harder than ever, on a force Friedman’s metaphor didn’t account for: network effects.

Now the real flattening is arriving. Intelligence, delivered through that same internet, globally and instantly - and suddenly anyone can build almost anything. Here’s the whole argument up front: what goes flat is the ability to build. What does not go flat is who wins. The spoils will concentrate again, likely into bigger spikes than last time - and the force that makes them won’t be network effects. It’s something harder to see, and it’s the most human thing in the whole stack.

What actually happened last time

Friedman’s flat world was real, as far as it went. Coordination costs collapsed. Capability democratized. A developer in Bangalore or Tel Aviv could sell to the world. The tide genuinely rose: the median participant in the global economy got better off, and tools once reserved for giants became nearly universal.

But a rising tide only lifts things that float. It also drowns what can't move with it - and it produced the most concentrated winners in economic history. The same connectivity that leveled access built network effects and data advantages that compounded without limit. The field got flatter, and the outliers got taller. The median rose; the average rose faster, because the average is dragged up by a handful of trillion-dollar companies a “flat” world was supposed to make impossible. He was half right. And the half he got wrong is the half that made people money. A lot of money.

And here’s the part everyone forgets: the giants of the old world were supposed to win the new one. GE, Xerox, AT&T, Siemens - they owned pieces of the actual infrastructure the internet would run on. They had the capital, the labs, the distribution, the position. Xerox PARC literally invented much of the future, and Xerox still couldn’t ship it. They lost anyway - partly because someone else’s creativity proved too much for them, and partly because corporate inertia is a real force: the stronger your position in the old era, the harder it is to bet against the business model that built it.

So the real lesson was never “advantage disappears.” It was advantage relocates. The old moats (geography, distribution, physical scale) got bulldozed. New moats (networks, data, switching costs) got steeper than anything before. Flat where you inherited. Steep where you earned. And position going into a flattening has never once predicted who owns the spikes coming out of it.

The next flattening

Be precise about what AI flattens: the cost and time to replicate. What took a strong team two years to build can now be assembled in a weekend on top of a frontier model. The half-life of a product moat is crashing toward zero. Whatever advantage you built into the product stops protecting you almost immediately. Your customers can increasingly rebuild what you sell them, cheaper. Your competitors are one model release away from your feature set.

The internet wasn’t enough to flatten building. It moved information; it couldn’t move capability. Intelligence over the internet moves capability - everywhere, at once. This is the flattener Friedman reached for, twenty years too early.

And both fashionable readings of it repeat his mistake. The pessimist says everything commoditizes: race to zero, invest in nothing. The optimist says it’s a land grab: grow at any cost, margins are tomorrow’s problem, and negative gross margin is actually good because it means usage (a claim with a silent condition attached: only if you can control it). Both mistake flat for uniform. The tide will rise again: the median AI-powered company will be built faster and serve customers better than the median SaaS company ever did. But the spikes are coming too. Possibly the tallest we’ve ever seen.

Watch who understands this. The smartest AI companies are already realizing that laying the pipes is not enough, that shipping intelligence, like laying fiber, is the flattening force, not the high ground. You cannot ride the previous era’s business model into this one; the GEs of the world tried exactly that. You have to push forward and unlock a new kind of advantage - the way the internet took network effects, a force as old as the telephone, and paid them out at a scale no railroad ever saw.

What a moat even is

A moat was never a feature. It was never being first. Those are things you inherit from a moment in time, and inherited advantages are exactly what the rising tide drowns.

A moat is whatever your competitor cannot replicate by writing a check or a prompt. Last era, that meant networks and data. This era, look where durable positions are actually forming: not in the models, those are the flattening force, not the high ground, but in the application layer wrapped around them. AWS’s moat was never compute; it’s the hundreds of services your product is built on. Google’s is not an algorithm; it’s Gmail, Search, YouTube - surfaces you’d have to pry out of daily life. Workflow embeddedness, proprietary data that accrues with every use, trust at the edge, switching costs your customer feels. Same rules as last time: flat where you inherited; steep where you earn - and you earn it every single day now, because the terrain no longer holds it for you. Yesterday’s novelty is tomorrow’s mundane commodity.

The part that shows up in the numbers

Strip valuation to one sentence: a business is worth the profit it can durably throw off. That’s it. That’s the whole religion of investing.

SaaS let everyone forget this, because durability came built in: 85% gross margins and predictable retention meant revenue growth was future profit, so we priced on growth and skipped the rest. Those assumptions are gone. In AI-powered companies, margins move (sometimes violently) with every model release, and retention is contestable in a world where customers can rebuild you.

Gross margin, it turns out, is margin for error. SaaS had so much that companies could be sloppy and thrive; the margin absorbed the mistakes. The phrase “move fast and break things” literally defined the era. AI companies have a fraction of the cushion, which means execution isn’t one input anymore, it’s the whole game. In SaaS, growth hid your flaws. In many cases it cured them. In AI, growth reveals them because every new customer now scales your costs too, and stresses every function at once.

So what makes the spikes?

A company is just a group of people organized to turn effort into value customers will pay for. Everyone calls the soft version of that "culture," but culture isn't the posters on the wall. It's whether the organization can absorb a shock and keep converting it to value, release after release. That's the thing that makes the spikes. And like the last flattening's spike-maker, it's an old force about to be paid out at a scale it never could before.

With the internet, we eventually learned the answer: network effects - compounding, self-reinforcing, and visible enough to name in advance. But notice: network effects weren’t new. The telephone had them, the railroads had them, credit cards had them. The internet didn’t invent the dynamic, it stripped away the friction that had always capped it, and a force that used to top out at national scale suddenly compounded worldwide. Every flattening works this way: it takes an old advantage and pays it out at a scale it never could before.

This time the force is harder to see, because it isn’t an asset a company holds. It’s two human things (old ones) that AI is about to uncap.

• Creativity: the genuinely inverted take that flips how people expect to behave. The best products in the world were never faster copies; they were reframings, and only a person operating at full imaginative stretch produces those.

  • BlackBerry answered ‘how do we build a better keyboard?’ Apple asked ‘why is there a keyboard at all?’

• Adaptation speed: the discipline to convert each new wave of intelligence into something durable before the next wave lands and yesterday’s innovation becomes a commodity.

  • When the world moved to cloud, Adobe aggressively rebuilt and repriced their products, actually shrinking their revenue in the near-term, but 10x-ing their market cap over the following decade.

One decides how tall the spike goes. The other decides whether it endures.

We don't have a name for it yet. But it shows up in how a company behaves under pressure - and the part that shows up in the numbers, you can measure. That's what Business Model Logic is about.

Subscribe to Business Model Logic

Very True Podcast

We’re open to any questions to cover in future episodes, just fill out the Q&A form.

Remember to subscribe via Spotify or Apple Podcasts.

Portfolio Highlights

Alta builds an AI workforce for sales and revenue teams, it closed a $25 million Series A. Instead of one general-purpose assistant, Alta deploys specialized agents with actual job descriptions: Katie handles personalized outreach, Luna digs into analytics and insights, and Alex runs automated calling. The agents plug directly into the tools teams already use across sales, marketing, billing, and collaboration, taking over the routine work so reps can focus on selling. The result is a revenue org that moves faster and makes decisions backed by data rather than gut feel.

Aligned announced a $60M Series B led by PeakSpan Capital. They give B2B sales teams a shared workspace where buyers and sellers can collaborate on deals in one place, replacing the chaos of scattered emails, links, and attachments. Its AI agents surface hidden stakeholders, flag deal risks, and help equip internal champions with the content they need to build a case with leadership. More than 70,000 sellers and 1 million buyers now use Aligned every month to run deals end to end, from first meeting to close. The company says ARR has tripled over the past twelve months, and the new capital will go toward building out a fully agentic deal-execution layer and expanding engineering.

MotorMia builds an AI copilot for car enthusiasts planning vehicle modifications. Its AI, Mia, analyzes a user's specific car, down to the year, make, model, and current setup, then recommends personalized upgrades across six core systems: engine, transmission, brakes, suspension, cooling, and exhaust. Rather than leaving builders to guess or dig through scattered forum threads, Mia verifies that a given part will actually fit and work on their car, and projects the real-world performance gains before they buy anything. Users can also browse real builds from the community to see how similar upgrades played out on comparable vehicles, turning what's traditionally a trial-and-error hobby into a guided, data-backed process.

Who we are

Verissimo Ventures is an early-stage Venture Fund based in Israel and the US. We take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our third fund.

Thoughts on Investing and Starting Up

This month is about origins and transitions. Mine begins picking stocks with my dad’s help at age 11. Then internships, a mechanical engineering degree, and a job at Morgan Stanley tech banking in Menlo Park, where I got to work on game-changing deals like the Facebook IPO. Then Washington DC, to join NEA in 2013. That was the first time my entire background coalesced around a single approach to business and finance.

And it landed at a genuinely unique moment in technology history: the arrival of the SaaS business model. When I was at Morgan Stanley from 2011 to 2013, half the software companies we covered were still running the perpetual license model for all or part of their customers. By the time I was a year into NEA, it was all SaaS. The big drama of the moment was watching Adobe cut its recognized revenue to transition to a full subscription model. Novel, gutsy, and ultimately very successful. But already a relic of the past. The transition era ran from about 2010 to 2015. After that it was all SaaS all the time, until the word became essentially synonymous with software. And that’s the real win, when people forget a transition ever happened and it just becomes the new reality. I was fortunate to have a front-row seat to that transition. And I have another front-row seat now, to the one standing in front of us: the transition to AI.

I showed up at NEA in 2013, a 24-year-old mechanical engineer who had spent two years at Morgan Stanley. Harry Weller was genuinely confused when I said I wanted to work on enterprise software companies. It sounds about as crazy today as it did 13 years ago, but for completely different reasons. But Harry quickly clocked me as an old-school crossover nerd, finance and engineering, and not just that I’d done both, but that I nerded out on both. From that moment on, I was his SaaS business model guy. He handed me a bunch of NEA SaaS portfolio board decks and the latest research from David Skok and Joel York on the model. And most importantly, he started sending me to sit with executive teams outside of board meetings to dig deep on their business models. His ask was simple: learn it.

Harry had a way of knowing things before other people. People love to throw around the word “calculus” when they really just mean calculations, but this time it was a premonition (undoubtedly coming from his physics background). He just hadn’t put his finger on it. He knew I would.

This newly invented metric called ARR (annualized recurring revenue), the thing that actually captured the growth trajectory of a subscription business, had a calculus relationship with the accounting metric we all know and love: recognized revenue.

ARR is the derivative of recognized revenue.
It took me a few months, but I found it in mid-2014.

Here’s the cleanest way I’ve found to explain it. ARR is to recognized revenue what miles per hour is to distance. ARR is your speed at this instant. Recognized revenue is the ground you actually cover. Hold $10M of ARR for a year and you recognize $10M, the same way 60 miles per hour for an hour gets you 60 miles. Accelerate, and the distance compounds. I’ve been teaching it off that same slide for a decade.

Flipping to the financial side, cumulative revenue is the integral of ARR. Once you see it that way, you can’t unsee it.

When you apply it to an MRR chart, the math comes to life in the nerdiest way possible:

Everything I built afterward flowed from this singular idea. My first real piece of work at the firm was a deck I called SaaS 101. Then SaaS 202, which went deep: Salesforce as the native-SaaS archetype, Adobe as the case study in transition, the full ARR build, the magic number, the churn and cohort mechanics. Then a SaaS model template I’m still proud of, because it went on to help dozens of companies understand their own model and raise their rounds. As my friend explained, it gave founders the controls in the cockpit that they needed to drive their businesses.

By 2014 I was, immodestly, the SaaS business model expert at NEA. As a first-year associate. Partners on the other coast I had no formal working relationship with were asking me to visit their portfolio companies. My banking-style weekly SaaS comps package made its way around the entire firm and out to a lot of the founders. By 2018 saying any of this out loud sounded absurd, because by then everyone was a SaaS expert. Honestly, if you weren’t, you had no business being in venture. And that’s fine. That’s actually the tell. When an insight becomes obvious to everyone, it usually means it was done well. SaaS 101 stuck around as firm lore long after I left, which is the nicest thing I can say about my own work.

Over time I pivoted it from SaaS metrics to SaaS unit economics, then to unit economics, period. All of it collapsing back to the same place. The DCF. Not case by case. The fundamentals. The things that always matter and are always true.

Here’s why I’m telling you this now.

That whole era, the one my decks described, was gold rush for one specific reason. The SaaS business model optimized the DCF. Predictable recurring revenue, unbounded growth, fat margins, expanding addressable markets. Those are the four variables that drive a DCF, and the subscription-based enterprise software model nails all four.

SaaS turned good old boring software into the best discountable cash flow stream anyone had ever seen, and the market paid up accordingly. The math was clean, and clean math compounds.

And then it wasn’t.

If you understood what made SaaS work at a fundamental level, the turn was easy to see coming. A gold rush means more competition, which means less pricing leverage, which means lower gross margins, and higher CAC to compete for customer attention. And it means relentless investment just to keep the products relevant. The exponential venture boom made its predictable transition into a linear private equity grind.

We are at the end of one era and the start of another. Like the last transition, some of what comes next will be obvious extrapolation, and some will be genuinely creative new applications of a powerful new technology. And when everything is stripped away, the business model is the single source of truth.

AI is the next thing, and the next thing is murky. Exciting, but nebulous. Growth is faster than anything SaaS ever produced, which is itself a natural extrapolation of what multi-tenant SaaS did for growth potential. Adoption that used to take years now takes weeks. But the two questions that made SaaS such a beautiful asset, margins and predictability, are suddenly open questions again. Inference costs real money. Models commoditize. The moat that felt permanent last quarter looks rentable this quarter. Faster growth, far less certain economics. Plenty of excitement and energy, and very few sober voices in the room asking where and how this plane lands safely, and what the shape of its approach looks like. We've already learned this much: the oversimplified SaaS rules of thumb break on AI. Lean on them to value or steer one of these companies and you'll be confidently wrong, because what actually makes AI different is precisely what those shortcuts can't see.

This is exactly the moment Munger and Buffett would slow down and ask the only questions that ever mattered to them. Not “How fast is it growing?” Rather, “How durable is it?” What are the real, lasting drivers of cash flow here, and which of these businesses will still be throwing off earnings in ten years, versus which are this cycle’s flash in the pan? Durability is unglamorous, but it is the whole game.

So I’m hunting for the AI version of that 2014 insight. The single function that sits underneath these AI business models the way the derivative sits underneath SaaS. I don’t think it’s been written down yet, the same way mine hadn’t been in 2013. It won’t come from a dashboard or a tweet. It comes from analysis, from creative thinking, and from sitting with founders and pulling their economics apart line by line, which is exactly where I’m spending my days. It also has me back in my mechanical engineering textbooks from 2009, looking for the function from the physical world that happens to describe this one. I have an idea, but it’s still a work in progress (and it’s much more complicated than a first and second derivative to understand). Once I get it clear, you can be assured that I will share it with you.

Here’s the part I keep coming back to: SaaS is what we call it now, but that name never came from the technology. It came from the business model. Before the model was nailed, the technology was just cloud software. Instead of sitting on a local server, it was hosted in the cloud and delivered over the web. The moment the subscription model proved itself, SaaS swallowed cloud software whole, and now nobody remembers there was ever a difference. The business model became the name, and the name became the permanent identity.

Sit with that, because it tells you exactly where we are with AI.

Right now we have AI. That’s a technology. It is not a business model. We’re in the cloud software phase of this cycle, still naming the thing by its tech because the model underneath hasn’t been found, let alone named. There’s no working theory of the AI-native business model yet. There’s barely a vocabulary for it.

And this time I’m earlier than I was last time. In 2013, SaaS was already humming. I was reverse-engineering a model that mostly worked. With AI the model isn’t humming yet, which is harder, and far more interesting.

So here’s the bet. The function comes first, and the name follows. When I find the equation that actually describes how these businesses create and keep value, it won’t just let me evaluate them. It will reveal the shape of the AI-native business model itself, the real economic engine hiding inside this science experiment we keep calling AI. Find the function, and you name the category.

Subscribe now

Portfolio Highlights

Novella announced its Series A funding round, bringing its total capital raised to $21 million to super-power insurance brokers. We were one of a handful of investors in their pre-seed round in February 2024.

They are an AI-native wholesale insurance broker for the excess and surplus market, the corner of insurance that covers complex, hard-to-place risks that standard carriers turn down. Its AI agents handle the heavy operational work behind a placement, from binding and form comparison to policy review, inspections, billing, and renewals, so human producers can spend their time on client relationships and growing their books. With a research base in Tel Aviv and a growing US footprint, Novella is building the modern, AI-native broker for a market that has long run on manual work.

Permit.io is a full-stack authorization platform that handles the permissions logic inside applications, the rules that decide what each user is allowed to see and do. Instead of having engineers rebuild access control for every app, Permit provides a no-code policy builder that removes the need for in-house authorization expertise and lets non-technical business and security stakeholders take part in managing permissions directly. As software splinters into micro services and, increasingly, autonomous AI agents, that challenge only grows, and Permit gives teams one consistent foundation for managing access across users, services, and agents alike.

Very True Podcast

We’re open to any questions to cover in future episodes, just fill out the Q&A form.

Remember to subscribe via Spotify or Apple Podcasts.

Who we are

Verissimo Ventures is an early-stage Venture Fund based in Israel and the US. We take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our third fund.

Thoughts on Investing and Starting Up

What Does Enterprise Software Become?

