Subscribe now

Zillow, Redfin, and realtor.com make 80-90% of their revenue from buyer-agent commissions, so they’ll never be the ones to shrink that pool. Traditional brokerages can’t either: their agents are 1099 contractors who’d walk the moment commissions dropped.

A traditional agent closes 5-10 deals a year. Homa’s in-house agents, freed up by AI doing the pricing, document review, and buyer Q&A, close 200-250. That math is what lets Homa hand buyers up to 2% of the purchase price back.

We also get into why this might work even better in a cutthroat market like the Bay Area than in the Sun Belt, an unreleased survey showing 63% of buyers feel guilty asking their agent for help, and what happens to the other 1.9 million agents if this catches on.

Full conversation below:

Marcus was Goldman’s business, which meant funded by Goldman’s balance sheet, staffed by Goldman employees, built under Goldman’s post-financial-crisis risk appetite. It worked, but Omer says the two things missing were exactly what OnePay was built around: distribution and incentives.

Every Marcus customer cost money to acquire. At OnePay, a Walmart employee can open a banking account in two clicks and 20 seconds while setting up benefits. That’s zero CAC at the exact moment of intent.

Four years later: 7 million monthly actives, a $4B valuation, and a customer base that actually looks like America’s middle 70-80%, not the narrow slice most neobanks start with.

We also get into how AI has reshaped how OnePay builds and where the roadmap goes next: family banking, mortgage, and stablecoin-powered remittances.

Full conversation below:

Subscribe now

Dalton’s sharpest point wasn’t about the tech, but imagination. He calls it “the aperture,” or the narrowing range of ideas founders actually conceive of.

When YC started, there was no content machine telling founders what a good startup idea looked like, so people pitched wildly. Dalton even remembers a founder proposing a truck that baked muffins en route to delivery. Now, with startup advice everywhere and AI slop flooding every feed, founders converge on the same ideas.

OpenAI wasn’t a business when it started, it was a research nonprofit nobody could underwrite. Space companies were nearly unfundable a decade ago. Internet banks were considered a one-way ticket to a bank-partner lawsuit until they weren’t.

Dalton says the biggest outcomes come from founders willing to sit outside the aperture.

Eighteen months ago, Seema hosted a dinner for CMOs and was stunned by how few touched AI beyond quick research questions. Marketing is the most obvious use case there is, and they still didn’t trust it to run anything.

That’s what changed, but unevenly.

The internal work is getting automated first:

-segmentation

-attribution

-pulling insight from messy data

Why? Because you can check it. The external side (live campaigns, dynamic creative) lags because trusting an AI-generated ad to go out unsupervised is a far higher bar.

Nearly every marketing tool today lives in the analytics layer. The GEO tools tell you where you rank and why. The missing piece isn’t intelligence, but the execution you can trust. For the first time, the autonomous team reads like a roadmap instead of a fantasy.

Tune in to the full interview below. And thanks to Plaid sponsoring!

But the number that stuck with me wasn’t about growth, it was that one in six AI signups are coming from the same bad actors.

Subscribe now

In the old SaaS world, free-trial abuse and multi-account fraud were annoying but mostly harmless. The incremental cost of one more SaaS license was close to zero. So who cared if someone strung together a few fake trials?

But tokens have real economic value. When a bad actor chains together a bunch of free trials to siphon off inference, that’s not a rounding error, it’s a direct hit to the unit economics of the business. And because these companies are scaling so fast, the abuse scales with them. If you can’t get ahead of it, it compounds.

Stripe is partnering directly with OpenAI, Anthropic, and Cursor to fight a category of fraud that essentially didn’t exist before. The attack surface is new, the economics are new, and nobody has a decade of playbooks to fall back on.

Tune in to the full conversation below!

Today, Gusto is using Privy‘s infrastructure to pay contractors in Nigeria who’ve never touched crypto. Ramp is building business accounts that run entirely on stablecoins. DoorDash is using the same rails to pay out workers globally.

