I am not a security researcher. I’m a marketer with a lot of technical ability. I read the same stuff you can read, I talk to people who understand this stuff far better than I do, and then I try to figure out what it means for how I operate as someone who, as part of my job, oversee’s web developments. So this is not a technical breakdown. There are smarter people writing those, and they’re worth your time if that’s what you need.

I also want to be clear, I’m not crapping on WordPress - I'm being critical of the nature of dev’s who use the platform that have enabled a mentality of “it’s just secure, it’s fine” when it really isn’t. It needs your attention, like everything else in our lives.

What happened

On July 17th, WordPress pushed out three emergency releases at once and then did something unusual and honestly a little startling: it forced updates onto sites all over the world, whether the owners knew about it or not.

The bug, menacingly name wp2shell, let an attacker take over a site without logging in. No password. No stolen credentials. No tricking an employee into clicking anything. Just a request sent to the site, from anywhere, by anyone who knew how.

Within about four days there was working attack circulating publicly, the U.S. gov’s cyber agency had added it to its “actively being exploited” list, and security firms were reporting real break-ins. Estimates of exposed sites ran into the tens of millions.

That’s the gist of it. Now here’s the part I got wrong in my article “Your Website Agency Is Holding You Hostage — And WordPress Is Letting Them.”

My March argument was too comfortable

When I wrote about WordPress in the spring, I hung a lot of the security case on plugins. About half of WordPress sites are running at least one plugin with a known exploit in it. It’s a real stat and it’s genuinely alarming to me, and should be to you too.

It was also the easy argument, and I should have seen that at the time. I too, make mistakes.

Because if the danger lives in plugins, the story has a clean ending. You’ve got a discipline problem. Audit what’s installed, rip out the abandoned junk, patch on a schedule, hire someone who cares. Annoying, but solvable. Everybody gets to keep their site and feel a little better about it.

wp2shell didn’t leave that door open. The flaw was in WordPress itself. A clean install, brand new, nothing added, nothing exotic — still wide open. There was no sloppy plugin to blame and nothing to clean up, because there was nothing there in the first place.

I was right that it was going to hurt. I was wrong about where it would come from. Little did I know it was going to come from the mothership itself.

The forced update tells you more than the bug

If you take one thing from this, take this:

WordPress didn’t just release a patch and send an email. It reached into millions of websites it doesn’t own and updated them without asking, because it had every reason to believe a huge share of owners would never get around to doing it. Is that ethical?

Sit with that for a second. The people who maintain the platform looked at their own user base and concluded that trusting those users to protect themselves was a losing bet. Not a wild guess, a judgment based on years of watching what actually happens with time and human behavior.

That’s a confession about a sprawling, aging, loosely-supervised installed base where the person nominally responsible for your site’s security is often an agency that stopped returning your calls two years ago.

The forced update mostly worked, by the way. Which is a genuine credit to them. But it only reaches sites where that machinery is turned on and functioning. Every security firm covering this is saying the same thing, so in case you did not get the memo: go check your version number yourself.

“The people who maintain the platform concluded that trusting their users to protect themselves was a losing bet. That’s a quiet confession about the whole arrangement.”

Now imma let you finish, but let me argue with myself for a minute

I use and promote the alternative to WordPress in most cases. Take what I say with as much salt as you need. I’d rather put my objections on the table than let you find them later and wonder what else I skipped.

WordPress handled this decently. Objectively, and better than most larger companies would have. They quietly warned the major providers before going public, so protective rules were live within hours of the announcement. They shipped worldwide in under a week. That’s a solid response for a bad day, and it deserves praise.

The static stack isn’t bulletproof. It just fails in different places, and I’d be doing exactly what I accused those agencies of doing if I let you believe otherwise.

Here’s my honest version. A pre-rendered site doesn’t have a program running on the server waiting for a request, which is why this particular category of attack doesn’t reach it, there’s nothing on the other end to confuse. But the site still gets built, and building it means pulling in hundreds of small packages written by people I have never met and will never read a line of code from. If one of those gets poisoned somewhere upstream, it rides straight into my build and I’m the one who signs the delivery. That’s a smaller door than the one WordPress left open last week, but it’s a door. That’s the nature of development though. There will always be a risk in some capacity.