I spent an hour this week chatting with Roy Rubin, and I haven’t stopped thinking about one question since.

Roy built Magento out of a freelance shop while he was still a student. He gave the software away for free, watched his customers fund his growth, and eventually sold the company to eBay. He now invests in enterprise companies at the earliest stages. So when he talks about how this business works, he’s not theorizing, he’s describing a thing he lived through from the inside.

The question we kept circling was simple to ask and hard to answer. What is enterprise software, actually, anymore. And the more we talked, the more I realized the answer lives in a part of the industry most people stopped paying attention to a long time ago.

The part of the story nobody tells

When Roy released Magento, he assumed the business would be services and support. His intuition, after years of picking up the phone in the middle of the night for e-commerce customers, was that mission-critical software means people want a vendor on the other end of the line. He was right about the need, wrong about who would fill it. The systems integrators who downloaded his open-source product turned out to be better at support than he was, and overnight he had an ecosystem servicing his customers for him.

So he leaned into the product business. He intentionally left holes in the product, gaps for third parties to come in, plug, customize, and ultimately own the customer relationship. The product company built the core. The integrators did the work.

This was not a Magento quirk. This is the whole history of enterprise software. Oracle, SAP, Salesforce, every large platform got distributed, implemented, and made to actually function inside real organizations by a vast layer of value-added resellers and systems integrators sitting on top of it. While valuable, building the software was the easy part. Making it work was the business.

We tend to forget this because the SIs and VARs were never the glamorous companies. Nobody grew up wanting to work at one. They trade at half a turn of revenue while the platform vendors trade at ten times or twenty. But they were always there, and they were always essential, because software has always been a bunch of screens on top of a database that somebody still has to install, configure, and teach people to use.

The premium was a wedge

For about twenty years, the most valuable software companies shared two traits. They were a system of record, and they encoded a work methodology. A dollar of revenue at a company that was both was worth more than a dollar of revenue at a company that was only one. Salesforce is the obvious case. While it is a fairly straightforward database with screens on top, it also defined the modern CRM. It became the way an entire profession does its job, to the point that "Salesforce admin" is a job title.

That second trait, the methodology, is what makes revenue sticky. It is what lets incumbents survive on friction alone. Moving off a deeply embedded workflow was so painful that the new thing had to be not a little better but dramatically better to justify the switch. The friction is the moat.

What’s changing now with AI is that the friction is collapsing. Migrating from one system to the next used to be a year-long project. With APIs, MCPs, and agents that can run a migration cleanly in a week, the switching cost that protected a generation of incumbents is dropping toward zero. The wedge that the SIs drove in, and that the platforms turned into pricing power, is getting thinner.

But what comes out of this may not be what you first assume.

The labs just answered the question, sort of

Here is the part that made me sit up. For the entire history of this industry, the model was the platform vendor builds the product, and a separate ecosystem of integrators does the services. The two stayed separate. Service revenue was even vilified on public company income statements, low-margin, non-recurring, the stuff you wanted off your S-1.

This month, that separation broke in the most visible way possible. Anthropic launched an enterprise services firm with Blackstone, Hellman & Friedman, and Goldman Sachs. Days later, OpenAI launched its own majority-owned deployment company and bought an applied AI firm to staff it. Both are copying Palantir’s forward-deployed-engineer model, putting their own people on-site inside customers to make the technology actually work.

In other words, the platform vendors looked at the services layer that every prior generation handed to an ecosystem, and decided to keep it. They are betting that the bottleneck in this cycle is not the model, it’s implementation, and that whoever owns implementation owns the customer.

Roy left holes in his product so an ecosystem would form. The labs are trying to do the exact opposite, pulling the work back in house before an ecosystem can form at all. Same structural question, opposite answer, and it’s genuinely unclear who will be right.

So what does it become

If the model layer is a commodity, and the switching cost between models is zero, and the platform vendors are absorbing the services layer themselves, then the comfortable distinctions we’ve relied on for two decades stop holding. A lot of what’s being sold as vertical AI software turns out, on inspection, to be a system integrator wearing a SaaS costume. The startups discovering this are learning they have to charge like Accenture to deliver like Oracle, because customers need so much help that the product alone was never the thing being bought.

That doesn’t mean the startups lose. It means the thing that wins looks different from what won last cycle. Not a thin wrapper on a model, but a genuinely deep workflow that the model providers won’t bother to build and the incumbents are too slow to reach. The value moves back toward the part everyone undervalued, the work of making the software actually fit the business, and away from the part everyone overvalued, the interface on top of the database.

Which is to say the future of enterprise software might look a lot like its past. Customers funding growth. Services and solutions as the real business. Integrators, by whatever name, capturing the value. The picks and shovels outlasting the gold rush, the way they always do.

The full conversation with Roy Rubin is on the Very True podcast.

Very True Podcast

We’re open to any questions to cover in future episodes, just fill out the Q&A form.

Remember to subscribe via Spotify or Apple Podcasts.

Portfolio Highlights

Rossum builds AI agents that handle the paperwork behind everyday business operations, and this month it was acquired by Coupa to power its push toward autonomous spend management. Rossum's agents read invoices and orders, check the data, send follow up emails, and pass everything into a company's core systems. It works across messy real world formats, from scanned mail to PDFs, and is trusted by companies like Siemens, Bosch, and Panasonic.

CCX is a software platform that streamlines how healthcare companies and insurers negotiate access to new medicines. It spans planning, workflows, contract administration, and digital negotiation, compressing deals that used to take months into hours. It is a massive, complex category still largely run on email and spreadsheets, and CCX is quietly becoming the system of record for an industry where faster, cleaner contracting means better patient access to the treatments they need.

CoLab helps large engineering teams review designs together and reach decisions faster. Engineers run virtual design reviews in the browser, and as they do, CoLab automatically captures their feedback and expertise. That knowledge then powers AI agents that flag errors in drawings and surface lessons from past projects, so teams catch problems early instead of after something ships. Customers include Ford, Komatsu, and Bombardier, which recently signed a multimillion dollar AI contract with the company.

Certa gives companies one place to vet and monitor the outside vendors, suppliers, and partners they rely on. The platform uses AI to automate the full relationship, from onboarding and background checks to ongoing risk and compliance monitoring, a process that has long been slow and manual for most companies. Customers like Honeywell, Uber, and Box use Certa to onboard new partners up to three times faster, and it was recently named a Leader in the first ever Gartner Magic Quadrant for third party risk management tools.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our third fund.

The Director and the Critic

In February I wrote about the End of Lazy SaaS. The thesis was that AI isn’t killing SaaS - it’s purifying it. The fundamentals haven’t changed; the margin for error has. Good enough is no longer good enough. System of record plus methodology is still the moat. The market is finally separating the platforms that define how work gets done from the products that just sit on top of someone else’s workflow.

That was the diagnosis. This is the prescription - what we’re funding at the early stage as the new rules take hold.

Start with an uncomfortable truth.
For a very long time, software was too easy. Not the building of it - that part was hard. The winning part was easy. The barrier of syntax and engineering meant good-enough really was good-enough. Whoever could hire enough engineers to make the damn thing work, then put a strong sales-and-marketing CEO on top, won. Look at the leading SaaS companies of the last fifteen years and the CEO archetype is a pattern, not a coincidence. Marc Benioff at Salesforce, who came up through Oracle sales. Bill McDermott, who started his career selling Xerox copiers door-to-door and ended up running both SAP and ServiceNow. Frank Slootman, the canonical “professional CEO,” who rode the same playbook through Data Domain, ServiceNow, and Snowflake - three IPOs, zero engineering background, brought in each time specifically to scale GTM. Tom Reilly at ArcSight and then Cloudera, same model. Sales-led, GTM-first operators, running businesses where execution was the limiting factor and creativity was already in the rear-view mirror. The engineer-founders - the Drew Houstons and Eric Yuans - were the celebrated exceptions, and the celebration itself proves the rule.

This worked because execution was scarce. Creativity was a luxury because syntax was a tax. You couldn’t build unless you could code, which meant the people in the room were a narrow slice of the people who had ideas worth building. The slice that could build was small enough that whoever got there first with a working product and a real GTM motion ran the table.

And here’s the part I want to say out loud, because it gets to what actually killed SaaS: that environment didn’t just reward sales-led companies. It created a VC-subsidized race to the bottom. Infinite competitors chasing finite markets. Capital flooding in faster than software could find new problems to solve. Valuations climbing because of supply and demand of capital, not software. Growth expectations becoming insane because the alternative was admitting that the next round wouldn’t clear at the last round’s price. Software was supposed to scale infinitely - that was the entire point of the business model. Instead, the spending to acquire customers scaled infinitely, and the businesses underneath got hollower and hollower.

That era is over. AI isn’t what ended it - the structural rot was already there - but AI is what’s now exposing it.

This is where I want to draw a line that I think most VCs are about to draw in the wrong place.

The difference between a movie director and a movie critic.

AI is going to replace the critic. The truly great director has a job for life. Critics react to what already exists - they evaluate, compare, refine. Directors decide what should exist in the first place. One is downstream of creation. The other is creation.

The question I’m asking on every deal right now is: are we reacting or innovating?

Reacting is a great-looking AI feature on top of a known workflow. Reacting is “the same thing, but smarter.” Reacting is a model wrapper with a thesis that gets cheaper every month. It feels like progress because it produces output, but it’s downstream of someone else’s creative decision. And here’s the inconvenient truth about reacting: the LLM developers themselves will probably eat most of this market. It’s right there for the taking. They have the models, the distribution, the capital, and the surface area. If your edge is a thinner version of what GPT or Claude or Gemini can already do, you’re not competing with another startup - you’re competing with the foundation model companies’ next product update.

What the foundation model companies don’t have, and frankly can’t have, is the deep domain expertise and creative conviction to truly innovate inside every niche industry. The economics don’t work for them to go vertical. The depth of insight required - actually understanding how a specific job function operates, what it actually needs, what nobody has bothered to build for it - isn’t something they can acquire at scale, because depth and scale are fundamentally opposed. That gap is where the real opportunity sits. Not at the model layer. Not at the obvious horizontal applications. In the messy, overlooked verticals where someone with conviction and taste can define a category before the giants notice.

Innovating, in other words, is harder to spot and easier to underestimate. It looks like a founder defining a category that didn’t have a name yet. It looks like a workflow that didn’t exist before someone insisted it should. It looks, often, a little crazy at first.

And here’s the part that should make every investor lean in: software is finally becoming what it was always meant to be - creative expression. The barrier to entry was never insight. The barrier was syntax. AI is removing it. The expression layer is being massively democratized. Which means a wave of people who always had the creative conviction but never had the language are about to start shipping.

Some of them will be the next Salesforce of [insert overlooked industry or function here]. Most VCs will miss them because they’re still pattern-matching on the old signals - the technical co-founder, the polished demo, the familiar GTM motion. The new signal is creative conviction in a previously locked-out builder.

That knife’s edge - between reactive AI features and genuinely creative founders newly able to express themselves - is where we’re trying to make better calls than the next firm.

What this means for what I’m funding right now:

I’m not backing prettier versions of existing tools.
If your edge is “Salesforce but with better AI,” you’re funding a feature, not a company. The incumbent has the workflow and the mindshare. You have a model that’s three months from being commoditized. You might be a player, but it’s their universe.

I’m looking for founders who’ve decided what should exist, not what could be improved.
Not “this would be useful.” A specific, opinionated, almost stubborn view of how a process should operate, and the conviction to mandate it on customers. As I keep telling companies: be someone’s Salesforce. Be the tool a specific job function sits inside all day - not just for automation, but for an opinionated framework for how that job should get done. Most people don’t want a creative outlet in their day-to-day work. They want a system that tells them what good looks like. The biggest opportunities are in markets that have been long overlooked by VCs and founders, where adoption patterns are finally moving fast enough to matter.

I’m overweighting workflow gravity over feature parity.
The question isn’t “is this product better.” It’s “once a customer adopts this, can they ever leave?” If the answer is no - if the company’s process now lives inside the product, if the team’s resumes reflect it, if the data model has rewired the org chart - that’s the bet. This is also why so many AI-bolt-on attempts at existing workflows have struggled. There are too many variables, too many unknowns, to plug a brand-new technology into a system that was built around different assumptions. It’s almost paradoxical to expect otherwise.

I’m underwriting character harder than ever.
When code is cheap, the bottleneck is judgment. The opinion has to come from somewhere, and the somewhere is the founder. This is why my Ratzon, Mazel, Middos framework matters more in 2026 than it did in 2020. The founder’s clarity about how the world should work is now the actual product. Swimming through the technology, media, and market waves while shipping product and building real customer relationships is the only durable competence left.

One more thing.

When everyone was asking for more buttons, Steve Jobs gave them zero.

That’s the whole argument. Great software isn’t built by surveying the market and shipping the consensus. It’s built by founders with the conviction to say “no, this is how it should work” - and the taste to be right.

Here’s where I’ll push back on myself. A few model generations from now, AI will have taste. It’ll evaluate designs, recognize what’s working, and execute against a defined aesthetic better than most humans. Taste is pattern recognition at scale, and that’s what these systems are built to do.

But taste isn’t the same as creativity. Taste is judgment over known patterns. Creativity is generating a new pattern worth having taste about. When Jobs killed the buttons, he wasn’t applying taste to existing phones - he was inventing a category that didn’t exist yet, then defining the taste for it. Models interpolate within the distribution of human output. The genuinely creative move is always outside that distribution, which is why it looks insane until it’s obvious.

That’s the layer that stays human. And in software, it’s the founder.

AI made the cheap part cheaper. It will eventually have great taste. What it won’t have is the creative conviction to mandate a new opinion on a category that doesn’t know it needs one yet. That’s the expensive part. That’s the founder’s job.

That’s what we’re funding.

Share

Portfolio Highlights

Stack is a unified platform for solo operators building AI consulting businesses. Members get a structured technical curriculum, an AI copilot called Stacey, mentorship from practitioners at companies like Cursor, Lovable, and Lindy, and the back office software to run a real services business, including invoicing, payments, and lead routing.

If you want to hire AI native us based freelancers or employees, be sure to reach out.

Kaipod is the largest software and support network for microschool founders in the country. As families increasingly seek smaller, more personalized alternatives to traditional school, KaiPod gives educators the operational stack to launch and run their own microschools, including the Newton management software, founder training, and a national community spanning 150+ schools across 30+ states.

Micro is the all-in-one tool and system of record for the agentic era combining email, messaging, CRM, projects, and AI in a single interface that organizes itself and gets work done for you. Built for early-stage founders, Micro pairs the industry's most advanced agentic memory with a growing library of plugins and skills to eliminate roughly 90% of the busy work around fundraising, hiring, and sales.

Very True Podcast

We’re open to any questions to cover in future episodes, just fill out the Q&A form.

Remember to subscribe via Spotify or Apple Podcasts.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our third fund.

Subscribe now

Thoughts on Investing and Starting Up

The Three Types of Seed Investors
(and How to Close Each One)

I’ve been having the same conversation a lot lately. A pre-seed founder hits an inflection point and it’s time to raise a seed round. The instinct is always the same: build a massive list, schedule a ton of meetings, and “run a process.”

That instinct is usually wrong. Here’s why.

Fundraising is a market, but it’s a time-based, limited market. Everyone is looking at you at the same time, and only once. You don’t get to iterate (unless you’re really smart about it). Founders think they need a big top of funnel. They don’t. Word gets around when you’ve talked to everyone, and it’s probably not good. Let word get around about who you haven’t talked to and you become a hot commodity.

And unlike your customers, investors are playing a zero-sum game. That competitive energy is a tool you can use, if you know how.

The Three Types

After years of sitting on both sides of these conversations, I categorize seed-stage investors into three types. This isn’t academic. It changes how you sequence your raise, who you prioritize, and how fast you close.

Type 1: The Lead.
The firm that writes the $2-5M check, sets the valuation, and anchors the round. Every founder starts by trying to get this meeting. But leads are not as independently minded as you’d think. VC is a herd business. FOMO matters. Social proof matters. The theater of fundraising matters.

Going straight to leads before you have any momentum is usually a mistake. You’re walking into a market with no signal, and they’ll read that correctly.

Sometimes the stars align: you identify one perfect-fit fund, engineer every interaction, and close in three weeks. But it takes real insight and relationship to make that work. And boiling the ocean hoping someone fully “gets” what you’re doing without a lot of your help is a Cinderella story, not a strategy.

Type 2: The Small Check Writer.
The angel, small fund, or operator-investor writing $100K-$500K. Most founders treat these people as a consolation prize. That’s backwards.

Add $500K to the balance sheet and you buy yourself months of additional traction. While that traction may seem small, remember that the entire business is small-angle approximation and every step forward adds critical fidelity to your big vision. You also add nodes to the cap table, which creates signal and leverage. And they won’t run you through the wringer on diligence or distract you from your business (don’t let them if they try).

Have a SAFE open. Set a reasonable cap ($15-25M) with a ~10-20% discount for early conviction and a simple MFN clause. Close these people. Send wire instructions. Put the money in the bank.

“Won’t a future lead be annoyed?” They’re not here yet. You need money in the business. Running your business to please yet-to-invest investors is not a recipe for fundraising success.

Type 3: “Let Us Know When You Have a Lead.”
This is the default position of most small VCs. It sounds supportive. It feels like progress. It’s neither. Once you have a lead, everyone comes in. That’s how herds work.