Subscribe now

All of these are things 🏕️ Henri Stern, Privy’s founder, says would have taken them eight years to build as a standalone company. They did it in one thanks to their acquisition by Stripe.

I caught up with Henry at Stripe Sessions, where Stripe just announced digital asset accounts, or an API that bundles storage, transactions, yield, and trading into one interface.

Tune in to the full interview below!

That’s how Gusto shipped Co-Founder, their new AI-native product that will generate hundreds of millions in revenue.

Edward Kim, Gusto’s CTO, thinks the build story is more interesting than the product itself.

For context: Gusto crossed $1B in revenue last year serving 500,000 small businesses. Their growth playbook has been pretty familiar: build great payroll, expand into adjacent products, acquire where the build timeline is too long.

Guideline for 401k.

Mosey for compliance.

Subscribe now

But Eddie was one of those five builders on Co-Founder, submitting pull requests again for the first time in years. And the comparison he keeps coming back to is M&A. He’s been in acquisition conversations that couldn’t even close in nine weeks.

“In the age of AI, there are things that are probably not worth buying anymore.”

Co-Founder itself is a pre-configured AI agent connected to all of Gusto’s existing payroll, HR, and compliance infrastructure. One that small businesses can talk to over text, Slack, or the Gusto interface, no login required. It ships knowing your business already: your employees, your pay schedules, your open positions. Eddie’s framing is what he calls the “work before the work,” the spreadsheet gymnastics a massage spa owner does every week to calculate therapist pay, commissions, and tips before they ever open Gusto to run payroll. Co-Founder closes that gap end-to-end.

The deeper implication Eddie is sitting with: if you can ship something acquisition-scale in nine weeks with five people, the whole framework for how you prioritize, staff, and invest changes. Build vs. buy isn’t a procurement question anymore. It’s a strategy question.

Tune in to the full interview below!

That’s an audacious bet, so I sat down with Dennis Yang, Head of Generative AI at Chime, to understand how they’re actually pulling it off.

Subscribe now

The short answer: AI running three layers deep.

-inside the org

-inside the product

-inside the infrastructure for what comes next

On the operational side, engineers using AI are shipping nearly twice as fast, with ~80% more code changes and ~60% faster reviews. Dennis is pushing every Chimer to find something AI can’t do today and check back in ninety days.

That same compounding logic applies to user research. A team of 15 researchers that used to run 20-30 interviews a week can now run 2,000. AI conducting real conversations, aggregating insights, and delivering them to PMs in hours instead of weeks.

On the member-facing side, Chime’s AI assistant “Jade” now resolves over 70% of support interactions without a human. Up from under 20% on the old IVR. Dispute resolution time dropped from 42 minutes to under 20. Fraud loss rates fell 29% over three years.

Dennis is also very interested in agentic commerce, where Jade could eventually act on behalf of members, negotiating bills, pausing subscriptions, surfacing benefits they didn’t know they had.

Watch the full interview below!

Last week I sat down with Zach Abrams, co-founder of Bridge, which was acquire by Stripe for $1.1 billion in October 2024. At the time, it was the largest crypto acquisition ever, and plenty of people thought they overpaid.

Twelve months later, Bridge’s closest competitor, BVNK, just transacted at $1.8B. Bridge itself is reportedly growing 4x year-over-year, per Stripe’s own annual letter. The cross-border stablecoin rails Stripe bought are quietly becoming one of their highest-velocity businesses.

I sat down with Zach Abrams to talk about all of it. The valuation gap, the growth engine, and whether he thinks he sold too early.

The valuation gap is real

BVNK at $1.8B vs. Bridge at $1.1B isn’t just a headline. It’s a signal about how fast the market repriced stablecoin infrastructure once the regulatory environment started to clear. Bridge was acquired before the GENIUS Act momentum, before the stablecoin market cap crossed $170B+, before it became obvious that cross-border payments were the killer app.

Zach’s honest about it: the $1.1B wasn’t a bad deal, it was the right deal for the information available at the time. But the market moved faster than almost anyone predicted.