And my whole arsenal is in a handful of accounts. Get into those and there is no second lock waiting behind them (don’t even try, pal). Fewer moving parts is genuinely an advantage right up until it isn’t.

So the claim I’d actually defend is narrower than what I wrote in March. I also think it’s sturdier for revision:

With a modern static setup, your exposure is mostly a function of how the thing was built. With WordPress, it’s mostly a function of how well somebody has been keeping up with it. One of those survives your vendor having a rough quarter. The other one doesn’t.

What you should do now!

Ask these four questions to your site admin. Send them in writing, so you have the answers in writing. Stress urgency!

Which version is every one of our sites running, right now?

Not “auto-updates are on.” The actual number, site by site, including that campaign microsite from 2023 that nobody has opened since. The safe releases are 6.8.6, 6.9.5, and 7.0.2 or newer.

Has anyone checked whether we were already hit?

This is the one everybody skips, and it’s the one that matters most. Updating locks the door. It does not remove anyone who already walked in. Attacks started within days of the announcement, so “we patched it” is not an answer to “were we breached.”

Who is watching for the next one?

There will be a next one. Is it the agency? The hosting company? A specific human being with a name and a phone number? If nobody can answer that cleanly, your plan is basically hope, and hope only worked this time because WordPress stepped in and did the job for you.

What’s actually sitting in there?

Form submissions, contact records, whatever’s accumulated in that database over the years. And there’s a nastier angle worth knowing about: hijacked marketing sites are frequently used to show visitors a fake error message that talks them into pasting a command into their own computer. Your website becomes the thing attacking your customers. That’s a brand problem long before it’s an IT problem.

The part that has meaningful long-term impact

I’m not telling you to tear out WordPress (right now, at least). Migrations are insanely expensive and disruptive and I’d be a massive hypocrite to pretend otherwise given what I do for a living.

But this should change what you think you’re paying for. Open your eyes.

If there’s a monthly retainer, some slice of it is supposedly maintenance and security. Last week was a free, unannounced, real-world test of that line item to scare you into re-imaging your sites security, and ability to react to something catastrophic. You now know how your provider behaves in an actual emergency. Did they call you, or did you call them? Could they produce your version numbers in an hour, or did it take a week and three follow-ups? Have they looked for damage, or did they just say it’s handled? Details matter.

You couldn’t have bought that information. You got it anyway.

Most companies are going to file this under “taken care of” and get back to work. Probably nothing bad happens. Good.

The ones who treat it as a diagnostic, who actually read what it revealed about their own setup and the people running it, are going to be standing somewhere very different the next time this comes around.

And it will come around. I promise. It always does.

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world.

Cheers!

Please Subscribe & Give Me a Like or Two on YouTube

Share

Subscribe now

Leave a comment

On July 7, the most capable model Anthropic makes stops being included in your Claude subscription and starts charging you by the token. Most marketers will read that as pricing news and scroll past. The scroll is the mistake — the date is a deadline on a skill you were going to need the moment your AI work turned into a real cost.

Let me give you the facts, then the one reframe that makes them matter, then the audit you should run before Tuesday.

What’s changing with Claude Fable 5 on July 7?

Through July 7, Fable 5 is included in Pro, Max, Team, and select Enterprise plans for up to 50% of your weekly usage limits. After that, it moves to usage credits, you keep the model, you just start paying per token to run it.

Fable 5 came back on July 1 across the Claude Platform, Claude.ai, Claude Code, and Claude Cowork after a short export-control pause. The catch is the seat you sit in. Standard Enterprise seats get no included window at all — it’s usage credits from day one, and if credits aren’t enabled, no access. So the flat-rate era for Anthropic’s best model has a hard end date, and for some accounts it’s already over.

How much more expensive is Fable 5 than the other Claude models?

Per million tokens, Fable runs $10 in and $50 out — exactly double Opus 4.8, five times Sonnet 5’s current rate, and ten times Haiku 4.5 for the same job.