The skill is converting Type 3 into Type 2. When you feel someone drifting toward “keep us posted,” push the conversation forward: “We have a SAFE open now. This is how you lock in your allocation.” Nobody likes being called a follower. Give them the framing of early conviction, but also give them a real mechanism to act on it.

At this stage, getting a “let us know when you have a lead” may actually be worse than “we’re out.”

The Closer Mentality

The thread connecting all three types is what I’d call the closer mentality. Not pushy. Not desperate. Just clear-eyed about what you’re asking for and unwilling to let conversations drift into comfortable ambiguity.

Don’t prompt an investor with a live opportunity and expect them to come to you. You’re in control. Send the SAFE docs. Send the wiring info. When someone says “I’ll come in behind a lead,” respond with a concrete alternative. Convert the passive into the active. Give them a chance to be a hero. Saying you came into a company before First Round or Founder Collective carries real weight in this business.

Run a Tight Process

Don’t build a big CRM. Don’t buy Affinity. Use a Google Sheet. Four columns: Fund name, person at fund, status, who makes the intro. Share it with your key investors, advisors, and friends. Let everyone run it down. Simple scales. Complicated stalls.

Be a sniper, not a shotgunner. For each target: is the intro strong? Is this on-thesis? If you can’t answer yes to at least one, you’re wasting cycles.

On data rooms: think of it as the appendix to your deck. Anticipated follow-up questions, ready to go. Forward modeling if it’s relevant. Don’t spend ten hours on market research nobody asked for (you probably should have done this before you started the company anyway). Think carefully about what you want to disclose and when. You can and probably should hide customer names. But Bookings, Revenue, Pipeline and Usage tell investors two things: there is real momentum and you actually track the stuff that matters.

Most seed investors’ only KPI is to get your company a strong Series A. If you show you have those chops now, it derisks the work they’ll need to do and gives them the space to get excited about what you’re building.

Use AI as Your Second Set of Eyes

One tactical note most founders are sleeping on: use your note-taker and AI tools to interrogate your own pitch. Put all your investor conversation transcripts into a prompt and ask real questions:

• Which questions come up most often, and how do I modify my pitch to pre-empt them?

• Where am I losing engagement?

• Am I hitting all my key thesis-forming points in the first ten minutes?

• Can I lead investors to start giving me ideas that create buy-in for their participation?

Take investor feedback with a grain of salt, but learn to listen through the feedback rather than to it. AI can help you do that at scale.

The Meta-Lesson

This framework isn’t really about investors. It’s about markets. Momentum compounds. Social proof cascades. The participants largely follow each other. The founders who raise well don’t fight that structure. They use it. They sequence conversations, build momentum with small wins, and maintain the closer mentality throughout.

My dad is a CFO. He taught me early that your negotiating position is fundamentally different when there’s cash on the balance sheet. That lesson applies to everything, including how you raise a seed round.

Start there.

Share

Portfolio Highlights

Namespace just announced $23M in Seed and Series A funding led by NEA, with Susa Ventures and Burst Capital. We’re happy to have invested in their first round in 2021.

Founded by Hugo Santos, Namespace is building the unified command center for software development, bringing the same speed and efficiency used inside Google to the broader market. Historically, software engineers and DevOps teams have worked in silos with fragmented tools, creating bottlenecks and costly delays. Namespace solves this by providing an all-in-one platform that allows these teams to collaborate in one place, bridging the gap between writing code and launching it so companies can ship products significantly faster.

Nurra, founded by Ben Lauzier and Greg Kuwaye, is a hybrid human + AI care navigation platform. While most AI in healthcare focuses on provider workflows, Nurra tackles the far bigger problem: helping patients actually move through a fragmented system. By pairing dedicated care navigators with AI that automates outreach and follow-ups, Nurra delivers concierge-level chronic care at lower cost while improving engagement and outcomes. The team is ex-Lyft, Thumbtack, and One Medical, and has already signed a major enterprise contract they’re now scaling.

Subscribe now

Very True Podcast

We’re open to any questions to cover in future episodes, just fill out the Q&A form.

Remember to subscribe via Spotify or Apple Podcasts.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our third fund.

Thoughts on Investing and Starting Up

The End of Lazy SaaS: Same Game, Harder Rules

There is a lot of noise right now about “the end of SaaS”. Everyone is looking to AI as the great disruptor, but it’s really an evolution of digital products - the same way the cloud brought multi-tenant software to the masses rather than heavy appliance-based solutions. But if you look back at the SaaS analysis I was doing ten years ago, the fundamentals of what makes a company truly valuable haven’t actually changed.

What has changed is the margin for error.

The game is still the same, but the difficulty setting just got cranked up. The market is becoming sharper and tougher. The era of mediocre businesses slipping through the cracks and maintaining market share simply because they exist? That’s over.

And frankly, that is a good thing. Too many products got too lazy for too long.

The Two Pillars of the Moat

When we strip away the hype, the most valuable companies (the ones that survive this disruption) are the ones that do two specific things:

1. They are the System of Record:

   1. They define how the data is stored. They own the database.

2. They Define the Work Methodology:

   1. They define how the work actually gets done.

There is a massive difference between a tool you use for five minutes and a platform that dictates your workflow.

As I often tell companies: “Be someone’s Salesforce.” Meaning… the tool that an individual contributor is inside ALL DAY.

The Resume Paradox

Here is a contrarian indicator I’ve watched for years: The Inverse Proportion of Usability.

If you see a specific piece of software listed as a “skill” on a resume, it usually implies the product is difficult to use. It has a steep learning curve. Intuition tells us this is bad product design.

Business reality tells us this is a massive moat.

If a product is hard enough to use that you have to be “trained” in it, it means that product has successfully defined the methodology of the industry. It means the switching costs are astronomical. That is the “rarefied air” of enterprise value. It’s not just about UI; it’s about deep, structural lock-in.

The New Standard

For a long time, companies could get away with loose retention and weak expansion if their top-of-funnel was loud enough. That window has closed.

We are seeing a return to rigor across three layers:

• The Numbers: High retention and aggressive expansion are no longer “nice to haves,” they are survival metrics, even at massive scale.

• The Relationship: Do you have genuine lock-in with the customer, or are you just a vendor? Do you provide the relationship and insights that will weather the storm?

• The Tech: Does your product sit on top of the data, or does it define the data?

The disruption we are seeing isn’t destroying SaaS; it’s purifying it. It’s forcing a cleanup of the ecosystem. The companies that are merely “nice-to-have” features are going to evaporate. The ones that serve as the fundamental nervous system of their customers will not only survive, they’ll get stronger.

The market is demanding we stop building lazy products. It’s time to get sharp.

Bonus: These are slides from a presentation I put together in 2017… let me know if you think anything has fundamentally changed (besides the names and valuations).

Portfolio Highlights

RallyUp turns your biggest cost center into a revenue-generating engine by empowering employees to become micro-influencers by leveraging employee- generated content (EGC). We are excited to invest in Alec Kremins at the early stage as he builds RallyUp in this emerging category, which we believe can become a massive new distribution channel for companies.

Mixus is the trust layer for High-Stakes AI. Founded by Shai and Elliot, Mixus provides the path to leveraging AI in high-stakes environments by building trusted AI agents for high-stakes work. Just as Stripe unlocked payments and AWS unlocked the cloud, Mixus acts as the essential “trust layer” required for safe, enterprise-scale AI deployment. They allow companies to push AI usage to its operational limit by intelligently keeping a human in the loop right where the work already happens: your inbox. Mixus turns email into a controlled execution layer where the AI agent does the work, and a human approves the output before execution - meaning zero workflow changes for the user.

Share

Very True Podcast

We’re open to any questions to cover in future episodes, just fill out the Q&A form.

Remember to subscribe via Spotify or Apple Podcasts.

Your Business Model Probably Sucks - Part 1

Your Business Model Probably Sucks - Part 2

Living in Sprints: Navigating the Eras of a Founder’s Life with Todd Saunders

How Bad Finance Kills Good Companies with Ariel Menche of Raftel Strategy

Subscribe now

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our third fund.

Thoughts on Investing and Starting Up

We’re switching things up this week. Instead of our usual deep-dive written commentary, we’re letting the founders and investors do the talking. We’ve curated our top recent podcast conversations covering everything from venture math to autonomous checkout.

The purpose of this podcast is twofold:

1. To share many of the same ideas, analyses and hot takes in a more digestible, flexible format.

2. To share unique and engaging conversations with founders, other investors and friends, which cover a variety of industry-related topics.

We’re open to any questions to cover in future episodes, just fill out the Q&A form.

Looking forward to a great 2026!

Very True by Verissimo

Investing in AI, Selling Secondaries and Venture Math with Nick Fitz of Juniper Ventures

From Craft Beer to Autonomous Checkout: Living Life in Chapters with Avi Moskowitz, CEO of PDQ

It’s Never a Straight Line: Why the Path to an Exit Often Starts with a Stop - Neri Bluman of XFunnel (acquired by HubSpot)

The Frameworks You Need to Navigate VCs and Raise Your Next Round

Portfolio Highlights

CoLab has announced a $72M funding round led by Intrepid Growth, Insight Partners, and Y Combinator.

CoLab is the AI-Powered Operating System that enables engineers to ideate, review, and make their most critical decisions effectively. CoLab brings your experts, collective knowledge, and AI agents together so decisions can be made quickly with context and confidence—getting life-changing products into the world years sooner. CoLab is trusted by leaders like Ford, GE, Johnson Controls, John Deere, and Schaeffler.

Datafold is the unified data-quality platform that helps data teams ship faster and with confidence. By combining proactive automated testing and observability, it prevents bad code and data issues before they reach production, integrating directly into development, migrations, and monitoring workflows.

Preql automates the hardest part of AI adoption: cleaning, reconciling, and contextualizing messy enterprise data. Their AI agents transform fragmented ERP, CRM, HR, and expense data into structured, auditable, AI-ready pipelines that scale across the enterprise.

P.S. Back in the Saddle

On a more personal note, as some of you may have seen from the videos I’ve been posting on LinkedIn, I’ve officially gotten back on the bike.

It’s been a journey—I took a few years off after a crash, and returning to the sport required overcoming some real mental hurdles. That experience is exactly why I’ve fallen in love with Get Into Geerz.

It is an incredible organization here in Israel that uses mountain biking as a therapeutic tool for children and teens. It’s not just about exercise; it’s about teaching kids how to navigate difficult terrain in life. I’ve signed up for their charity ride to support this mission.

If you’d like to cheer me on (and support kids conquering their own uphill battles), you can check out my fundraising page here. Thanks!

Support the Ride & the Kids 🚴

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our third fund.

Thoughts on Investing and Starting Up

Want to hear this as a podcast with some additional anecdotes? Find it HERE.

When I started in venture capital in 2013 at NEA, it was hard to call VC an “asset class.” NEA was the biggest game in town and it still operated like it was in a cottage industry. I think it’s easy to see how it has evolved into its own asset class (rather than a small subset of private equity) in the last ten years, if only just by looking at the volume of capital that has poured into VC.

But I believe there still remains a cottage industry inside this new asset class. And it is not just based on investment stage.

Whether early stage or late stage, there are two fundamentally different approaches to making venture capital investments:

1. The first is the old-school cottage approach - work with high relationship quotient founders building companies with a path to sustainable revenue growth. Work closely with them on company building. Accrue value much faster than anywhere else in the market. Liquidity from strategic or financial M&A, sometimes IPOs.

2. The second is the $100Bn exit or bust and hype chasers. Whether early stage or late stage, it represents a fundamentally different approach. Whereas the cottage approach relies on operational and decision frameworks, this part of the asset class focuses on sales and marketing as core competencies (for investors). It’s all about access, intros, and attention. Whereas the cottage approach could stumble upon a $100Bn outcome which would drive insane returns, this approach abandons ship as soon as it becomes clear that the $100Bn outcome is off the table.

Neither one is “better” than the other. But they are fundamentally different. The issues occur when there is crossover between the two approaches - inside a company, inside a portfolio or inside a partnership. Early-stage founders often get stuck in the middle - they aren’t sure if they want to be the next $100Bn company or if a solid $200M exit, which results in them personally making $50M, is really what they’re really excited about. The reality is that all VCs basically look the same to founders, and many VCs still haven’t decided what they want (you’d also be hard-pressed to find a VC who says they’d prefer not to have a company like Databricks in their portfolio, but few would admit that their strategy is $50Bn+ or bust). Most early VCs invest cottage-style into strong businesses and founders and then hope for that outlier upside, but if it doesn’t come, that is also OK and strong returns are still to be had. Later stage founders on the other hand usually have much more clarity about their identity - they know which capital sources to target: either PE-style growth equity or something like the SoftBank Vision fund. Not long ago, all early stage venture models had $500M-$1Bn exits as their ideal successful exits to drive fund-level returns. The emergence of the $50Bn exit created insane returns in the asset class based on the original fund models. By definition, fund models built around finding the next batch of those $50Bn companies will not generate nearly as good of returns as those same exit sizes produced in earlier vintages. The biggest winners here are the AUM aggregators, who use these $50Bn exits as justification to raise 5-10x larger funds without meaningfully increasing their cost basis. Guaranteed fee streams drive personal wealth, and carry becomes an afterthought. I truly don’t know where this all ends up, but I can assure you that secondary sales are part of our strategy.

At Verissimo we made the decision at inception that we would invest in founders who meet three key criteria: Ratzon (desire & willpower), Mazel (luck & right-place-right-time), and Middos (character traits). Focusing on founder middos allows us to build an incredible community of founders with shared values and a vast network of other founders who come with them. It also selects out a handful of founders, many of whom have proven to be the best fundraisers and recruiters, but many of whom have also destroyed tremendous value, wreaked havoc on their networks, and even gone to jail.

As we continue to invest out of Fund 3 in a weirder and weirder market, we remain convinced that at our fund size, there exists more than enough alpha inside our network of founders who want to build strong, sustainable businesses and do so with vigor and integrity, delivering fantastic returns to our LPs. We believe that AI opens up a new opportunity for unit economics that allow even small venture checks to drive massive returns. A new set of rules for customer acquisition and product building are having a massive positive impact on the economies of scale of the typical startup. With our fundamentals-driven philosophy we sit perfectly at the intersection of all these factors and are extremely excited about what lies ahead in the next few years.

Check out the embellished recording here:

Subscribe now

🎙 Podcast Feature:

Very True by Verissimo

Check out Verissimo’s new podcast, Very True by Verissimo.

PLG & User Activation with Yaakov Carno of Valubyl

Is ARR Dead? Wait, What even is ARR?

The Financial Operations Flow: A Comprehensive Review

How to be Wrong in Early Stage Venture

Revenue at Seed - is it a game changer or a waste of time?

Tech on the Rocks

Tune in to Verissimo’s Nitay Joffe, and Kostas Pardalis, founder of Typedef, as they chat with tech geniuses on Tech on the Rocks - where hardware, cloud, and all things future-tech meet over a virtual drink!

🎧Episode 22 : Navigating the Future of AI and Data Infrastructure with Bauplan

https://techontherocks.show/22

Portfolio Highlights

Congratulations to the XFunnel team on their acquisition by HubSpot. Our partnership was brief but fantastic, and we’re proud to have been part of their journey.

XFunnel helps companies manage their visibility across large language models and AI answer engines. The platform tests different prompts across leading AI tools, analyzes how brands are mentioned, and pinpoints where they can gain ground. Marketing and growth teams use XFunnel to move beyond traditional SEO and build a repeatable strategy for winning attention in AI first discovery.

Gelt announced a $13M Series A led by Zvi Limon of the Rimon Group, Vintage Investment Partners, and TLV Partners.

Gelt builds software to automate tax preparation, generate ongoing optimization recommendations, and organize audit-ready documentation. By combining AI with licensed CPAs, Gelt monitors rules year-round, surfaces actionable steps, and implements client-specific strategies. The result is simpler workflows, clearer guidance, and real savings for high-net-worth individuals and growing businesses.

Namespace brings software engineers and DevOps together on the same page with a new framework for managing the entire lifecycle of code. Inspired by a widely used internal Google approach, it standardizes how teams build, test, and deploy so work flows in one shared language. The result is fewer handoffs, fewer bugs, and faster releases.

Substrata applies deep learning and reinforcement learning to decode human dynamics in real-time, powering socially aware AI that helps dealmakers read the room, navigate complex relationships, improve win rates and optimize sales velocity.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our Third Fund.

Thoughts on Investing and Starting Up

As summer comes to a close, we look forward to an exciting fall for our portfolio companies and potential new investments. We feel that the last 12-18 months have represented a non-linear leap forward in the technology world. New underlying technology is coming out of Silicon Valley faster than ever and new business disruption has come into focus like never before, especially in legacy industries. The amount of creativity and speed that we are seeing now is like nothing we’ve experienced before.

There is an old quote, which I first heard from Bill Gates: “We tend to overestimate what will change in the next 2 years and underestimate what will change in the next 10 years.” I think these numbers may be shifting - 1 year from now it’s hard to know what will be automated and what our new challenges will be, and a 5 year horizon feels like eternity.

Anyone who has read my posts before knows that I am a big calculus fan because of its accuracy in describing objects in motion. Position, velocity, acceleration - all derivatives that help us understand where things are headed. But there are higher order derivatives as well: jerk, snap, crackle, pop (not a cereal joke).