The 4x number deserves more attention than it’s getting.

Most acquisitions kill product velocity. Corporate gravity sets in. Procurement cycles, headcount freezes, brand consolidation. Bridge is doing the opposite. Zach credits Stripe’s unusual operating model: acquired teams retain meaningful autonomy, and distribution is additive rather than cannibalistic.

4x YoY growth inside a $1T TPV company isn’t a rounding error. It’s a validation of the thesis that stablecoin rails can run through, not around, existing financial infrastructure.

In January 2022, UBS agreed to acquire Wealthfront for $1.4 billion. By September, the deal was dead. Most CEOs would have spent a year licking their wounds. David told the team something different:

“We will be profitable shortly. We are growing extremely quickly. And I think we’ll look back and say we were incredibly lucky this acquisition fell through.”

Three years later, Wealthfront IPO’d. The $1.1B market cap is roughly what UBS was going to pay. But now Wealthfront owns its own future.

Schwab-level margins at 1/30th the AUM

Here’s the number that stopped me: Wealthfront runs roughly 47% EBITDA margins on ~$94B in AUM. Schwab runs similar margins on trillions. The reason isn’t magic. It’s incremental gross margins above 90% on both the cash and investing products.

David spent 10 years as CTO before becoming CEO. His job during that decade was to find every variable cost in the business and turn it into a fixed one. They optimized print statement margins. They brought services in-house. They took the marginal cost of managing another dollar on the platform down to nearly zero.

The model: ~38bps blended, 60bps on cash. Half of new clients come through referrals. CAC stays low. Margins stay high. Scale does the rest.

Wealthfront is a case study in what happens when you hold the line on technology investment before you scale. Most consumer fintechs spend on growth first and figure out unit economics later. Wealthfront did it the other way around.

Mortgages: the industry’s margin is their opportunity

Wealthfront launched home loans in Colorado, Texas, and California in late 2025. This is their biggest product launch since the cash account, and it’s a deliberate swing at one of the most operationally brutal products in financial services.

The logic David laid out is clean:

• Non-bank mortgage originators spend ~$10,000 per loan in origination costs. Mostly humans, not software

• They have purely transactional relationships with borrowers, so they’re spending that $10K cold

• Wealthfront has 1.4M existing clients, long-term data on their finances, and established trust

• They’re already seeing $2.4B in wires to escrow and title annually, which is implicit demand hiding in plain sight

Early results: over 50bps better rate than the national average. The playbook is to prove it works with existing clients, let word-of-mouth pull new ones in, then expand.

I asked if they would have tackled mortgages without AI. The honest answer: probably not. The pandemic forced the mortgage industry to digitize fast. Counties went online, APIs proliferated, OCR got good enough. That infrastructure shift is what made the opportunity real. AI accelerated the timeline.

The CEO who Googles his own data

David asked an engineer a granular question about fraud decline rates. The engineer’s response: “You could have answered that yourself,” and sent him the chat log where he’d just asked their internal AI tool, got the query written, pulled the data, and had the answer in two minutes.

The broader point David made is important: AI doesn’t just change what CEOs can do. It changes what everyone can do. His job now is less about information gatekeeping and more about setting context so that a more distributed, faster-moving team can make better decisions without his sign-off on everything.

Internally, every engineer has AI coding tools. They’ve built a proprietary agent that can write pull requests for specific features, runs automated code review on every PR, and has deep context on their codebase. The small team that maintains it is one of the highest-leverage groups in the company.

Not anti-advisor. Anti-bad-fit.

The robo-advisor narrative of “AI will kill financial advisors” has always been too blunt. David’s framing is more precise: financial advisors serve older clients who grew up valuing human relationships for money decisions. That cohort is aging out. The generation behind them grew up buying groceries on their phones at midnight. They want their finances to work the same way.