Here’s the full picture:

• Fable 5 — $10 / $50 per million tokens (input / output)

• Opus 4.8 — $5 / $25 → Fable is 2× Opus

• Sonnet 5 — $2 / $10 right now (introductory through August 31; $3 / $15 after) → Fable is 5× Sonnet today, ~3.3× once the intro rate lapses

• Haiku 4.5 — $1 / $5 → Fable is 10× Haiku

Same task. Same tokens. Up to a 10× cost spread depending on which model you point it at. That number is the only one you need to internalize.

Should I stop using Fable 5?

Wrong question. The move to usage credits looks like an access problem - “do I still get the good model?” - but it’s a budget problem: which of your tasks actually deserve the good model. Capability just stopped being a default you pick out of a dropdown and became a line item you approve.

That’s the reframe, and it’s the whole piece. Not “can I use Fable” — “should I, for this one.” Every task you run has a difficulty. Every model has a price. The job is matching them, and until now you never had to, because the best model was free inside your plan and there was no penalty for pointing everything at the top of the menu.

Isn’t this the same discipline as media buying?

Yes, you already own this instinct, you just know it from ad spend. You don’t put Super Bowl money behind a post announcing your new office hours; you match the spend to the stakes of the placement.

Tokens are that same math now. Fable is premium inventory. Right up until July 7, that premium inventory is sitting inside your subscription for free — which is exactly why people are about to burn it on the equivalent of footer copy. The meter just makes visible a decision you should have been making all along.

How do I decide which AI model to use for each task?

Run a routing audit: list your recurring AI tasks and tag each one with the cheapest model that still clears your quality bar. That single table replaces the Fable decision with a routing map for everything.

The tiers land cleanly once you sort by what the work actually demands:

• One-shot thinking, where being right once pays for itself — strategy synthesis, messaging architecture, untangling a positioning mess. This is where Fable or Opus earns its rate.

• Drafting, editing, most production content, most agentic runs — Sonnet, at roughly a fifth of Fable’s cost today and about a third once the intro pricing ends.

• Tagging, extraction, classification, routing, bulk transforms — Haiku, at a tenth.

What falls out of that table isn’t a Fable decision. It’s a cost model for your entire workflow. And the July 7 date is just the forcing function that makes you finally build it.

Why does this still matter after July 7?

Because the date, the model name, and the price sheet are all disposable — Fable becomes something-6 by fall, the numbers move — and the discipline underneath doesn’t rot. Matching task difficulty to the cost of reasoning is effort allocation: the operational craft that comes after prompt engineering, and the one that quietly protects your margin once tokens are your cost of goods.

This is the bigger pattern I broke down in my AI bait-and-switch video: the tools onboard you for free, then quietly hand you the token bill. Fable’s July 7 meter is just the cleanest example of it yet.

Learn it on Fable this week and you already know it for every model that ships after. That’s the point of running the audit now instead of waiting for the next pricing change to force it.

One honest caveat, because I won’t sell you false precision: Anthropic hasn’t published what “50% of your weekly limits” actually equals in tokens or dollars. So anyone handing you an exact “you get X Fable runs before the meter” number is making it up. Watch your own usage dashboard, not their math.

Three moves before Tuesday

Capability is now something you budget, not something you default to. Act on it before the window closes:

1. List your recurring AI tasks. Actually write them down — the stuff you run every week.

2. Tag each with the cheapest model that clears your bar, and route accordingly.

3. Spend this week’s free Fable window on the two or three hardest problems on that list, the durable, high-leverage work, not on anything Haiku could’ve done for a tenth of the price.

Do that, and July 8 isn’t a downgrade. It’s the day your token spend finally maps to the value of the work.

I'm Nicholas Putz. Systems & Signal is where I break down moves like this one into operating systems you can actually run — for marketers who'd rather understand what's happening than panic about it. If this was useful, the video above is the fast version, and a forward to one operator who needs it is the best thing you can do for the channel.