For the first time in my career, I think we are starting to experience these higher order derivatives. On a recent trip to San Francisco, multiple friends attested to the investment banking-like culture inside the top AI firms. A healthy dose of FOMO keeps people in the office late and back in early, 7 days a week. And this FOMO is real - things are moving so fast and changing so quickly, that people really just don’t want to miss a beat, especially as technology advancements accumulate. It’s not about the next deal or release, it’s about a fundamental shift in what is possible. This is extremely exciting and healthy for technology and the promise it offers to improve our lives. It feels like things had become more linear in tech for many years - changes got more predictable, incremental. And now, in the best way possible, things are once again unpredictable, and there is unprecedented excitement and fervor.

Some key takeaways from a summer spent thinking and chatting with some of the smartest people I know:

• Silicon Valley is back in the spotlight, leaving every other tech ecosystem in the dust (with one notable exception: Israel)

  • In times of linearity, satellite tech ecosystems can catch up. They can also use their local industrial context to build game-changing products based on the cutting edge core tech. In times of non-linear development, like we’re experiencing now, the tech-for-tech nature of Silicon Valley (and Israel) puts them at a huge advantage wherein the application and business builders are in close proximity to the core tech developers.

• Verticals are experiencing a renaissance

  • Vertical SaaS has always been a game of timing: build the tools for the verticals and hope they’re ready to adopt. AI changes this equation because it allows tool builders to close what I call the last 5% gap in automation, which essentially means that it works accurately to solve the finer industry-specific details. Automating the last 5% requires nuanced expertise and institutional knowledge - something that the new underlying tech enables. Bringing products that accomplish this to users forces them to say YES (or be left behind).

• The rate limiting step is now very clearly just humans

  • The inability and unwillingness to learn how to implement the next level of automation is holding everything back. As people open themselves up to changing how they think, how they pursue opportunities, and how they learn new skills, the pace of solution development can start to catch up to the pace of the underlying tech development.

• Abstract thinking will win the day

  • When the acceleration is accelerating, the only solution is to think on a higher order. Harness the power of human creativity and brain function to think abstractly about fundamental problems and novel ways to solve them, then imagine and build the technology to make it happen. It’s what everyone in tech has been waiting for, and now people just need to get comfortable being uncomfortable and think big.

We are extremely excited about what lies ahead and the people pursuing it most aggressively.

I look forward to spending more time in San Francisco and Tel Aviv soaking up the ever-changing ecosystem of builders who are pulling the future into the present.

P.S. Anyone with senior connections at United or El Al… let’s get the direct TLV-SFO flight back in the air!

Subscribe now

Programming, Events, Content and More

🎙 Podcast Feature:

Very True by Verissimo

Check out Verissimo’s new podcast, Very True by Verissimo, where we spotlight our It’s All About Everything founder series. Hosted by the Verissimo team, each episode features candid conversations with early-stage founders, operators, and investors shaping the future of tech. From behind-the-scenes startup stories to honest takes on scaling, fundraising, and staying resilient, Very True by Verissimo offers a real look at what it takes to build something bold.

Tech on the Rocks

Tune in to Verissimo’s Nitay Joffe, and Kostas Pardalis, founder of Typedef, as they chat with tech geniuses on Tech on the Rocks - where hardware, cloud, and all things future-tech meet over a virtual drink!

🎧Episode 21: Email as a Knowledge Graph: Micro CEO Brett on Rebuilding CRM at the Inbox

Portfolio Highlights

Receive announced a $4M seed funding round led by NGVP. We participated alongside Insight Partners, Blank Ventures, Corner Ventures, Clocktower Technology Ventures, and others.

Receive offers merchants immediate access to revenue tied up in slow payouts by leveraging real-time operating data. With no interest or fees, it provides a faster, more affordable alternative to traditional financing, built for the pace of modern commerce.

Row64 recently announced a $4M seed round led by Galaxy Interactive, with participation from Alumni Ventures and Differential Ventures.

Row64 is a visual, interactive, real‑time BI platform that brings game‑engine responsiveness to analytics. A a GPU- and CPU-accelerated stack delivers high‑frame‑rate interactivity and sub‑millisecond record‑level updates, even on datasets with hundreds of millions to billions of rows. The result: faster diagnosis, faster iteration, and decisions that keep pace with the tempo of modern-day operations.

XFunnel is an experiment-driven AI search engine optimization platform. XFunnel tracks, tests and boosts brand visibility inside AI-generated answers on tools like ChatGPT, Claude, Perplexity and Gemini by simulating real user prompts and analyzing how - and why - brands surface. XFunnel closes the loop, turning continuous experiment data into a proprietary data moat to create tailored optimization playbooks and workflows. Founded by serial entrepreneurs Neri Bluman and Beeri Amiel, XFunnel serves high-profile clients such as HubSpot, Monday.com, Wix, Fiverr, Check Point and many more.

Stack has built a business-in-a-box for AI consultants. It has already helped 50+ people across the US launch their solo businesses with a combination of business tools, training, mentoring, community and more.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We invest take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our Third Fund.

Thoughts on Investing and Starting Up

As a startup scales, something strange happens. The challenges don’t just grow, they compound. What starts as a tight, fast-moving team building a product turns into a complex organism with politics, legal exposure, edge-case customers, investor drama, hiring headaches, and pressure to "scale leadership."

In short: the emotional weight and headaches increase exponentially, not linearly.

At the same time, your ownership, the thing that made it all worth it in the early days, is shrinking. Even as your company’s valuation climbs, your personal equity utility curve starts to taper off. It’s now a smaller piece of a bigger pie, buried beneath a heavy liquidation preference stack and under rising board pressure. A personal $15M paper gain at Series A can feel thrilling. A marginal $50M of theoretical value at Series C (with 150 employees, 6 board members, 1,000 customers, and $80M of liquidation preference) - not so much if it costs you two years of stress, cortisol, and deferred life plans.

While the theoretical reward starts to flatten out, the risk and the stress both continue to increase non-linearly.

This leads to a predictable intersection point, what we call the Founder Expiration Point.

It’s the moment when the emotional cost starts to outweigh the economic upside. When you look at the road ahead - the next board meeting, the next 50 hires, the next midnight fire drill - and think: “The next $100M in company value just isn’t worth my next 2 years… I have $25M of equity on paper, but I’m still renting an apartment and I am exhausted.”

At that point, founders can typically do one of three things:

1. Step back or bring in outside leadership.
   Sometimes planned, sometimes quiet. Sometimes announced as "becoming executive chairman."

2. Push for secondary liquidity.
   If the equity can’t pay you emotionally, it better start paying you financially. We are big proponents of this approach. When big funds come in with big money, they are swinging for the fences. In order to maintain alignment between the founders and big money, founders need to be able to make the mental shift from playing defense/surviving/getting to the next stage, to swinging for the fences. That shift only happens from a position of personal financial stability (i.e., start playing with house money). This doesn’t mean massive $100M generational-wealth secondaries but rather $2-5M secondaries that take the pressure off and let founders live a bit closer to what everyone else seems to think they are worth.

3. Mentally check out.
   They’re still there in body, but the spark is gone. Execution suffers. Culture drifts.

You can think of this dynamic as two intersecting curves:

• The Tolerance Curve: how much pain you're willing to absorb.

• The Equity Utility Curve: how much marginal personal value you're getting from your ownership.

Where they cross, that’s the expiration point.

Not every founder hits it. Some grow with the company or have boundless energy and are willing to sacrifice it all. Some bring in a great COO early. Some get lucky with timing, market, or team. But for many, especially repeat founders, this moment comes sooner than outsiders expect. For investors, this is where it all comes back to the main reason you invested: the people. Pay close attention to what is going on in their world, how they look, how they talk, and figure out solutions to maintain alignment using the tools at your disposal.

Happy Fourth of July!

🇺🇸

Subscribe now

Programming, Events, Content & More

🎙 Podcast Feature:

Very True by Verissimo

Check out Verissimo’s new podcast, Very True by Verissimo, where we spotlight our It’s All About Everything founder series. Hosted by the Verissimo team, each episode features candid conversations with early-stage founders, operators, and investors shaping the future of tech. From behind-the-scenes startup stories to honest takes on scaling, fundraising, and staying resilient, Very True by Verissimo offers a real look at what it takes to build something bold.

🎧 Episode 1: Understanding Your Legal Docs
🎧 Episode 2: Tax Strategies and Wealth Management

Tech on the Rocks

Tune in to Verissimo’s Nitay Joffe, and Kostas Pardalis founder of Typedef, as they chat with tech geniuses on Tech on the Rocks - where hardware, cloud, and all things future-tech meet over a virtual drink!

🎧Episode 19: How Cloudflare Reinvents Serverless at Global Scale with Josh Howard

Portfolio Highlights

Typedef recently announced $5.5M in seed funding. The round was led by Pear VC, we participated along with Monochrome Ventures, Tokyo Black, and several angels.

Typedef is a serverless engine that helps teams run large-scale AI workflows, like summarizing documents or classifying support tickets, without building fragile, custom infrastructure. Today, many LLM applications rely on ad hoc scripts to chunk text, manage rate limits, call the models, and store results. It works, for a while, but quickly becomes brittle and expensive at scale. Typedef replaces that mess with a single declarative job: it handles batching, retries, cost caps, and output storage automatically, turning generative AI into something you can trust in production

Mixus announced a $2.6M pre-seed round, and we were proud to participate alongside Hannah Grey VC, Liquid 2 Ventures, OVTR, EverywhereVC, and FullCircleFund.

Mixus embeds AI agents into your team's workflow, blending the speed of leading models with the critical oversight of human expertise. It turns tasks that took hours into seconds, without sacrificing reliability, by addressing the demand for AI agent verifiability and control, making these tools accessible and safe for everyone.

Clara Health lets teams build HIPAA-compliant AI automations in chat, tailored to their business, with help from a network of AI experts and partners. Customers include major home-care providers in the US.

OurRitual is modernizing relationship support with a blend of expert guidance and personalized digital tools. While therapy has become more accessible, couples often lack affordable, tailored options. By combining weekly sessions with self-guided app experiences, OurRitual makes it easier for partners to build stronger, healthier relationships.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We invest primarily in enterprise software companies and take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our $26M Fund 2.

Thoughts on Investing and Starting Up

It’s been a big week in AI. Google, OpenAI, and Anthropic all had major releases, and one clear throughline was the push toward increasingly autonomous coding agents. So we figured this was the perfect moment to talk about how unreliable Large Language Models (LLMs) are as a base technology, and what that means for builders trying to work with them.

Unreliability is the core bottleneck to unlocking the full power of LLMs. For all the deserved excitement around LLMs, most users still engage with them only occasionally. Daily active use remains comparatively low. You could read that as limited utility or slow dispersion, but we think a major contributor is unreliability. When a system can’t be trusted to work consistently, its real-world utility collapses. It’s no coincidence that the clearest value from LLMs so far has come from code generation, where outputs are not only useful, but highly verifiable. You can run the code, test it, compile it. It either works or it doesn’t. And the product it supports—software—is deterministic by design.

This piece explores:

• The core challenge: LLMs are fundamentally unreliable, a characteristic we believe is unlikely to change soon.

• Implications for builders: Success in building AI-powered products doesn’t come from waiting for perfect models, but rather from from building around their inherent variance.

• Framework: We outline four distinct approaches builders can take to manage variance. We'll map the core characteristics of each, offering a clearer lens for assessing whether a given team's capabilities and chosen strategy align.

LLMs are temperamental

We’ve spent the past few years closely tracking the LLM space, following the research, watching how companies are deploying them, how users interact with them, and using the models extensively ourselves. These observations have made one thing abundantly clear: large language models are fundamentally unreliable.

• Their outputs often deviate significantly from the intended result.

  • This unreliability is persistent. It might be low in some cases, but it never fully disappears, and it’s largely unpredictable ahead of time.

  • Even in the most constrained, well-defined tasks, LLM behavior is not deterministic.

• Crucially, when LLMs are asked to perform tasks that involve any kind of real agency—multi-step actions, tool use, autonomy—these reliability issues get far worse. Failure rates spike to levels that make these systems unusable in real-world scenarios.

• We believe these core reliability dynamics of LLMs aren’t likely to change meaningfully in the short to medium term.

Some concrete findings

1. LLMs hallucinate—that is, they generate incorrect outputs. The most reputable benchmarks we have, including those used by leading labs in their model cards, indicate that baseline hallucination rates, in scenarios where the model relies on its “own memory” (a useful proxy for many real-world failure modes), are around 50% even for the best models.

   

   1. There are narrow usage patterns, e.g., when the answer is explicitly included in the prompt where error rates can drop to ~1%, but this is fragile: these rates are heavily dependent on the type of question, and even then, it still falls well short of software-grade reliability.

      

   2. Theoretical work indicates that a non-zero probability of hallucination is an inherent and unavoidable characteristic of LLMs.

   3. As avid users, we can attest firsthand: while skill can help mitigate failures, it is nowhere near sufficient for full reliability.

2. These reliability issues get worse when LLMs act as agents, systems that must orchestrate tools, interact with software, and carry out multi-step tasks. Empirical results across a range of the most rigorously constructed agent evaluation benchmarks show agentic AI performance well below human levels, and worse, the rate and nature of these failures (their unpredictability and chance that they can happen at “any time”) make these systems unusable.

   1. Partial list:

      1. The Agent Company, Vending-Bench, Realistic Evaluations for Agents Leaderboard, HAL: TAU-bench Airline, OSWorld, GAIA Leaderboard

3. Code generation is one of the most mature use cases for LLMs—and even here, the same reliability issues persist. Widely cited research from METR, arguably the most thorough study of LLM code performance to date, suggests that reaching 99% correctness is only feasible for code a human could write in under a minute. There are caveats (including how human baselines are estimated), but the core insight holds: even with strong feedback signals and tightly scoped tasks, LLMs still make small, consequential mistakes—making truly high-accuracy code output surprisingly difficult to achieve.

4. LLMs are highly input-sensitive. Minor, even purely cosmetic, changes to prompts can produce wildly different outputs, undermining predictability and confidence in test results

   1. This ties back to the core strength of LLMs: their flexibility. You don’t need to know the exact input in advance, that’s why we use them. But it also means that performance on “in-distribution” test cases doesn’t tell you much about how the model will perform in the wild.

5. Alignment. We won’t dive into the full complexity of the alignment discussion here—it’s a deep and important topic—but the basic idea is this: alignment is the discipline concerned with making sure that models behave, at a high level, the way we intend them to. It’s what tries to prevent scenarios where an LLM blackmails its developer to avoid being shut off, or gives users instructions on how to build bioweapons. It’s a critical area of research, and very much not a solved one. Beyond just illustrating how fundamentally opaque these systems still are, unsolved alignment directly impacts the viability of agentic LLM use cases. But as noted, this is a topic worthy of its own deep dive, so we’ll leave it at that for now.

Projecting forward

Forecasting AI progress is notoriously difficult, especially in a domain evolving this quickly. That said, we have reason to believe that the core reliability dynamics of LLMs aren’t likely to change meaningfully in the short to medium term.

Here’s why:

1. A key reason we believe these reliability issues will persist is the way LLM errors interact with task complexity. Our own observations (and interpretations of much of the available research and empirical findings), suggest that LLM failure modes compound with task length in ways that exceed typical human error patterns. This tendency for AI errors to multiply and cascade will, we suspect, remain a fundamental barrier for truly reliable agentic systems for the foreseeable future.

2. The fact that no LLM use case has demonstrated true determinism in real-world applications is significant. It strongly supports the theoretical arguments discussed earlier, which posit that hallucinations are an unavoidable, and likely unsolvable, aspect of current LLM architectures.

3. A striking example of LLMs' operational opacity comes from Anthropic's research, illustrated by how a model might calculate 36+59=95. Internally, it uses a sophisticated, multi-path strategy, approximating the sum with one heuristic while precisely determining the final digit with another. Yet, if asked to explain its calculation, the LLM describes the standard 'carry the one' algorithm taught to humans. This suggests the model lacks the mechanism to introspect and articulate the actual computational path it took. This inability to reflect post hoc—even to itself—underscores the black-box nature of LLMs. While speculative, we suspect this missing self-awareness contributes to their unreliability.

   

4. The jaggedness of AI capabilities remains a major driver of unreliability. Models can demonstrate superhuman performance on advanced mathematics, yet still fail spectacularly on basic arithmetic. This inconsistency—seen across many domains—makes complex, multi-step tasks especially fragile: a single failure in an otherwise simple step the model happens to be bad at can drastically increase the overall error rate.

   

What this means for developers

We see two strategic tracks for building around LLM variance—(1) systems that operate without user verification, and (2) systems that manage verification explicitly. Each further breaks down into two distinct approaches.

1. Autonomy

   1. Determinism

   2. Accurate enough

2. Human in the loop

   1. End-user verification

   2. Verification at the provider level

These paths differ significantly in their strengths, the demands they place on development teams, and the sources of competitive advantage they offer. The following sections will unpack each one in turn.

1. Autonomy

This first broad strategy focuses on creating systems where the AI's output is intended to be used directly, without an explicit verification step by the end-user for each instance.

This includes systems that incorporate internal exception handling, where the system itself can detect inputs or scenarios it cannot process to the required standard of reliability and then gracefully manage these exceptions (e.g., by flagging them, routing to a different process, or informing the user it cannot proceed). The key concept here is allowing the system to 'run without babysitting', requiring additional capacity for when the system flags exceptions (usually means you need manpower) doesn't change this underlying dynamic.

This is the only category that enables fully autonomous usage, systems that can operate without human intervention. That makes it arguably the highest-margin path, with the clearest ROI story: either an existing process becomes faster, cheaper, and more scalable, or a new one becomes viable, and in both cases, the value is easy to measure. You can point to what the system does, what it improves, and what it costs.