The structural data supports this. The industry isn’t training many new advisors. Those who exist skew toward serving high-net-worth, older clients. The advice gap for younger, mass-affluent investors is real, and it’s Wealthfront’s market.

When I pushed on where human advisors still win, David was direct: if you want one, you should get one. Just watch the fees, prefer fixed-fee or fee-only structures, and understand that the emotional management function, keeping you from panic-selling in a downturn, is genuinely valuable. Wealthfront tries to do that through product design and content instead.

Wealthfront is not a robo-advisor anymore. It’s a technology-first financial institution with Schwab-comparable margins, a product roadmap that’s getting longer not shorter, and a credible wedge into the largest consumer asset class in America.

The mortgage bet is the right move at the right time. If they can crack the automation problem in origination the way they cracked it in investing and cash, the unit economics could be extraordinary. And they’ve got 1.4M clients already holding the door open.

That’s the infrastructure Figure and Valon are replacing.

I sat down with Michael B. Tannenbaum (Figure) and Andrew Wang (Valon) to dig into their new partnership.

The early numbers are hard to ignore:

• 40% improvement in delinquency outcomes

• 23.3x increase in originations Q1–Q4 2025

• 2x HELOCs funded per loan officer

• 2.9x improvement in app-to-fund conversion

The idea is that when you pair blockchain-native origination with purpose-built servicing, the results compound.

• better borrower communication reduces delinquencies.

• better recapture rates unlock negative CAC.

• better infrastructure means smaller originators can compete with the Rockets of the world.

Full interview below!

I recently sat down with Mike Cagney, founder of Figure, after they made history with the first public equity natively listed, traded, and settled on blockchain.

• If you hold margin stock at Robinhood or Schwab, they can lend it to short sellers without your consent, and you get nothing

• The DTCC processes $74 trillion in equities. Figure’s thesis is that entire vertical stack (registry, exchange, prime broker) gets replaced by a wallet

• DATs like MicroStrategy are the likely first movers onto OPEN. The arbitrage mechanic Cagney lays out is genuinely compelling

We’re still early. But the profit pools at risk here are huge.

Full interview below👇

But was this a sale, or a calculated move to access a bank’s balance sheet and go to war at enterprise scale?

Today I’m sitting down with Brex CFO Erica Dorfman to find out. We’re going to unpack the bull case, the bear case, and the biggest misconception floating around right now.

Including who Brex’s real competition actually is.

Hint: it’s not Ramp.

If you want to understand how the competitive landscape in banking, finance, and tech is shifting, this is the conversation to watch.

The catalyst? Mercury Personal, the most ambitious product launch Mercury has ever made. And this week, I sat down with Immad Akhund, Founder & CEO, at their San Francisco offices to dig into why they built it, who it’s really for, and whether it can actually win.

The narrative here is interesting.

While Chime and Robinhood chase the mass market, Mercury is going the opposite direction. Targeting what they call the “Super Prime” customer: founders, operators, and builders who carry serious balances and want a bank account that works like software.

• 200K Business customers

• $80K Avg. customer balance

• 3.25% APY on deposits

I’ve been a happy customer, but I also don’t think Mercury Personal is for everyone, and Immad agrees. That honesty is part of what makes this conversation worth watching.

We cover:

• Who the Super Prime customer actually is, and who Mercury is not targeting

• Why Mercury launched a consumer product after years of staying laser-focused on business banking

• How they plan to win with a $240/year paid model in a world of free neobanks

• The roadmap: what features are coming next

• Why Mercury is pursuing a bank charter, and what that means for their consumer ambitions

Mercury Personal is now open to everyone after an alpha with ~4,000 users. If you’ve ever felt like your personal banking is years behind your business tooling, this one’s for you.

WATCH THE FULL INTERVIEW

I Finally Broke Up With Bank of America.

Before you read the numbers, watch my new interview out this morning.

I drove down to Robinhood’s HQ in Menlo Park to sit down with Abhishek Fatehpuria, their VP of Product. He joined as an engineering intern in 2016 when the company was 40-50 people. Today he runs product for their largest division: Brokerage.