Please Subscribe & Give Me a Like or Two on YouTube

Share

Subscribe now

Leave a comment

TL;DR: For two years, AI cost was invisible — flat fee, write an email, move on. That era is ending, not because anyone's punishing you, but because what you're asking AI to do has changed. As you move from chatting to building, token cost becomes your cost of goods. Here's what a token is, the four shifts that made this matter, and five moves to build a token instinct before the bill scales.

For the last two years, we’ve used AI like it was free. You pay your $20 a month, you write some emails, you brainstorm subject lines, you move on. The meter was never running, because there was no meter.

That’s coming to an end. Not because anyone’s punishing you — because what you’re asking your AI to do is evolving. And the one thing nobody explained to marketers is the one thing that’s about to decide whether your AI workflows make money or quietly bleed it.

It’s called token utilization. And it’s the most important boring concept in marketing right now that almost nobody is talking about.

Why this never came up

When you use ChatGPT or Claude to write a single email, you’re doing the AI equivalent of sending one text message. Tiny. Instant. Effectively free. The flat monthly fee covers it a thousand times over, and the AI companies have been subsidizing heavy usage to win market share.

So an entire generation of marketers learned AI in an environment where cost was invisible. You never saw a bill that scaled with usage. You never had to ask “what did that cost me?” because the answer was always “basically nothing.”

That created what’s about to become a very expensive blind spot.

What a token actually is

A token is a chunk of text — roughly 3 to 4 characters, or about ¾ of a word. “Marketing” might be one token. “Unsubscribe” might be two or three.

Rule of thumb: 750 words ≈ 1,000 tokens.

Two things cost tokens, and this is the part people miss:

• Input — everything you send in: your prompt, the document you pasted, the brand guidelines, the chat history.

• Output — everything the AI sends back.

You pay for both. So a “short prompt” that includes a 40-page PDF isn’t short. You just paid for 40 pages of input every time you hit enter.

That’s the whole concept. Now here’s why it suddenly matters.

The four shifts that changed everything

A year or two ago, AI was a chatbot. You typed, it answered, done. Low token use by design. Today AI is becoming infrastructure — something you build workflows on top of. Four shifts blew token usage wide open:

1. Agents. AI doesn’t just answer anymore — it runs. It takes a goal, breaks it into steps, calls itself ten, twenty, fifty times, uses tools, checks its own work. One “task” that used to be one message is now dozens of calls under the hood. Every call burns tokens.

2. Context windows exploded. Models went from holding a few pages to holding entire books — hundreds of thousands of tokens. That’s powerful: dump your whole content library in, ask anything. But “dump everything in” means you’re paying for everything in, every single call.

3. Reasoning models. The newer “thinking” models generate enormous amounts of internal text working through a problem before they answer you. You don’t see most of it — but you pay for it. A single hard question can quietly cost 10x a simple one.

4. You stopped chatting and started building. The moment you move from typing in a chat box to wiring AI into a workflow — auto-generating 500 product descriptions, summarizing every support ticket, running a content engine — you’ve left the flat-fee world and entered the pay-per-token world. That’s the API. And the API doesn’t subsidize you.

Put it together: the tasks got bigger, the models got hungrier, and the pricing got metered. The cost that was always invisible is now attached to volume — and marketers are the ones building the high-volume stuff.

Where it gets real for marketers

The second you productize an AI workflow, token cost becomes your cost of goods. It’s no longer a $20 subscription — it’s a per-unit expense baked into every output you ship.

So do the math marketers never do:

• A workflow costs 10 cents a run. Fine.

• Now run it 100,000 times a month. That’s $10,000. Suddenly it’s a real budget meeting.

A workflow that costs a dime is harmless — until you multiply it by volume. Then it’s a line item.

And the silent killer is context bloat. Say you built a content generator that re-sends your 5,000-word brand guide on every single call. At scale, you’re paying to read the same document a million times. Trim that context and you can cut cost 60–70% — same output.

Same story for anyone selling AI-powered services. If you charge a client a flat retainer for an AI tool, your margin is the gap between what they pay and what the tokens cost. You don’t control that margin unless you understand tokens. This is the difference between an AI offering that scales profitably and one that eats you alive at volume.

The five moves that separate marketers who scale AI profitably from the ones who get wrecked at volume — below, for subscribers.