This can be pursued in several ways:

1.A. Determinism

The aim here is to engineer an LLM-based system that, for all practical purposes, achieves complete determinism, mirroring the reliability of traditional software. Realizing this requires a system capable of exhaustively testing its outputs against rigorous criteria and, critically, remediating nearly all conceivable LLM-generated issues. Companies like Skyramp, focusing on deterministic AI for test generation, are exploring aspects of this space.

In practice, attaining this level of comprehensive determinism is extraordinarily challenging. It necessitates absolute confidence in having identified and addressed all potential edge cases and failure modes, a feat few if any have achieved in practice.

What defines this approach

• This approach demands deep AI technical expertise. Getting an LLM-based system to behave deterministically is extremely hard, and being confident that it actually does requires a sophisticated understanding of how and where LLMs can fail. Unless the problem is already widely solved, teams must be able to both achieve high accuracy and rigorously validate it across edge cases.

• Such a system likely requires a very sophisticated engineering layer, IP-worthy at that, this dynamic serves as a moat from both “models getting better" and competitors.

  • This does mean that there are likely serious demands for quality engineering talent as well, and depending on specifics, specialized talent (who else has experience in thinking about a system’s weak points?).

• Your customer needs to trust you, really trust you that your tech does what it says it does, earning their trust may not be easy, but if you do, that is another source of competitive advantage.

1.B. Accurate enough

This path focuses on achieving a level of accuracy that is 'good enough' for a particular application where direct, instance-by-instance user verification is not part of the primary workflow, and occasional errors are within acceptable tolerance.

A critical requirement is achieving and validating a high degree of confidence that you have thoroughly accounted for all significant edge cases and the varied ways your LLM product can fail within the specific operational context.

Examples:

• An AI classifier designed for very large, unstructured datasets, used for broad-based trend analysis. Here, occasional misclassifications can be tolerated as they are unlikely to skew the overall insights significantly.

• Automated structured data extraction from documents (like invoices or forms), where the system achieves a high enough accuracy for direct use in downstream processes, with the understanding that a very small residual error rate is acceptable.

• AI-powered note-takers for meetings. In many common usage patterns, there's no explicit human check on the generated notes. The underlying logic is often that notes with minor errors or omissions still offer greater utility than the alternatives of laborious manual note-taking or having no record at all.

What defines this approach:

1. For this strategy, it's critical to deeply analyze the user's workflow and the actual consequences of different AI failure modes (e.g., overgeneralization vs. complete fabrication, missed information vs. incorrect information). This involves an in-depth mapping of how various errors will impact the user's objectives and risk exposure. Crucially, developers must recognize that users themselves can be poor judges of these risks or may be overly trusting, as evidenced by ongoing issues like professionals misusing LLMs in critical fields despite widely known failure patterns (e.g., at least 20 cases of legal decisions where AI produced hallucinations just this month, despite hallucinations in law being a high-profile failure mode for years). The onus is on the product team to anticipate these issues, understand the true (not just stated) risk tolerance, and design accordingly, rather than solely relying on users to self-regulate their trust.

2. While perhaps less demanding than striving for proper determinism, it still requires strong enough AI skills to know what accuracy actually means in context, how to measure it, interpret it, and design around it.

3. For this strategy, the central aspect of product design is ensuring the product's core AI performance (its accuracy and reliability) is properly calibrated. While usability is always a factor, it's secondary here to achieving the target performance level.

4. Customers must trust the product's defined level of accuracy. This requires demonstrating consistent performance within those bounds, which can then become a key differentiator and competitive moat.

2. Human in the loop

Unlike approaches that strive to engineer out all AI variance, the second broad strategy for LLM product development accepts a degree of inherent model variability and focuses on managing its impact by integrating human verification into the process.

2.A. End-user verification

This common approach involves creating products where the end-user directly reviews, validates, and potentially corrects the AI's output as part of their workflow. To succeed, the value proposition must clearly justify the user's verification effort; the AI's assistance, even if imperfect, must be preferable to alternatives.

• Examples

  • Search & discovery tools (e.g., web search, shopping search): Users often don't need every possible correct answer and can tolerate some irrelevant results if the tool quickly surfaces useful or intriguing information.

  • Information triage (e.g., "important document" finders in email): An AI might flag potentially missed invoices or critical communications. The user quickly verifies if the flagged items are indeed relevant, a task easier than manually sifting through everything.

  • Powerful general-purpose chatbots: LLMs via the chat interface are so broadly capable that users find ways to derive utility despite needing to critically evaluate responses and manage occasional errors.

  • "Copilot" paradigms (e.g., coding assistants like Cursor, legal tools like Harvey, etc.): The AI generates drafts or suggestions (code, legal clauses, analyses), and the expert user, who is "in the loop" at the right time, can quickly identify and correct issues.

  • Creative “content generation”: These are the canonical examples. Users can easily assess if the output’s style and basic coherence are suitable.

What defines this approach

• Workflow augmentation: These products almost always enhance a specific user persona’s workflow rather than fully automating a standalone function (there’s also a small class of “alerting” use cases, but we assess that category to be relatively limited). This stems directly from the need for human verification: when humans are in the loop, they’re usually using the AI to do their job better, not babysitting it. In other words, the AI augments the human, not the other way around. While you can design a system where a human simply verifies AI outputs, that pattern rarely proves compelling in practice. The output has to be so valuable that it justifies paying someone to check it, and most applications don’t clear that bar.

• Ceiling more than floor: Since this approach often involves users adopting new ways of working, the "ceiling" of value, the significant benefit derived from the AI's best or most insightful contributions, must be exceptionally high. This potential for substantial gains (e.g., speed, unique insights, quality improvements) is instrumental in justifying a change in work pattern and earn the user's verification effort. The focus is less on average AI accuracy ("floor") and more on the transformative potential of the AI's peak contributions.

• Usability is central: Since user interaction and verification are integral, nailing the UI/UX for verification, correction, and seamless integration into the user's existing or new workflow is vital. Quality of user experience is then the primary differentiator and source of competitive advantage.

• Designed for verifiability: A central characteristic is the design of AI outputs that are easily and reliably verifiable by the intended skilled end-user. This involves careful consideration of information presentation, the nature of the AI’s output, and ensuring the right user is targeted (e.g., AI accounting tools for accountants who can vet the suggestions).

2.B. Verification at the provider level

This distinct model places the responsibility for verification and ensuring the quality of the final output squarely on the product or service provider. This involves the provider's own internal human review processes, often working in concert with AI and specialized software to deliver a "done-for-you," result. From the customer's perspective, the AI's imperfections are handled "behind the curtain."

The end-user perceives this as an "AI-powered service". Usually, this involves the provider taking over a specific, often narrow, function previously handled by the client, fulfilling it in a third-party services style. The client receives a completed task or a digestible deliverable, effectively outsourcing the complexity of managing the AI and its potential variance.

What defines this approach

1. Specialized, narrow-scope offerings with optimized playbooks: Real value and defensibility are typically unlocked when a provider identifies a narrow scope—a specific, well-defined problem or function that can be demonstrably and massively improved with AI. This focus is critical because:

   1. Identifying such improvable functions and developing effective playbooks is a key capability in itself; achieving this across broader, more general functions is significantly more challenging.

   2. As empirical findings to date suggest, simply layering AI onto existing broad processes (as some "rollup" strategies might attempt) often falls short of delivering significant productivity gains. True transformation typically requires a fundamental redesign of how work is done, which is most feasible within a narrow, provider-controlled scope.

   3. Investing in a dedicated software layer around the LLM for a provider-managed workflow (e.g., specialized internal UIs, tailored integrations, automation scripts) typically yields a high return on investment. However, this level of impactful, specialized engineering is generally only feasible when the operational scope is narrowly defined; it becomes impractical or offers diminishing returns for overly broad, general-purpose applications.

2. AI integration & workforce management: The provider's operational ability to effectively integrate AI tools with their skilled human workforce to genuinely maximize productivity and quality is essential.

3. The pivotal role of domain expertise: This expertise is necessary for designing effective AI-augmented workflows, facilitating effective quality control, and ensuring the final output meets exacting client standards. This expertise becomes a significant competitive differentiator.

4. Provider control optimizes outcomes: A key advantage is that the provider designs and controls the optimal internal flow for how the AI is best used. There's no need to train end-users on the nuances of AI interaction (e.g., prompt engineering, watching for hallucinations), which removes a whole class of failure that no longer needs to be accounted for and gives the provider quite a wide margin of tolerable variance rates.

5. Providers have a strategic choice regarding transparency: whether or not to explicitly inform end-users that AI was involved in producing the service or deliverable.

6. The business dynamics closely resemble those of specialized, high-value third-party service providers. Brand reputation for quality, reliability, and deep expertise is paramount.

7. Subject to the amount of specialization, the addressable market for any single offering may be smaller, making a strategy of "dominating your pond" or achieving leadership in a specific niche essential for success.

Oh we should have mentioned, there is technically one more strategy available…

Final thought: constraint as catalyst

The unreliability of LLMs isn’t just a bug, it’s the foundational constraint that defines what’s worth building, how it should be built, and what defensibility even looks like in this new wave of AI-native products.

Indeed, one could argue that unreliability is the raison d'être for much of the application layer, it's the challenge that creates the opportunity for builders to deliver value.

Builders can’t wish the variance away. They have to choose where and how they handle it, and that choice is what determines whether their product is viable, differentiated, and durable.

The best builders aren’t trying to beat the model. They’re building systems that expect it to fail, and still work anyway.

Subscribe now

Programming, Events, Content and More

🎙 Podcast Feature: Tech on the Rocks

Tune in to Verissimo’s Nitay Joffe, and Kostas Pardalis founder of Typedef, as they chat with tech geniuses on Tech on the Rocks—where hardware, cloud, and all things future-tech meet over a virtual drink!

Episode 18: Business Physics: How Brand, Pricing, and Product Design Define Success with Erik Swan - https://techontherocks.show/18

🚀 Founder & Community Programming

Reminder to check out our It’s All About Everything Series!

Portfolio Highlights

Celery recently announced a $6.25 million Seed round led by Team8. We wrote their first check and are excited to have participated in this latest round.

Celery is an AI-powered financial control platform that helps organizations catch costly payroll and revenue errors before they impact the bottom line. Designed for finance teams in high-volume, labor-intensive sectors like healthcare, Celery automatically audits thousands of records, flags policy violations, and uncovers unprofitable clients. By shifting from manual oversight to intelligent monitoring, Celery helps teams save time, improve accuracy, and take control of their financial data.

Receive helps small and medium-sized businesses unlock instant access to cash from pending customer payments. Instead of waiting days for transactions to clear, merchants receive a virtual card with immediate funds tied to their receivables. Repayment happens automatically once the original payment is processed, making the experience seamless and hands-free. Offered both directly and through payment providers, Receive gives businesses faster cash flow without disrupting existing systems.

Monto is streamlining enterprise billing by helping companies get paid faster with less manual work. Instead of navigating complex customer payment portals one by one, Monto connects to them all and automates the process of sending, tracking, and collecting invoices. By reducing friction in accounts receivable, Monto improves cash flow and frees up internal teams to focus on growth.

Trace Machina builds foundational software infrastructure for teams working on high-performance, safety-critical systems. It’s platform, NativeLink, is designed to accelerate complex code development, enabling faster builds, better simulation workflows, and significant savings on compute costs. It's especially well-suited for industries like aerospace, robotics, and advanced hardware, where reliability and speed matter most.

🎉 Events

Verissimo NYC Dinner
Incredible night with the Verissimo Ventures community in NYC—founders, LPs, and friends coming together to explore what’s ahead in AI, tech, and venture.
These moments remind us that meaningful relationships are at the heart of everything we’re building.
Grateful for the energy, insights, and shared vision around the table.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We invest primarily in enterprise software companies and take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our $26M Fund 2.

Thoughts on Investing and Starting Up

Anyone who has followed my writing for the last few years knows that I love DCFs as a great way to conceptually map startup characteristics to value drivers through four key variables: short term growth, long term growth, discount rate and free cash flow margin.

Even when I was in investment banking working with companies with $50M+ revenue, it was often hard to accurately perform a DCF analysis to point to a real valuation. In fact, during training at Morgan Stanley, a certain VP said “DCFs are just fun with excel - you can make it say whatever you want.” For investment bankers this is fine because more often than not, you are starting with the answer to the valuation question and then trying to apply rigor to justify that valuation.

The investing world is fundamentally different because buying an asset means that you’re stuck with it and will either need to sell it or illustrate value accretion to shareholders (or risk losing your job). Selling an asset as a broker is a much simpler game: get the highest price you can while still getting a deal done.

With the limitations of DCF analyses for high-growth, unprofitable tech companies, many look to comps multiples for valuation. I have written before about the limitations of the revenue multiple and the fact that it is an “output metric” based solely on comparison of value, but we are going to run with it for now with all of those considerations in the background. In most cases, it’s the best we’ve got.

There is a well-known phenomenon in investing that revenue is worth more if it’s growing faster - this makes sense on many levels, and it manifests in premium revenue multiples. As a result, there is a strong positive correlation between revenue growth and revenue multiple, with profitability playing a role as well (especially over the last couple of years).

This correlation works best with companies that are around the same revenue scale and growth. If revenue scale is far below or growth is far above the median, then the analysis breaks down. If, for example, a company is at $65M revenue scale and is growing 100% annually, then this correlation has limited value and we need to figure out another way to compare it to the public market comps.

I’ll introduce another valuation method (which we used at Morgan Stanley) called the Discounted Equity Valuation Method. (caveat: it’s super hand-wavy and can make a DCF for a highly unprofitable company look rigorous…). But as always, there is a broader point, and while the method may not yield definitive results, it provides a valuable framework for valuing companies without great public comps.

(You’ll see how this gets back to seed stage startups soon, so sit tight.)

We’ll start with example projections of a hypergrowth mid-to-late stage software company:

This company is currently at ~10% of the revenue scale of the 25th percentile and ~10x the median revenue growth of the publicly traded comparables shown above, which makes it very hard to use them as comps. But if we look a few years out, the revenue scale and growth get much closer to where comps are actually trading today. How do we compare 2027 numbers for this company to 2025 numbers for public comps?

Enter: The DEV (Discounted Equity Value) Method

The basic idea:

• Take 1, 2 or 3 year forward multiple on the public trading comps

• Apply it to a 3, 4 or 5 year forward revenue number on the target company (i.e., a 2-year shift)

• Discount the valuation back to present value for the number of years that you shift using a discount rate

This is a very basic version of a DEV analysis. For comparison, this yields current company valuations of $1.7-2.6Bn whereas a simple comps multiple would yield $845M. It is effectively rewarding the company for its aggressive growth prospects. (Note this only works if the projected growth rates are higher than the discount rate.)

Naturally you can play with just about every variable here:

• Comps multiples can change meaningfully based on comps selection and the current market

• Shifts can be 1, 2, 3 or more years

• Discount rate can vary greatly (and is probably the hardest variable to nail down without a reliable way to calculate WACC)

If you were trying to do a rigorous valuation analysis here, I would recommend a bunch of sensitivity analyses. The large range of outcomes that this analysis produces is also a good reason why you likely will see bankers using it much more than investors.

So who cares?

As per my normal theme, I am trying to find fundamentally sound reasons for many of the market dynamics and trends in the early-stage venture market. A strongly held belief is that investing is investing no matter the asset class or company stage: the job is to buy an asset for less than it is worth and then sell when the price corrects (or overcorrects). It doesn’t matter if it’s an ice cream shop, gold bars or a nascent AI company.

A long time scale makes this more complicated. If you are a day-trader and find a mispriced asset, you could be in and out within hours. For an early-stage startup investor, the corrections take much longer and the time-value of money is a much larger factor.

In many ways, early-stage VCs are betting on long-term enterprise value accretion in its purest sense. Sustainable revenue growth is always the key to success. High risk, high reward, low liquidity, long hold periods.

Underwriting at the early stage

Ultimately startups are trying to go from 0 to $1Bn+. But why does it need to go to $1Bn+? Naturally many startups never get there - having middling success or failing completely.

It comes down to underwriting the early-stage investment using the key variables:

• Future Valuation

• Discount Rate

• Time

30% IRR for a VC fund is excellent, but would LPs in an early-stage venture fund get excited about a 1.3x? The simple answer is no. Early-stage venture targets high multiples of capital over long periods of time.

So what is the right mix of these variables to deliver exciting fund-level returns?

I recently met a founder who asked about changing the equation… and aiming lower - he was wondering if a multi-billion dollar outcome was really needed if the chances of success are higher. The theoretical answer is that if the time to exit is shorter and the discount rate is lower, then a smaller target future valuation should be acceptable. But are great VCs getting excited about those sorts of opportunities? The answer is generally no.

Why is the theory different from practice?

• Outliers drive exciting fund returns - the real reason to invest in an early-stage venture fund is that one of the $1Bn target outcomes may end up being a $10Bn outcome and fund-level returns go from a solid 4x to an outstanding 20x.

• Aiming low is deemed a bad signal on founders - their abilities, grit and vision.

• Saying that discount rate is lower for one reason or another is a difficult claim to make.

• Most investors want it all: de-risked founders & markets, incredibly high upside, medium time to exit - this becomes the optimization equation.

What is the key takeaway?

This has been a very long-winded explanation behind why “great markets” and “great founders” demand higher valuations even at the earliest stages - it is because they are deemed higher upside and lower discount rate.