We covered a lot:

→ How Robinhood’s free stock referral program went from cash incentives to random stocks, and why people cared more about getting a share of Apple than the dollar value

→ Why Robinhood starts product development with “swipies” instead of PRDs (”If you can’t capture them in one screen, it doesn’t matter what you build after that”)

→ How building for active traders, who demand pixel-perfect design and millisecond latency, created a discipline that makes every other product better

→ The shift from winning active traders to winning share of wallet, and what it means that the average account transfer is now ~$80K

→ Where Cortex, tokenization, and the super app strategy are headed

Robinhood crossed $100B market cap last year. But the most interesting story isn’t the stock price. It’s how their product philosophy is positioning them to compete with JPMorgan and Schwab for mass affluent customers.

Full conversation below. 👇

The Links

1️⃣ Claude’s latest demo of their new AI Finance Agent builds a DCF live on-stage in < 2 minutes.

2️⃣ The biggest neobank you’ve never heard of now offers free tax filing.

3️⃣ Highlights from Mercury’s annual letter

The rise of Stablecoins is one of the most important trends in finance right now. James and I covered two pretty under appreciated aspects of Stablecoins.

The first is the network effects that underpin both digital and physical currencies.

The second is why Stablecoins are important, not just as a digital utility layer for the movement of money on the internet, but also as defense tech, something that serves the national interest.

Tune in to the full interview below!

In case you missed it, I just dropped my exclusive interview w/ the CEO of the latest Stablecoin unicorn. Mesh and Bam Azizi just raised a $75m Series C at $1b valuation.

Mesh powers products at companies like PayPal , Revolut and Coinbase.

Full interview below!

The Links

1️⃣I had the first interview with Brex after it announced it was selling to Capital One

2️⃣How to get to $1B in revenue faster than some of the biggest companies in the world

3️⃣The value of free products vs the value of AUM

I just released a new video on YouTube, and all about what it takes to hyper-scale.

Deel hit the $1 billion revenue mark faster than the most successful companies of the last generation: faster than HubSpot…Palantir…Salesforce. Even faster than Stripe.

This is my interview, Alex Bouaziz, Deel’s founder & CEO.

I first met Alex 6 years ago at a table in South Park in downtown San Francisco, few months after his co-founder Shuo Wang had presented Deel to the Cambrian community at a fintech showcase AND a few months before my team at A16Z led the Series A.

Now, 6 years later, he and I are catching up to dive into the 11-point playbook the Deel team used to hyper scale. We go beyond platitudes to cover specific examples of how and why this playbook was developed at Deel with Alex sharing several anecdotes he’s never discussed before publicly.

Tune in below!

Links

1️⃣NYSE to offer tokenized stocks.

2️⃣Robinhood is rebundling everything.

3️⃣Visa running laps around Mastercard in Stablecoin card volumes.

Speaking of the new year, lots of big predictions for fintech in the next 12 months from both myself and Simon Taylor.

Simon’s are in my most anticipated video of the year, which just went live this morning. Check it out!

Top Links

1️⃣I created a prediction market. Cast your vote here 👇

2️⃣JPM just took over the Apple Card. Here are the important details.

3️⃣Banks will lose more mass affluent customers to fintechs in 2026 than ever before.

About to sit down with Alex Bouaziz, CEO of Deel.

Deel hit the $1B revenue mark faster than the most successful companies of the last generation: faster than HubSpot, faster than Palantir, faster than Salesforce. Even faster than Stripe.

Alex is going to share the 11-point playbook they used to execute that run. But this isn’t just a list of platitudes. It’s a set of hard-won, often counter-intuitive lessons.

Subscribe so you don’t miss it!

Links

1️⃣ Tether is looking to raise $20 billion at a $500 billion valuation and tokenize their shares

2️⃣ Interview with Wealthfront CEO on IPO day from the Nasdaq

3️⃣ Founders, if you are going to wind down, do it before EOY!

Subscribe now