You don’t need to be an engineer. You need a token instinct.

1. Learn to eyeball cost. Words ÷ 0.75 = tokens. Before you build anything at scale, estimate input + output per run and multiply by volume. If that number scares you, fix it now — not after you launch.

2. Right-size the model. You don’t need the most powerful model to reformat a list. Cheaper, faster models can be 10–30x less expensive and totally fine for routine work. Match the model to the job, and expect to adjust and maintain that mix over time.

3. Stop dumping everything in. Start thinking methodically. Send the AI only what it needs for this task, not your entire knowledge base every time. Context discipline is the single biggest lever on cost — and the behavior that’ll bite people later if they don’t correct it now.

4. Use the cost features. Prompt caching (for repeated context like brand guidelines) and batch processing (when you don’t need instant answers) can slash bills dramatically. If you’re building, learn these before you scale.

5. Think in unit economics, not subscriptions. Stop asking “what’s my monthly AI bill?” Start asking “what does one output cost me, and what happens at 10x volume?” That single reframe is what separates marketers who scale AI profitably from the ones who get a nasty surprise.

The bottom line

The flat-fee, all-you-can-eat era of AI was a phase — a land grab to get you hooked. The next phase is metered, and it rewards the people who understand what they’re spending.

You don’t have to be afraid of it. You just can’t be blind to it anymore. Build the instinct now, while the stakes are still low, and you’ll have a real edge when everyone else gets their first scary invoice.

I'm Nicholas Putz. Systems & Signal is where I break down moves like this one into operating systems you can actually run — for marketers who'd rather understand what's happening than panic about it. If this was useful, the video above is the fast version, and a forward to one operator who needs it is the best thing you can do for the channel.

Subscribe now

Leave a comment

Share

Round two is quieter, and most marketers are going to walk right past it, because it doesn’t announce itself as a skill. It shows up as a setting.

Claude’s latest model, Opus 4.8, put two controls directly in the interface: an Effort slider that runs from Low to Max, and an Adaptive thinking toggle. The slider defaults to High. The official framing is innocent enough… higher effort means more thorough answers, but it’s slower and burns through your usage faster.

Read that again. Burns through your usage faster. That’s not a quality dial. That’s a meter. And the moment a tool starts metering itself, you’ve stopped having a creative problem and started having a budgeting problem.

I broke down the practical version of this in five minutes. Watch it, then come back here for the system:

The reframe: compute is now a managed input

For two years, AI sat in the marketing stack as if it were free. One flat experience, ask it anything, get an answer. The cost was invisible, so nobody managed it.

That era is ending. When effort is exposed as a control with a real cost — your time, your limits, your team’s throughput — it joins the list of inputs you already manage as an operator: ad budget, headcount, agency hours. You don’t put your entire media budget behind a single retargeting banner. You allocate against expected value. Effort is no different. The skill isn’t writing a better prompt anymore. It’s knowing how much reasoning a task is worth.

The marketers who internalize that will quietly pull ahead. Not because they’re more creative, because they stop overpaying for thinking on work that needs none, and they reserve the heavy compute for the handful of decisions where it actually moves a number.

The system: an effort tier you can run as an SOP

Here’s the allocation model I’d hand a team tomorrow. Map the task to the tier, not the other way around.

• Execution work — run it on Low–Medium. Subject lines, ad variants, captions, reformatting, summarizing, first-draft boilerplate. High volume, near-zero reasoning. Running these on High is pure waste: slower output, faster limit burn, no quality lift.

• Strategy work — run it on High (the default). Newsletter drafts, positioning, campaign architecture, creative briefs, prospect research, messaging. Real reasoning pays off, but you don’t need maximum horsepower.

• High-stakes single-shot work — reserve Extra–Max. The analysis you’ll actually act on, exec and board deliverables, complex builds — anything where a wrong answer is expensive. The cost is justified only when the downside of being wrong is bigger than the cost of the compute.

And the toggle: leave Adaptive thinking on. It lets the model decide, message by message, when a turn actually needs deep reasoning — so a working session that bounces between “rewrite this line” and “build me a launch plan” doesn’t waste horsepower on the trivial turns. The only reason to switch it off is a dedicated strategy session where you want every question, even the small-looking ones, getting full depth.