What is the opportunity?

Arbitrage! Of course, that is the point of investing - buy underpriced assets and sell them when price corrects. When it comes to both founders and markets, we need to look for people and opportunities where the conventional market approach will undervalue them - only later on will the market recognize how great they and their markets really are, correcting valuation.

There is another popular approach: swing for the fences. Pay whatever you need to and make sure you’re absolutely getting into the best markets with the best founders and hope that you add an extra 10 turns on the return - anything else will yield mediocre-at-best returns. This approach can work in certain market environments, at certain stages and for certain investors. It is not our preferred method or approach.

Overall, we want to make sure that we optimize our ability to invest into the next cohort of $10Bn companies, but also that our returns will be outsized when that happens. We rely heavily on our network and pattern recognition to make this a reality.

We’re investing in great people from our network who fit our ideal founder profile and are bringing their energy to exciting markets with fundamental underlying shifts afoot. We believe these are the greatest opportunities of maximum returns. Some of these investments might seem weird or counter-intuitive, but at the pre-seed and seed, we believe strongly that that is the job!

Subscribe now

Programming, Events, Content and More

🎙 Podcast Feature: Tech on the Rocks

Tune in to Verissimo’s Nitay Joffe, and Kostas Pardalis founder of Typedef, as they chat with tech geniuses on Tech on the Rocks—where hardware, cloud, and all things future-tech meet over a virtual drink!

Episode 17: Incremental Materialization: Reinventing DatabaseViews with Gilad Kleinman of Espio -

https://techontherocks.transistor.fm/17

🚀 Founder & Community Programming

Reminder to check out our It’s All About Everything Series!

Portfolio Highlights

We’re excited to announce our investment in WatchWell for Fund 2.
Founder and CEO Jonathan Weber impressed us early on, and his strong ties to the luxury world — alongside a major U.S. dealer as co-founder — position WatchWell to lead innovation in this space.

WatchWell is building the modern, technology-driven companion for luxury goods. In response to rising theft, WatchWell is developing a discreet, embedded geo-tracking solution for luxury watches, a "LoJack for watches." Beyond theft recovery, the technology unlocks new possibilities for asset tracking, digital engagement, and streamlined insurance claims, offering a smarter, more connected ownership experience.

Airvet recently announced a $11M Series B-2 round led by HighlandX.

Airvet provides 24/7 virtual veterinary care through a platform that connects pet owners to licensed veterinarians via video and chat. Core features include immediate access to urgent care consultations, behavioral support, and preventative wellness advice without the need for in-person visits. Airvet also partners with employers to integrate pet telehealth as a workplace benefit, supporting employee wellbeing and reducing lost productivity.

Botika uses AI to revolutionize product photography in the fashion industry, offering a full-stack platform that enables retailers to plan, create, deploy, and optimize high-quality, scalable visual content in real time. Powered by custom foundation models, Botika provides a cost-effective way for fashion brands to generate on-brand imagery for their websites—forming the creative backbone of modern eCommerce.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We invest primarily in enterprise software companies and take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our $26M Fund 2.

Thoughts on Investing and Starting Up

When it comes to large funds making investments into seed stage companies, everyone talks about two main issues:

1. Signaling risk. If that fund doesn't follow on with a core investment, then it tarnishes the company for the rest of the market.

2. Incentive alignment. They are not economically incentivized to actually care about the company. A $3M check simply will not move the needle for a $1B+ fund.

But there is a third set of issues which may be even more problematic:

Involvement in the company - which includes:

1. Stage-specific advice

2. Attitude toward the status of the business

Stage-specific advice can be a real problem in one very clear, specific situation:

Giving genuinely great advice on how to grow a company/business before it is ready to do so can be totally detrimental to the mid and long term prospects for the company.

Premature growth is like lifting heavy weights before age 12... it can result in seriously stunted growth.

Attitude towards the status of the business may even be a bigger issue:

Once an early-stage investor has placed their bet, they (should) go from being a diligencing investor to a helpful teammate and advisor. This transition is fundamental to the relationship between founders and their investors. Once you're in a company, you ought to be in it to win it. Your job is now to do whatever you can to make this company successful. The problem with large funds taking option bets into seed stage companies is just that: they are option bets ahead of potentially much larger checks later on. The result is that those investors do not make the transition from diligencing investor to teammate after they wire the money - they are still a diligencing investor… except now they’re insiders. The feedback they give and the questions they ask will all be meaningfully colored by their role being fundamentally rooted in diligencing the company for a future investment.

While there is value to that critical eye in the room, too much of it often adds undue pressure to founders, preventing founders from expressing the sort of confident vulnerability which allows them to engage their stakeholders in the most productive way.

Even in a more positive light, their advice will always be directed specifically toward how the company can become a core investment for their fund, regardless of whether or not that still makes sense (because it is a KPI for any seed investment program in a large fund).

A lot changes in the first 2 years of any company - these stories are much more variable and complicated than what gets shared publicly. Having the sort of unwavering support, focus, economic alignment and stage-specific advice can be the game changer for founders.

So my advice to founders who are getting approached by large funds to lead small rounds: let those rounds be led by stage-appropriate investors, and let the big guys make their option bets alongside that (if they so desire).

A massive fund owning 20% of your company for $3M creates a very unstable situation which does not stand to benefit founders or investors. If it makes sense, let them own 5% and buy up to 20% in the next round—that’s their core strategy anyway! Alternatively, if they’re doing what seems like an excessive amount of diligence for what will be a small check for them, tell them you’re focused on raising from seed funds and should schedule a call for 9 months out. This will give them some relief that their multi-stage competitors are not taking the deal either, and then they can potentially invest in their sweet spot down the road.

In any case, founders should be asking VCs: “What percentage of your fund are you investing into this round?” This metric matters a lot.

Programming, Events, Content and More

🎙 Podcast Feature: Tech on the Rocks

Tune in to Verissimo’s Nitay Joffe and Kostas Pardalis founder of Typedef, as they chat with tech geniuses on Tech on the Rocks—where hardware, cloud, and all things future-tech meet over a virtual drink!

Episode 15: Apurva Mehta - Reinventing Stream Processing: From LinkedIn to Responsive with Apurva Mehta - https://techontherocks.show/15

Episode 16: From Data Mesh to Lake House: Revolutionizing Metadata with Lakekeeper - https://techontherocks.transistor.fm/16

🚀 Founder & Community Programming

Reminder to check out our It’s All About Everything Series!

Portfolio Highlights

Lenkie announced a Series A consisting of a £4 million equity raise and an additional £45 million debt facility, led by a large US private credit fund.

Lenkie provides businesses with upfront financing to fulfill new customer orders. Its technology platform evaluates creditworthiness in minutes by securely analyzing banking data and trading partner profiles. This automated approach enables fast, flexible loans—helping companies manage cash flow more efficiently as they scale.

Alta recently announced its $7 million seed funding alongside its official market launch. The round was led by Entrée Capital and Target Global—we were the third largest investor.

Alta provides an AI-powered workforce that streamlines sales and revenue operations. Its specialized AI agents—Katie for personalized outreach, Luna for analytics and insights, and Alex for automated calling—integrate seamlessly with existing sales, marketing, billing, and collaboration tools. By automating routine tasks and delivering actionable insights, Alta helps teams accelerate growth and optimize revenue performance.

Novella is a digital insurance wholesaler revolutionizing Excess & Surplus (E&S) insurance through AI-powered solutions. Its technology transforms the traditional quote-to-bind process by ensuring brokers submit complete, accurate information upfront—streamlining operations and cutting processing times from weeks to days.

Curvenote is a web-first publishing platform transforming how researchers create, share, and communicate scientific work—moving beyond static PDFs. The platform combines an intuitive authoring environment—supporting Markdown, interactive figures, and real-time previews—with powerful automated integrity checks and computational reproducibility. Trusted by prestigious organizations including the American Geophysical Union and Microscopy Society of America, Curvenote is bringing scientific authoring into the 21st century.

Novodia delivers fast, affordable, and high-quality custom curricula by combining proprietary pedagogical AI with subject-matter expertise. It bridges the gap between school districts' demand for vetted, standards-aligned content and teachers' need for dynamic, engaging classroom experiences that reach every student.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We invest primarily in enterprise software companies and take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our $26M Fund 2.

Subscribe now

Lots of updates and thoughts to share this month, so bear with us until the end!

Thoughts on Investing and Starting Up

It's been more than two years since ChatGPT exploded onto the scene—and our collective consciousness. Its pace of adoption has been among the fastest of any digital “app” in history, a particularly remarkable feat given that we are talking about the adoption of a new category of technology—the data suggests that U.S. adoption of generative AI has been faster than adoption of the personal computer and the internet.

This viral and shocking spread—the good folks at OpenAI famously launched ChatGPT as a low-key research preview—has led to an interesting dynamic: while most people “know” what generative AI is, many users’ experience is so limited and unproductive that they end up with a (very) false sense of what Large Language Models (LLMs) are currently “capable of” and where they are headed.

Here’s just one example:

Today we’ll explore:

1. The two main usage types

2. Which type most lends itself to investment

3. The killer app formula

The bottom line:

• Large Language Models (LLMs) create value in two ways:

  • as knowledge repositories ("Genies")

  • as task executors ("Golems")

• The greatest opportunities likely lie in workflow-integrated applications, where LLMs (a) enhance existing workflows or (b) unlock new workflows that were not previously possible.

• Successful implementation typically requires domain expertise of the existing human-based workflow to select the best opportunities, insight to identify the creative improvements, and technical and product expertise to engineer the right solutions.

Let’s dive in…

I. The Two Main Usage Types

We find it useful to break down LLM utilization into two core usage archetypes: (A) Knowledge Interaction (it tells you things) and (B) Generative Tasks (it does things for you).

Let's explore these in detail…

A. Knowledge Interactions—The Genie

As the name would suggest this refers to interactions where the LLM provides users with information. This use case can span everything from learning a new skill from scratch and studying a new topic in depth (e.g., "What is General Relativity") to a quick lookup of how to solve a basic problem (e.g., “how do you accomplish X in Excel”). One incredibly useful example that I’ve personally used is uploading complex technical documents, academic papers, insurance forms, legal contracts, etc.—and having the LLM “break it down with you” as you read along. Try it, it will change your life. :)

Here we will cover three key points:

1. LLMs are incredibly knowledgeable

2. They suffer from important challenges

3. Nevertheless, they are useful even in their current form

1. LLMs are incredibly knowledgeable

Due to how they were trained, today’s LLMs contain the vast majority of humanity's accumulated knowledge, and are so capable at leveraging that knowledge that they are outgrowing the tests they are being evaluated against. We are literally running out of benchmarks to test new AI models on.

While benchmarks are certainly not a perfect measure (and most suffer from a number of known flaws), they are a good indicator of LLM’s capabilities, and feedback from real-world usage and exams both support this point.

2. LLMs suffer from important challenges

1. The non-determinism problem (they’re not reliably consistent): LLMs possess remarkable capabilities but struggle with consistent reliability. This gap manifests when models demonstrate impressive knowledge in one instance but fail unexpectedly in similar scenarios. While an LLM might brilliantly explain quantum mechanics one moment, it could confidently misrepresent basic physics concepts the next—without any indication of reduced confidence.

2. The hallucination problem (they make stuff up): LLMs are fundamentally prediction engines that generate text based on statistical patterns rather than understanding truth. This leads to hallucinations—instances where models generate false information, fabricate references, or create entirely fictional narratives with complete confidence. This tendency to confabulate poses significant challenges in professional contexts where accuracy is paramount.

3. LLMs are useful even in their current form

There are a couple of stories that help illustrate how LLMs are useful even with these challenges:

• A few months ago a particularly shocking story came to our attention. Some kid used a LLM to build a nuclear fusor in his bedroom – with zero hardware experience – in a couple of weeks.

• Another, more boring yet perhaps more illustrative case is a study at Harvard which compared physics students who were assisted by an AI tutor against students who participated in a more “traditional” human-led, active learning classroom, and found that: “when students interact with our AI tutor, at home, on their own, they learn more than twice as much as when they engage with the same content during an actively taught science course, while spending less time on task”—while feeling more engaged and motivated to boot.

More generally, the reason people are finding so much success is twofold:

• Models are getting better - hallucinations are increasingly not a problem for knowledge-based tasks.

• Users are becoming more skilled at working with their strengths and managing their weaknesses.

With regards to hallucinations specifically, our general sense here (supported by anecdotal experience and observation—and corroborated by other informed observers) is that that the typical hallucination one imagines, the kind where the model makes things up is becoming less and less common—and that a savvy user learns to recognize what types hallucinations are likely occur, when and how they are likely to occur, and they best ways to mitigate against them.

The inflection point seemed to occur during 2023 when a plurality of users became capable of deriving meaningful value and the incredible advancements that have since followed have further raised their reliability, usability and raw capability.

Adam Brown of DeepMind, and a theoretical physicist at Stanford, summarized the current state of Knowledge Interactions well:

The other thing that they're extremely useful for now that they were useless for before is just as a tutor. There is a huge amount of physics that a physicist would be expected to know that has already been done. No human has ever read the whole literature or understands everything, or maybe there isn't even something that you feel you should understand, or you once understood that you don't understand.

I think the very best thing in the world for that would be to phone up a colleague, if you knew exactly who to phone, they'd probably be able to answer your question the best. But certainly, if you just ask a large language model, you get great answers, probably better than all but the very best person you could phone. They know about a huge amount, they're non-judgmental, they will not only tell you what the right answer is, but debug your understanding on the wrong answer.

So I think a lot of physics professors are using them just as personal tutors. And it fills a hole because there are personal... If you want to know how to do something basic, it's typically very well documented. If you want to know advanced topics, there are not often good resources for them.

LLMs have tremendous utility in the Knowledge Interaction capacity as either teachers or oracles (give a fish or teach to fish). The main thing now holding back greater real-world impact is user creativity, awareness, and skill.

B. Generative Tasks—The Golem

As this name suggests, this refers to having an LLM do something useful. There are three basic categories of Tasks that LLMs can do:

1. Language

2. Data

3. Processes

In this section, we will define each one and discuss its impact

1. Language

These are the most well-known use cases (e.g., summarization, first drafts, translations, Q&A, etc.). Virtually all of these use cases are “horizontal” in nature - i.e., they cut across industries - and hence are getting incorporated into seemingly every existing software tool. Despite the widespread adoption of this form, it adds relatively little value compared to the other types (with the exception of code).

2. Data

LLMs enable users to query data, enabling executives, employees, and customers to interact with relevant information in useful ways. The basic idea here is giving an AI access to a quantitative knowledge base (internal or otherwise). Doing so unlocks the ability to query a company's (proprietary) unstructured data in natural language and enables everything from automated customer support to a transformation in internal visibility (BI+). While we do believe that the enterprise knowledge bot offerings will deliver real value—the space already has a number of serious players with real traction—we suspect that there is somewhat of a ceiling here and that in the long run this category will account for a comparatively small fraction of the overall Generative AI market.

3. Processes

Users are able to enhance processes by interweaving LLMs into them. These are the most useful Tasks that LLMs can accomplish. They tend to be highly context specific, usually verticalized, non-obvious, and paradigmatically new in that they require real creativity and care in their design. This third category is where a lot of the “power users” are gaining their productivity increases, and perhaps more interestingly (and scalably) where most of the (very few) third party killer apps are being created. In practice, domain expertise/real experience with the workflow is usually required for success. Some eclectic real world examples include: historical text transcription and analysis, “help me beat a car rental damage claim,” or “see what CEOs said about the labor market in their conference calls.”

II. Process-based “Tasks” lend themselves most to investment

A. Knowledge Interactions (“The Genie”) do not easily lend themselves to investment.

Ultimately, we think Knowledge Interactions will not be the main investment space.

The question of investability of third party apps in the Knowledge Interaction space is quite a complicated one (and worthy of its own deepdive) and the answer varies based on specific context, but high level, we believe a few things are true here:

1. Most of the “tricks” that are used to help with hallucinations don't solve the fundamental problem—enabling you to blindly trust the AI’s output, and are often no better a mitigation strategy than a quick follow-up Google search

2. The mechanisms to try to solve the problems tend not to scale beyond their immediate use case. Inevitably the vast majority of potential LLM interactions (literally most of humanity's cumulative knowledge stores) will simply never be “reached” by tools—and this holds for most potential UI/UX improvements as well.

3. We expect most of the value creation here to accrue to the big model providers directly, and not third party applications—though there will be exceptions of course (AI-powered TurboTax?). Tools may be necessary now, but they won’t be in the future. The Genie use case is in most ways the core of what LLMs are, and so, as intrinsic capabilities continue to increase (which we are optimistic about) so too will the LLM’s value itself.

B. Generative Tasks (“The Golem”)

Language and Data “tasks” suffer from key weaknesses when it comes to building investable products. As noted, the language tasks are of relatively little use when compared to the other categories, and the data tasks space is already saturated. In contrast, process tasks don’t suffer from these weaknesses, and therefore will continue to be the most investable opportunities.

Process tasks have outsize impact and are limited in their current development. Significantly, they also require third-party tools to deliver their maximal impact, as most users are not able to reach maximal impact in their private use. There are at least three key reasons why this is so:

1. Improved UI/UX: APIs can streamline processes beyond the more basic interface that LLMs currently allow (e.g., chatting, copying-and-pasting).

2. Expanded capabilities: LLMs are pretty bad at complex, real-world open ended reasoning—and as such are quite bad at thinking at an abstract or high level. They can execute well for narrowly defined tasks but if you try to give it real responsibility like an AI software engineer or AI scheduling assistant, they will quickly go off the rails. Hard-coding a reasoning layer provides a possible solution.

3. Better error-checking: Tools can catch or otherwise mitigate hallucinations and non determinism at a level that humans cannot. Here too a coded reasoning layer (rules, plain old software) to manage this variability offers a possible solution—indeed this is an absolute necessity for the vast majority of LLM-powered apps (when one does choose to go the tool route), as lack of certainty about how your application will behave (i.e., the chance that your system will simply invent data/do something unexpected) is a feature fundamentally incompatible with mature, real-world usage.

III. The Killer App Formula

The most successful apps tend to democratize the value of LLMs, even beyond what power-users can achieve because of digitization and automation. Most killer LLM-powered apps we've seen seem to follow the same basic structure:

1. Select: Take an existing workflow—usually one that is some combination of complex, time-consuming and annoying.

2. Diagnose: Identify how timely and specific calls to a LLM carefully interspersed throughout its lifecycle can improve it.

3. Engineer: Construct an engineering layer that authors the workflow—a UI that guides users to work the way you need them to—and a hard-code a reasoning layer that calls the LLM at the right time and in the right way—and voilà, you get magic.

A great example of a killer app following our formula is perhaps the most popular LLM-powered app we know of, Cursor the AI powered integrated development environment (or code-editor). The genius of Cursor is two-fold: (1) understanding that LLMs can give extremely useful coding/formatting suggestions that can incorporate context – a new dimension of autocomplete, and (2) understanding that UX is what will make or break the value proposition here—and nailing it.

Looking into the future

We believe that LLMs will have the major economic impact many are forecasting (though we can quibble about magnitude, and we don't agree with everyone). A major component of their value will be delivered via third party tools (the “Application Layer”) that productize these magic use cases, and that this category will expand over time to include entirely new workflows centered around this magic LLMs can do.

Even though LLMs are improving, they’re not getting better in the three main ways covered above that would make tools obsolete (i.e., enhanced UI/UX, expanded open-ended reasoning, and reliable error-checking). Going forward, we will be looking for products that augment workflows and founders that are likely to identify these most beneficial use cases.

A version with footnotes and sources can be found here.

Subscribe now

Programming, Events and More

We are super excited to have kicked off our “It’s All About Everything” Founder Series, with our first Expert Sessions and Zone of Genius Cohorts beginning next week. You can learn more here: https://verissimo.vc/its-all-about-everything-founder-series

Last week, we hosted an incredible evening at the Tasting Room in Tel Aviv, bringing together founders, investors, and key players in the tech ecosystem for a night of great conversations over wine and cheese. It was amazing to see so many familiar faces and new connections being made.

Stay tuned for more gatherings soon! 🚀🍷

🎙 Podcast Feature: Tech on the Rocks

Tune in to Verissimo’s Nitay Joffe and Kostas Pardalis, founder of Typedef as they chat with tech geniuses on Tech on the Rocks—where hardware, cloud, and all things future-tech meet over a virtual drink!

Episode 13: Yaron Hadad - From black holes to AI in mathematics: AI Innovation in Mathematics and Health - https://techontherocks.show/13

Episode 14: David Jayatillake - Semantic Layers: The Missing Link Between AI and Data - https://techontherocks.show/14

Portfolio Highlights

PromptLayer recently announced a $4.8 million Seed raise led by ScOp Venture Capital. We participated alongside Stellation Capital and several Angels.

PromptLayer is a centralized prompt management platform designed to streamline the deployment of prompts in LLM-powered applications. PromptLayer's platform enables teams to store, organize, iterate, and assess prompts in a single location, collaboratively. PromptLayer’s easy-to-use interface empowers non-technical domain experts to play a key role in the development of AI applications.

Flare’s software platform simplifies the delivery of complex professional services while addressing consumers' expectations of ease and clarity throughout the process. Their platform automates time-consuming administrative tasks and streamlines client communication, freeing lawyers to concentrate on their core expertise—practicing law.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We invest primarily in enterprise software companies and take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our $26M Fund 2.

Thoughts on Investing and Starting Up

One of the hardest parts about early-stage investing is that you never truly know how well your fund is performing prior to liquidity. Contrast that to trading stocks in the public markets: at the end of every day, you know the exact value of each of your positions. This knowledge can be more stressful or less stressful depending on your personality. But we get back to the same issue… even on a quarterly or yearly basis, it’s hard to know how an early-stage fund is performing.

The prevailing metric is TVPI (“Total Value to Paid-in” Capital). But this begs the question:

The standard method for early stage companies is to just mark up to the last preferred share price. The public market-based comps method would in theory be better, but applies much better to later stage companies who have the sort of metrics that are more easily comparable to a public company.

For example, a company with $100M of revenue, burning $20M per year and growing at 50% might trade at 10x forward revenue, which is pretty straightforward, and you could readily apply this same method to a company at ~$25M of revenue, burning $20M per year and growing 150% using a correlation analysis. But what about a company at $2M of revenue, burning $12M per year and growing at 400%? This method starts to break down quickly.

But even if you could use this method, while a company is private it as a capitalization stack of preferred stock. Preferred stock can have debt-like and other unique features that create large value discrepancies between Series D shares and Series A shares, for example. Many secondary funds are great at running the sort of analyses on liquidation preferences, cap stacks, ratchets, etc. that are needed to really understand the difference in value between a Series A share and Series D share or a common share. But with a relatively small capitalization stack, still just a couple of years into the game, marking companies is much more arbitrary than we would all like to believe.

We’ve got to remember that all of these methods are really just a proxy for understanding absolute value accretion. Venture is meant to be an uncorrelated asset class where the only thing that matters over long periods of time is value accretion driven by profitable growth (or at least the potential for it). And of course, getting lucky with a low interest rate exit market many years down the road.

The result of all of this is that “valuation stories” tend to attract all the attention, acclaim and focus for early-stage investors (and their LPs and the media), rather than what really matters: sustainable revenue growth. (Note that very few funding announcements include revenue numbers and almost all of them include capital raised, and several of those include valuation information. Might it be because there is a big disconnect? I don’t buy the argument that it’s because they would be disclosing somehow valuable confidential information.)

There is an argument to be made that when a company gets “written-up,” that should be validation of the progress of the underlying business, and map to the absolute value accretion. One big problem: free market dynamics are not at play in venture. One single investor can write up a company big time - and they just may be right - but it certainly does not represent the market. Unlike the public markets, there is no mechanism to short stocks, which balances this out, so you still really don’t know if a markup means progress or just the perception of progress by one person or fund.

So now I’ll make my aggressive argument:

Lower valuations (within reason) are better for everyone until exit. Investors, Founders, Employees, Acquirers. (The only losers are journalists who don’t get to knight another unicorn for their useless market reports.)

Why?

As an early-stage investor pretty aligned with founders, we have to remember that write-ups look and feel good, but they may not actually be good for any of us.

Since 2013, I have invested in well over 100 startups. Having a front row seat to all of these stories has given me a real (often painfully earned) appreciation for modest valuations (relative to capital raised, of course).

I would like to argue that write-ups actually represent Beta (volatility) rather than Alpha in venture because of the preferred-stock dynamics. I’ll illustrate with an example:

In this case, a company “gets hot,” and raises a big up-round: the founders feel rich, the early-stage VCs look smart - everyone is happy. But regardless of how well the business is doing, they might be looking at a long period of share price stagnation despite excellent progress due to market dynamics completely out of their control. (No responsible VC investor will look at a company that raised at $100M with $1M of revenue in 2022 and say “well now they’re at $5M, so it must be worth $500M now in 2025.”) In the above case, the company raised a down round at a lower share price - exactly what happens to founders, employees and early-stage investors is totally dependent on the details of the closing docs. It could be inconsequential or it could be a total wipeout.

The consequence of not clearing this hurdle? See the middle 15 pages of your closing documents - it can be extensive, complicated and brutal. So even though you might be thinking, “well I’ll take the good market price now and deal with it later.” It is unfortunately not that simple and introduces cram-downs, recaps, ratchets, and other nasty down-round related terms which get buried in the docs.

Everyone would do better if that same company had a share price trajectory that looked like this:

So we’ve got to remember that write-ups feel good in the moment, but immediately after closing, there is a new, very real hurdle to clear, often with dire consequences for missing it.

Events & Content

🎙 Podcast Feature: Tech on the Rocks

Tune in to Verissimo’s Nitay Joffe and Kostas Pardalis, founder of Typedef as they chat with tech geniuses on Tech on the Rocks—where hardware, cloud, and all things future-tech meet over a virtual drink!

January 2025 Episodes:

Episode 12: Building a Native Search Engine in PostgreSQL: ParadeDB's Journey to Replace Elasticsearch Featuring Philippe Noël

Episode 11: Optimizing SQL with LLMs: Building Verified AI Systems at Espresso AI Featuring Ben Learner

Portfolio Highlights

Aligned recently announced an $8M Series A round led by JAL Ventures. We were proud to be the first fund to back Aligned over three years ago, and have since watched them grow into a category-defining platform. We couldn’t be more excited about their trajectory and the tremendous potential ahead.

Aligned is a B2B customer collaboration platform designed to help sales teams boost transparency, reduce friction, and close deals faster. The platform offers a shared digital workspace where buyers and sellers can align on goals, track progress, and manage deal-related content seamlessly. Key features include Mutual Action Plans for structured deal execution, AI-driven insights to identify buying signals and optimize engagement, and smooth integrations with CRMs and sales tools. Additionally, the AI can autonomously engage with customers, answering queries and providing real-time support, enabling the sales process to continue seamlessly even when the sales team is unavailable.

Inigo provides complete visibility and control over production GraphQL APIs, empowering organizations to confidently adopt and scale GraphQL. Its essential tools include deep API analytics, schema-based role-based access control (RBAC), performance and error monitoring, dynamic rate limiting, and safeguards against schema breaking changes. Inigo delivers unique, in-depth insights into GraphQL usage, from field-level details and query paths to overall server health and performance. Its powerful engine enforces policies, blocks malicious queries before they reach GraphQL servers, and automatically identifies and resolves API issues.

NemoDATA is building an AI-powered predictive maintenance platform for vehicle fleets that analyzes sensor data to prevent failures. Seamlessly integrating with fleets’ existing telematics and data, NemoDATA’s proprietary virtual sensors help identify potential breakdowns before they happen, improving vehicle health, minimizing breakdowns, and increasing uptime.

Lighthouse is transforming how people rent, buy, and sell homes by cutting out the middleman and partnering directly with property suppliers. With a cash back model and flexible options for browsing or working with a specialist, Lighthouse makes the process of finding a home simpler and more rewarding.

Namespace is a purpose-built compute platform designed to accelerate development workflows by significantly reducing build and test times. It integrates seamlessly with tools like GitHub Actions, Docker builds, and Bazel, enabling rapid deployment of ephemeral instances in just a few seconds. Features include built-in Docker image caching, high-performance remote builders, and scalable infrastructure capable of handling numerous concurrent builds. Additionally, Namespace offers end-to-end previews, out-of-the-box observability with metrics and logs, and a private container registry for secure image storage. By optimizing performance and providing cost-effective solutions, Namespace enhances developer productivity and efficiency.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We invest primarily in enterprise software companies and take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our $26M Fund 2.

Subscribe now

Thoughts on Investing and Starting Up

We are excited to kick off 2025 with the launch of our “It’s All About Everything” Series for our portfolio founders. This series is aimed at achieving two main goals:

1. Working with founders to identify and get the most out of their “Zone of Genius.”

2. Providing founders with an expert-led survey course of each core operating function inside a company.

We believe that each of our founders has something special that they bring to their company and market, which is why we invested. After investing in over 120 companies over the last 11 years, I have come to realize that blind spots and distractions create some of the biggest barriers to success for early-stage companies.

My job as an early-stage investor is to pick great founders who have limitless potential and then do everything I can to help them avoid failure.

They will succeed because of their own talents, work ethic and market timing, but there are so many ways that they can fail. As an investor who has seen quite a few startup journeys, it is my duty to share the common pitfalls with founders and help them navigate accordingly.

Too often I hear “it’s all about growth” or “it’s all about recruiting” or “it’s all about product.” Scaled success requires the ability to focus on everything. This means leaning into strengths, supplementing weak areas with experts, and knowing exactly when each function should take the spotlight. Juggling this complexity is the hardest part of the founder job, so our aim is to give founders the tools to “focus on everything.”

A great way to avoid unnecessary mistakes is to understand the core concepts behind each function in a business. This is where repeat founders often excel: they know what comes next in the company building process, and therefore when they need to shift their focus from engineering to product to growth to recruiting to competitive strategy, etc. This instinct is undoubtedly what makes an outstanding and scalable founder.

We aim to help founders identify previously unknown blind spots and then solve for them accordingly. In this two-step process, and we aim to provide both the expertise and community to help founders avoid pitfalls and achieve their ultimate potential.

We are excited partner with my good friend David Rust to launch this new program for our founders.

Share

Events & Content

Alex recently spoke at an event hosted by Ameer Awadiyeh, CEO of Hasoub, a pioneering organization in Israel that is empowering Arab technologists and entrepreneurs, at the Google for Startups campus. In his talk, titled Leaning Into the Subjectivity of Early-Stage Investing, Alex shared insights on developing an investment methodology tailored to the unique challenges and opportunities of early-stage ventures. The event offered a great opportunity to discuss early-stage investing and connect with a community driving innovation in the Arab tech ecosystem.

🎙 Podcast Feature: Tech on the Rocks

Tune in to Verissimo’s Nitay Joffe and Kostas Pardalis, founder Typedef as they chat with tech geniuses on Tech on the Rocks—where hardware, cloud, and all things future-tech meet over a virtual drink!

December 2024 Episodes
🔟 Episode 10:
Security as Code: Building Developer-First Security Tools
Featuring David Mytton of Arcjet

9️⃣ Episode 9:
Dev Environments in the AI Era: Standardizing Development Infrastructure
Featuring Ivan from Daytona

Portfolio Highlights

Droxi recently announced a $21M Series A led by Drive Capital. We are strong believers in their vision and are excited about the awesome opportunities ahead.

Droxi's AI saves doctors time by showing them the most relevant parameters for each item in their EHR inbox, eliminating the need for manual searches, and clicking between multiple windows. Droxi is for every clinician who wants to spend more time treating patients and less time clicking buttons.

We are excited to co-lead the pre-seed round into Day 8. Day 8 upcycles discarded agricultural leaves into Rubisco protein. Their powder has neutral taste and color, is allergen-free, and functionally superior to whey and pea protein. In a world increasingly driven by the need for sustainable and reliable protein sources, Day 8 is paving the way for a healthier, more resilient food future.

Trullion is an AI-powered platform that automates workflows for finance and audit teams. By automating tasks like data entry, report generation, audit processes, and financial analysis, Trullion helps businesses of all sizes increase efficiency, accuracy, and scalability in their accounting operations.

Trullion was recently named one of North America’s 15 fastest-growing companies in Deloitte’s 2024 Technology Fast 500.

Komodor is a platform for managing Kubernetes at scale. It provides continuous monitoring, analysis, and visualization of the Kubernetes environment, offering clear insights to simplify reliability maintenance, real-time troubleshooting, and cost optimization. By bridging the Kubernetes knowledge gap, Komodor empowers teams to improve operational efficiency, and accelerate development.

OurRitual is making relationship coaching accessible to everyone through a dynamic digital coaching platform. Their app combines practical skill-building with guidance from human experts, powered by cutting-edge AI and backed by relationship science.

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We invest primarily in enterprise software companies and take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of our $26M Fund 2.

Thoughts on Investing and Starting Up

There is a line that I often find myself telling founders before they head into a new fundraising process: are you rehearsing or training?

I think there is a key difference in the approach that can be summed up with a simple understanding of who does each:

• Actors, singers, dancers rehearse

• Navy seals, basketball players and boxers train

So, as a founder, you need to ask yourself: Who are you more like? Who do you aspire to be like?

The answer is clearly those who train. Why? Because they are dynamic, interactive professions which require constant adjustment, and the only way to handle the infinite situations they might come up against is to train the mind and body to be prepared for whatever situation comes your way. Training is more about building instincts than anything else. As Mike Tyson famously said: “everyone has a plan until they get punched in the mouth.” Being an entrepreneur has some painful similarities to boxing, so just rehearsing all the moves in the ring before the fight isn’t going to result in many wins.

Now you might ask: don’t actors, singers, and dancers also train? And the answer of course is yes, they do train, but the last step and where they spend time before go-time is in rehearsal, getting every detail perfected before they go on stage. While there is a lot of overlap in these two approaches to practicing, there is also a distinction. And of course, there are those who think that the only real practice is game time itself.

So my advice to founders heading into a fundraise process is simple: if you want to look good, make sure you actually are good, and get good through training.

So what is the training?

Use the questions, concerns and potential challenges from VCs as fodder for your training - make sure you have good responses, but don’t rehearse and try to answer everything upfront in a magical perfect pitch. The best pitches aren’t pitches at all, they’re simply exciting descriptions of what you have accomplished and how that validates what you plan to do in the future.

The real work happens in developing the business, iterating and learning how every detail of the operation is informing future decisions. It is so easy to get bogged down as a founder “just trying to execute” - putting out fires, responding to customers, hiring & firing, pushing product updates, running marketing campaigns, etc. But sitting down for a proper strategy session that outlines how all of these things come together is a critical piece of the puzzle. Having detailed, strategic and vision-level expertise of the business will impress any investor (or customer, employee or partner) more than using just the right vocabulary to outline the market opportunity, product, etc.