If you do nothing else: the realization that 80% of marketing AI work is execution, not reasoning should change your default. Most teams are running all of it on High and getting nothing for the premium.

Why this is an operations problem, not a tool problem

Here’s the part that matters more than any one model.

The control will change. The names will change — Low, High, Max today; something else next quarter. What won’t change is the underlying discipline: matching the resource you spend to the value of the task in front of you. That’s the same instinct that tells you where your ad budget actually moves pipeline, or which meeting deserves your senior person and which doesn’t.

This is the whole thesis of what I write here. The teams that survive the AI era aren’t the ones chasing the cleverest prompt or the newest model. They’re the ones who understand their tools at the control level — and who build the allocation logic into how the team works, so it holds up when the tool underneath it shifts again. The model is the variable. The operating system is the constant.

Do this week…

1. Drop your default to Medium for execution work. Watch how much faster you move and how much further your limits stretch.

2. Turn Adaptive thinking on and leave it.

3. Pick the one task this week that genuinely deserves Max — and save the heavy compute for that.

That’s the entire system. Allocate effort like budget, and you’ll get more out of the same tool than the marketer next to you who never touched the slider.

I'm Nicholas Putz. Systems & Signal is where I break down moves like this one into operating systems you can actually run — for marketers who'd rather understand what's happening than panic about it. If this was useful, the video above is the fast version, and a forward to one operator who needs it is the best thing you can do for the channel.

Subscribe now

Leave a comment

Share

When Google’s VP of Search called their latest update “the biggest upgrade to our iconic search box since it debuted over 25 years ago,” my first reaction wasn’t surprise. It was relief.

Not because I have it all figured out, nobody does. But because I’d already spent the better part of two years quietly reshaping how I work around exactly this. The search box turning into a destination that answers, books, and builds instead of a doorway that sends people elsewhere. AI surfacing the sources it trusts instead of the pages that ranked. I’d been betting on it. So when the announcement landed, it felt less like an earthquake and more like a forecast I’d already packed for. Also let’s be honest, I have not been quiet about this.

I’m writing this because I suspect a lot of you got hit by that announcement differently. And I’d rather you interrogate your own practice now, on your terms, than get blindsided in a leadership review six months from now. So I’m going to walk through a few of the shifts I made not as a victory lap, but as a set of questions for you to hold up against your own work.

I stopped treating referral traffic as the scoreboard. Have you?

A while back I made a decision that felt heretical at the time: I stopped letting organic sessions be the headline number in how I judged whether the work was working.

Here’s what pushed me. I watched zero-click climb quarter after quarter and realized in a WTF moment I was about to be in a position where my traffic could decline while my actual influence grew and I’d have no way to tell that story if sessions were my only language. Now that Google has shipped AI Overviews to 2.5 billion people a month and an AI Mode that answers a billion users without sending most of them anywhere, that decline isn’t a risk. It’s the design.

So the question I’d put to you: if your organic sessions drop 18% next quarter for reasons that have nothing to do with your performance, can you walk into your leadership meeting and tell a story that isn’t an apology? Because if referral traffic is still your headline metric, you can’t. You’ll be blamed for the mall closing while everyone forgets the mall is now inside Google.

I’m not telling you to ignore traffic. I’m asking whether you’ve built a way to measure success that survives traffic going down. I had to. You might want to start.

I shifted my energy to the only thing that compounds. Where’s yours going?

Once I accepted that referral traffic was structurally leaking, the next question was obvious: what grows instead?

For me the answer became branded search and entity authority (thank you Mordy Oberstein for your inspirations on this) being one of the small handful of sources an AI model reaches for when it synthesizes an answer in my domain. Not ranking number one for a keyword. Being one of the three to seven sources the machine can’t build its answer without. So that’s where I started pointing my efforts: showing up consistently, across many trusted places, saying what I actually need to and what we do well, in our own voice, with points of view nobody else has.