One way to visualize this is the airplane metaphor: it’s either at 30,000 feet or on the ground most of the time, but ask any pilot and they’ll tell you what the most interesting and challenging part is: takeoff and landing. The old adage is that investors and advisors can offer the 30,000 foot view, and a great CEO will effectively oscillate between the ground and 30,000 feet: tactical execution and high level vision for the company. But there is a critical piece in between that is missing: understanding what to do at 3,000 feet or 15,000 feet on the way up. This level of detail is incredibly strategic and is vital to a company’s sustained ascent. Great investors and advisors won’t just help you refine the vision, but instead help map that vision to a medium level of detail for operating the the business. It is also the critical step that helps ensure that the detailed execution and the high level vision are actually in alignment.

Naturally, my favorite example of this is somewhere in the 15,000 foot range and involves building a model to quantify the mechanics of the business and make sure that operating assumptions will drive efficient growth. A slightly more detailed example, say 5,000 feet, would be a marketing channel strategy and rubric for measuring the success across various channels.

I don’t recommend that founders spend all of their time in these exercises, but that they devote the necessary time to each. One way to do this effectively is at board meetings. If you have a diverse set of investors who bring a variety of expertise, simply by talking through the steps between detailed execution and high level vision, you will bring out the best wisdom from the people around the table and take the business to the next level.

To sum this all up:

• Train, don’t rehearse

• Look good by being good

• Have board meetings (even if you don’t have a board)

• There is a strategic level between vision and execution that requires a village of experts to optimize

Subscribe now

Portfolio Highlights

OpenRoute (formerly Milio) leverages AI to simplify the analysis and interpretation of employee data. Their system empowers leaders to make strategic, data-driven HR decisions without requiring extensive training or disruption to existing workflows.

Celery has a suite of AI-powered tools that help payroll teams reduce costs and minimize risk to the bottom line, identifying discrepancies, inefficiencies and trends that aren’t visible in a spreadsheet.

Supermove is a vertical software company for the moving industry. Supermove’s easy-to-use platform enables moving companies to transition to paperless operations in minutes, not weeks. Their platform offers scheduling, billing, real-time reports, and various other workflows.

Substrata is a social intelligence AI that analyzes real-time human dynamics and communication to improve tone, balance, and effectiveness. By addressing gaps in sales and customer service interactions, Substrata enhances communication quality and leads the next wave of real-time, multi-channel conversation analytics.

Share

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We invest primarily in software business models and take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of Fund 2.

Thanks for reading Verissimo Ventures Monthly!

Thoughts on Investing and Starting Up

There has been an emphasis in the last several years in the venture ecosystem around access, or getting into great companies, but there is a much more important part of the business: exits, or getting capital out of great companies.

Of course when LPs are investing in funds, they are focused on what managers will do, and that is invest into new deals and manage their portfolio. But the rubber meets the road at exit, which is often many years down the road in early stage venture investing. Of course without massive value accretion in the portfolio, being good at managing exits will simply never matter, but when it does come to that point, it’s a vital part of the business. Put another way, you need to make sure that you win in your winners.

Price Targets and Analyst Ratings

Traditionally equity research analysts post a price target and a rating for a company. The three options for ratings are: BUY, HOLD and SELL.

Based on the perceived difference between the current price and the price target, the analyst will offer a strong or weak recommendation on the stock.

In the public markets, any investor can choose to BUY, SELL or HOLD their position or a portion of it, and these options are always on the table. While quarterly earnings and other news reports tend to draw the most attention and transactions, the financial instruments are liquid, and can be transacted at any time. The result: most of the time most investors are a HOLD, which is a decisive inaction.

Illiquid Investments

All of venture investing falls into the illiquid private equity investing category. Along with preferred stock, lack of public information, and off-market deal access, these are the key features of the asset class from a financial perspective, but let’s skip over to the end game: exiting an illiquid investment.

Of course if a company is acquired, there is forced liquidity, and if a company goes public, the decision on selling or distributing stock is a whole different subject.

The key distinction between illiquid and liquid investments is that for illiquid investments, when a moment of liquidity exists, there is no HOLD, just BUY or SELL. Meaning that in a moment of liquidity (i.e., there is an active offer to buy secondary), if you are not a seller, then you are by default a buyer and vice versa. Every time the opportunity arises to sell secondary stock, an investor needs to decide if they are a buyer or seller at that price. Why? Because there may not be liquidity windows in sight for a long time, so a HOLD is effectively a BUY, and BUY is a BUY&HOLD.

An example: And investor invests $100k into a company. The company raises 2 more rounds of financing and now there is a liquidity opportunity. The original investment is now worth $2M and there is demand for that $2M chunk. The investor now must make a decision: BUY or SELL (and how much). In other words, how much exposure do they want to the current risk/reward profile of the company? Knowing what they know about the business and its current valuation, how much would they invest now? And…is there a better risk/reward profile elsewhere for the proceeds of a sale?

Note that I did not mention entry or exit valuations here because they are totally relative and context-dependent. The entry valuation could be $5M or $500M and the current valuation could be $100M or $10Bn. The methodology stays the same. You need to ask the question: how much do I want to invest at the current valuation and what is my next best alternative for any proceeds from the sale?

There is another potential pitfall here: mixing knowledge of the business with emotional attachment. Working with a founding team for several years results in both in-depth knowledge, but also creates an emotional connection. These two things must be separated in order to make a clear decision about buying and selling.

Angel Investing vs. Fund Investing

The simple example I just walked through works great for an angel investment… but what about an investment in a closed-end blind-pool venture fund?

There are two main factors that a venture GP needs to consider:

• Each investment has recourse on the others in driving the overall outcome of the fund, meaning that an original $1M investment which is now a $20M position in a $40M fund might have another 2x upside, and while venture is not about 2x’s, returning half the fund vs. the whole fund on an investment makes a big difference (even thought it is an objectively great outcome in either case).

  • There are vastly differing opinions about what to do in a situation like this, but it is often very situation-dependent. If you’re in year 9 of a 10 year fund, then getting LPs their liquidity might be the priority, but if you’re in year 4, LPs may be more enthusiastic about the future upside of the business and happy to have their capital stay in it.

  • You’re now playing a different sport: trying to figure out whether a $20M investment will 2x, rather than your day job of trying to find $1M investments that will 20-40x.

• The next best alternative question becomes much more complicated because LPs may have a wide variance. They are each different diversified asset managers and trying to please all of them is nearly impossible.

In either case, it comes down to the judgment of the GP on the future prospects of the business. As mentioned above, this is more difficult than it seems because the in-depth knowledge also comes with attachment and hopefully a healthy dose of optimism. That optimism is a requirement for any early stage investor - without it, you’d be crazy for ever getting involved in early stage venture.

Why is this so Difficult?

There are so many factors that influence exit strategy (aka harvesting). They include: tax optimizations like QSBS, pressure to deliver cash, macro market conditions, founder signals, and proprietary info, among others.

Our approach is to develop the methodologies and models to support a dynamic strategy. This means setting clear priorities and thresholds that allow you to remove the emotion from the decision in the moment, and then update those models and thresholds based on the latest market conditions and company prospects.

Another key aspect is managing personal expectations. The likelihood that you exit a position (public or private) at the local maximum price is very low, and it’s easy to get frustrated and self-criticize when you don’t. Instead, it is much more effective to remove the emotion, sit down and build out an exit plan early, try to stick to it, and focus on the wins.

One of the best known examples of this approach is Fred Wilson of USV consistently selling secondary in Coinbase. On the way up, people thought he was crazy for selling, but when all was said and done, he posted the highest total cash return of any investor on the cap table. And I believe this is case more often than not. Why? Because when things seem like they are headed in a really good direction is often when you’re already at the top. And this is extremely hard to judge because we are all human. It is far too easy to look at specific examples and let that guide decisions rather than focusing on building a method that will work across a variety of scenarios for the best total outcome over time.

I illustrated a sell-down strategy in my post here. This is not meant to be prescriptive, but if you don’t have a model in place, then when the question gets asked “do you want to sell secondary in this round?” you may not even know where to start and get caught up in the emotional loops that result in bad decisions, lost opportunities and regret. You ought to at least have a pre-existing framework to fall back on rather than just an instinctual response to an investment question, especially because the skillset here is very different from what got you into this great company in the first place.

Also a shout out to David Frankel, whose discussion on 20VC about distributing stock vs. cash post-IPO prompted me to think more deeply about this subject.

Share

Portfolio Highlights

Congratulations to Taimur, Lukas and the entire Causal team on their acquisition by Lucanet Group. We are fortunate to have been some of the earliest backers of Causal and are excited for the future of the company as part of the Lucanet CFO Solution Platform.

MotorMia announced their Seed round and official launch at the SEMA auto show this year in Las Vegas November 5-8. Isaac and his team are revolutionizing the automotive aftermarket using AI, enabling real-time modification simulation and a much more efficient process to order and install parts. We were excited to have been the first backers in the pre-seed round and even more excited about the progress and future opportunities for the company. If you’re a car enthusiast, this is a new tool that you’ll want to check out.

We are excited to lead the pre-seed round into Self. Self partners with pediatricians to provide free, routine mental health checkups and a one-stop referral platform for all the behavioral health needs of their patients. They recently launched and already reach 10% of pediatricians in the US. Since launch, they have added new distribution channels and secured partnerships with dozens of platforms—connecting tens of thousands of therapists and providers. Mental health is a defining crisis of our generation, and Self is leading the way with solutions rooted in early detection and intervention.

Clara Health assists medium and large healthcare providers in managing high volumes of phone calls from patients and their care teams. Their realistic, multilingual AI phone agents handle both inbound and outbound calls, managing cross-department inquiries and ensuring follow-ups. This enables organizations to reduce staff and improve service quality.

Epsio is building a streaming SQL engine that boosts the performance of complex and repetitive queries in customer-facing applications by orders of magnitude. Unlike traditional databases that recalculate queries each time they run, Epsio incrementally maintains results for defined complex queries — resulting in a 100x+ performance boost, for a fraction of the infrastructure costs.

Subscribe now

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We invest primarily in software business models and take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of Fund 2.

Thanks for reading Verissimo Ventures Monthly!

Thoughts on Investing and Starting Up

When a company is first getting started founders are usually focused on a key problem in their market. The goal is to build technology to create value by solving that problem for as many people as possible. In the process of validating that problem with potential customers and studying the market, founders are often confronted with how they should deploy their limited resources in the near term, and the choice can be boiled down to:

Narrow the market scope OR Simplify the product.

Put another way: would you rather go a mile deep and an inch wide or an inch deep and a mile wide?

The reality is that the best answer is somewhere in between the two and the ultimate goal is to be a mile deep and a mile wide (i.e., Microsoft, Google, etc.) The question is, for a given market and the resources available, which dimension is the right one to prioritize in the early days - because when you are just getting started, you are an inch deep and an inch wide.

Let’s look at it through the lens of three of our key focus areas: application layer software, infrastructure software, and vertical software.

Application Layer Software

These markets are nearly all very busy these days. Many competitors (incumbents and startups) are offering a variety of solutions to companies, and customer acquisition cost is going to be a big driver for key business decisions. In many cases, going simple is a good bet. Trying to build a full operating platform from scratch in 2024 is a daunting task, both from a technology development perspective, but also from a market access perspective. Being able to build and message a very specific value proposition gives small, new companies a better chance of getting customers’ attention. It also allows them to iterate more efficiently on both product and GTM strategy until they figure out what works. Once they get some customer momentum, they can work from the inside with customers to develop the rest of their product roadmap. They will have the ability to collect data, iterate on product, and leverage their existing relationships to push new products and increase ACVs. The downside/challenge here is whether a simple product can establish enough of a foothold inside customers. Often customers are using entire suites of products from incumbent software providers, so integrating a new tool and/or deprecating a free product inside an existing platform takes a truly amazing product. Some successful examples have been Slack and Zoom - their products are very similar to free offerings from Microsoft and Google, but the adoption is still massive. The products were just that much better. They have both expanded their offerings in a meaningful way to increase their footprint inside customers and their ACVs accordingly. Other great examples are Loom and Docusign. Notably, Loom is owned by Atlassian now and Slack by Salesforce while Docusign and Zoom have remained independent and shifted to a strong platform strategy.

In summary: simplify, but the products need to be killer to stand out.

Infrastructure Software

Infrastructure software is hard to build: the standards are high and the contracts are big, but the customers can be both bureaucratic and fickle. The only solution I have seen work in infrastructure software in recent years is to raise enough money to build a great product that goes deep enough to be truly differentiated while also broadly applicable to a big market. Often times you will see a CEO/CTO team wherein the CEO could have been the CTO at any other company. This helps tackle the two-pronged challenge because the CEO brings their technical chops directly to the customers while the CTO and engineering team can build the high level products that the market needs for production environments. We currently have a portfolio company doing a great job at this balancing act: the CEO is working with sales executives to narrow the initial target customer profile and execute right-sized deals while the CTO is bolstering the product with scalability and robustness so that they can move upmarket into larger customers as soon as possible. Blitzing the market with a barely-working product will lead to failure, and taking a consultative approach with large customers and trying to meet all of their needs will prevent the team from achieving a scalable product-market fit.

In summary: you need to do both! Find the subset of customers whose needs map to a larger set of customers while building a robust and scalable product.

Vertical Software

Let’s start with what vertical software actually is: software sold into a specific industry to solve industry-specific problems. Historically vertical software companies were almost all platforms - Guidewire, Veeva, Textura, Ellie Mae, Instructure, etc. The thesis was that markets were smaller, but you could run a more efficient GTM strategy and almost always land-and-expand with additional products and features built specifically for that industry. When companies were looking for software solutions, they wanted something that went end-to-end or close to it, so platforms won the day. In that environment, narrow was the approach early: start with a specific submarket, build a full-fledged product for them and then expand into the immediate adjacencies. When the alternative was pen and paper, software needed to meet all customer needs to go anywhere. As software adoption across all industries has exploded in the last 5-10 years, there is now room for more specific solutions inside verticals. Most companies are now using some sort of software to run their business - maybe it’s Salesforce, Google, Zoho, but it could also be no-code tools like Airtable, Notion, Monday.com. There are two approaches to displacing or augmenting these tools: the first is to rip out the horizontal solution that has been customized for that business with a purpose-built tool (narrow). Affinity as a CRM for VC would be a good example of this. Our portfolio company Novodia is also working on this approach in the educational content sector: displacing low/no-tech solutions for educational content creation and management. The second is to build industry-specific integrated products (simple). A good example would be our portfolio company Droxi.ai, which works with existing EHRs or Nemodata, which is focused on fleet maintenance and works with existing sensors and platforms.

In summary: It is hard to generalize anything in vertical software because adoption patterns vary so greatly by industry, but either narrowing OR simplifying could work as an early market access strategy - though it’s likely both won’t work at the same time.

Narrowing or simplifying strategies can both work depending on the situation, but be very clear on the desired approach, and then make sure that there is a path to mile wide / mile deep success from either the mile wide / inch deep or the inch wide / mile deep initial approach.

Share

Portfolio Highlights

Rep.ai (formerly known as ServiceBell), is focused on using GenAI to help sales teams generate more leads. Rep.ai engages visitors in real-time, understanding their buyer journey and converting traffic into meetings more efficiently than humans. This AI-powered solution provides continuous, context-aware sales support to boost conversions and grow the pipeline.

KiaPod Learning provides small, in-person 'pods' where online and microschool students can learn in a supportive, co-working-style space. Combining the benefits of online and in-person education, KaiPod helps students access personalized curriculums while building meaningful social connections. KaiPod also supports parents by ensuring their children are on the right learning path. With locations in Arizona, Florida, and New Hampshire, KaiPod further helps educators through KaiPod Catalyst, the nation’s largest accelerator for launching microschools, offering mentorship, resources, and a community for founders to succeed.

Fintastic offers a full-stack FP&A platform that uses probability science to turn financial data into real business value. Their no-code solution creates a digital twin of your company, enabling continuous rolling forecasts, real-time scenario simulations, goal-driven budgets, and performance benchmarking. With seamless data syncing and extensive reporting, Fintastic provides the actionable insights teams need to optimize business performance without the hassle of complex implementation or maintenance.

FormLogic automates custom machining using software and robotics to deliver rapid, precision components for space and other frontier industries. Traditionally, custom machining relied on small shops with varying service levels, capacity, and lead times. Formlogic’s 24/7 autonomous factories streamline this process, offering clients speed, transparency, and economies of scale, making high-quality custom parts more accessible than ever.

PDQ (PrettyDamnQuick) helps SMBs and fast-growing Shopify brands streamline their delivery process. Their platform automates fulfillment with features like fraud detection, easy delivery assignment, and dynamic checkout. Customers have increased order capacity fivefold, with delivery times averaging less than two days, offering Amazon Prime-style speed and flexibility for small businesses.

Subscribe now

Who we are

Verissimo Ventures is a Pre-seed and Seed Venture Fund based in Israel and the US. We invest primarily in software business models and take a fundamentals-driven approach to early-stage investing. We work closely with founders to help them build the strongest, most fundamentally sound businesses with potential for explosive growth and a meaningful impact on the market.

We were founded in 2020 and are currently investing out of Fund 2.

Thanks for reading Verissimo Ventures Monthly!