This is the part I’d most want you to sit with. For two decades, our entire craft trained us to chase the exact words people typed and engineer pages to capture them. That muscle is now working against you if it’s the only thing you’re doing. The model doesn’t care about your keyword density solely. It cares whether you’re a recognized entity it can trust.

So look at your own calendar, your own budget, your own quarter. How much of it is still going toward capturing search demand with optimized pages and how much is going toward becoming the kind of brand a machine surfaces because it already trusts you? If the honest answer is “almost all of it is still the old way,” that’s not a failing. That’s just the gap between where the craft was and where it’s going. But it’s a gap you now know about, which means you own what you do next.

I said this back in January, in a piece I called “2025 was the year AI got loud. 2026 is the year brands get real.” I meant it then as a prediction. I’m restating it now as an instruction. The loud phase is over. The question is whether your brand is real enough that the machine vouches for you.

I stopped trusting Google’s guidance. Let me tell you exactly why.

I’ll say the quiet part out loud: I don’t trust Google’s marketing guidance. Not because anyone there is lying, and not as some edgy contrarian pose. I distrust it for a boring, structural reason - Google is a business, and businesses don’t hand competitors the playbook.

This clicked for me years ago, but the moment I started running my own sites across multiple AI surfaces, not just Google’s was proof. You’re going to see guidance from Google, and from a chorus of marketers echoing it, telling you not to bother with things like LLMs.txt, content chunking, or structuring your work to be cleanly extractable by AI. The framing always sounds reasonable: just write for humans, don’t overthink it.

Here’s the thing I noticed. Those exact techniques, the ones Google waves you off of, are what get you cited in ChatGPT, Perplexity, Claude, every AI tool that isn’t Google. Which means the guidance to skip them isn’t neutral advice about what works. It’s advice about what works for Google. And those are not the same thing.

That’s the whole reason for my distrust, and I want to be precise about it: Google’s recommendations are optimized for Google’s outcomes, not yours. When the company benefits from you staying inside its ecosystem and ignoring the surfaces eating its lunch, its “best practices” are going to quietly steer you away from the work that helps its competitors, even when that work is exactly what helps you. A company protecting its moat will always tell you the moat is for your own good. Of course it tells you not to optimize for the tools that threaten it. What else would it say?

So I learned to run every piece of Google guidance through one filter before I act on it: does this serve me, or does it serve them? Pretty much the same thing I do anytime anyone asks me to do something. Sometimes the answer is genuinely “both”, a lot of their technical SEO advice is still sound because clean, fast, well-structured sites help everyone. But on anything that touches the AI search wars, I assume the recommendation is written from Google’s side of the table until proven otherwise. ChatGPT alone reported 900 million weekly users earlier this year. I’m not leaving that surface on the table because the incumbent suggested I should.

The question for you isn’t whether Google is trustworthy as a company. It’s whether you’ve built the habit of asking whose interest any guidance serves, especially the guidance that sounds the most generous. The advice that costs you the most is rarely the advice that sounds self-serving. It’s the advice that sounds helpful while quietly protecting someone else’s business.

What I’d actually have you do today

I don’t write things I can’t turn into action, so here’s what the shifts above look like as homework, for me and for you.

Start tracking three things this week, alongside your traffic, not instead of it. Branded search volume: how many people search for you by name. It’s the trust scoreboard, it compounds, and it’s the number you bring to leadership instead of apologizing for sessions. AI Overview and AI Mode citations: go ask the AI the questions your customers ask, and see whether your name comes back. The gaps are your roadmap. And brand-building itself: treated as the search strategy it now is rather than the soft line item it used to be.

None of this changes overnight, and that’s exactly why it’s dangerous. Slow shifts don’t set off alarms. They just quietly make a playbook stop working while everyone’s still running it. The marketers who win the next three years are the ones who absorb this now and honestly, the signal’s been flashing since ChatGPT and the wave of new platforms started rewiring discovery several years ago. This update just turned the volume all the way up.

I started rewiring roughly two years ago and I still feel a step behind some days. So I’m not writing this from a finish line. I’m writing it from a little further down the same road you’re on, turning around to say: the thing in the distance is real, it’s closer than it looks, and the best time to start adjusting your practice was a while ago. The second best time is today.

Don’t be confused. Be early. The window’s still open, it’s just closing.

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world. Cheers!

Subscribe now

Share Systems & Signal by Nicholas Putz

Leave a comment

I use both. Every day. ChatGPT lives in a Workspace tab in my browser for work. Claude is where I actually get shit done. After months of running them in parallel, the differences aren’t subtle anymore. They’re the difference between a tool you use and a tool that changes how you think.

This isn’t a “which one wins” piece. This is a routing guide. By the end, you’ll know what each tool is actually good at, why Claude’s pricing model works differently than you think (and how to stop setting money on fire), and why the new Opus 4.7 release matters even if you don’t care about benchmarks. Which, let’s be honest, you don’t. Neither do I.

First — Calm Down, They’re Both Fine

Both of these are AI assistants built on large language models. Both will write you a LinkedIn post, summarize a PDF, and help you draft an email to the coworker who says “per my last message” unironically. The question isn’t whether they work. They both work.

ChatGPT (OpenAI) is the ecosystem play. It makes images, makes video, does voice mode, has a marketplace of custom GPTs, and is wired into Microsoft everything. It’s a Swiss Army knife — the one with 47 tools, including the little plastic toothpick you will never use.

Claude (Anthropic) is the depth play. Built by a safety-focused AI lab founded by a bunch of former (dare I say more ethical?) OpenAI researchers who apparently had opinions. It doesn't generate photorealistic images, doesn't do voice mode. What it does do, and this is the part that caught everyone off guard, is design. Claude Design (launched April 2026) turns prompts into on-brand prototypes, slides, and one-pagers that read your actual design system. Add in Artifacts, which renders live interactive components in the conversation, and Claude quietly became the tool that skipped the image-generation arms race and went straight for what marketers actually need: polished visual work that doesn't look like it was spit out of a fax machine.

One is broad. One is deep. You want both. You want to know when to reach for which. That’s the whole game.

I’ll also tell you right now, you need to pay for the Plus / Pro versions of these at least. Please. Do yourself a favor. The free versions… just stop. For many reasons not intended for this piece.

The 7 Real Differences That Matter for Marketers

1. Writing Quality: It’s Not Close and I’m (Not) Sorry

If you’ve ever read AI-written copy and thought “something’s off and I can’t name it” — you were almost certainly reading ChatGPT. Not because it’s bad. It’s competent, fast, and more consistent than most humans I’ve hired. But it has a tell. It defaults to bullet points like it’s being graded on them. It opens paragraphs with “Certainly!” or “Absolutely!” like a waiter at a chain restaurant. Every output is secretly a listicle wearing a trench coat.

Claude writes like an actual person who has read books written after 2008. Varied sentence rhythm. Real paragraph transitions. A vocabulary that doesn’t recycle “leverage” and “unlock” every 40 words. For brand voice, long-form, editorial, and anything a real customer is going to read — the difference shows up instantly in how much editing you actually do.

Practically: Claude’s drafts need maybeeee 20% of the cleanup ChatGPT’s do. Revision cycles eat your velocity. Claude wins the creative lane.

Where ChatGPT still wins: high-volume templated content. Product descriptions at scale. FAQ answers. 400 variations of the same landing page headline. When you need predictable and consistent more than you need soul, ChatGPT might still be the right call.

2. Context Window: The Gap Narrowed, But Claude Still Holds Its Memory Better

Here’s where a year ago, had I been using Claude then, I could have told you it had a huge structural lead. Today it’s more nuanced — which is the honest version.

Quick refresher so nobody’s lost: “context window” is how much text the AI can “hold in its head” during one conversation. Your prompt, your attached files, the whole back-and-forth — all of it competes for the same mental space.

• ChatGPT (GPT-5.4 line): ~400K tokens in the product, up to 1M via the developer API

• Claude (Opus 4.7, Opus 4.6, Sonnet 4.6): a full 1M tokens — now the default on paid plans, not a beta

So the raw numbers are in the same zip code. What hasn’t closed is how well each tool actually uses a long context. Claude’s recall across a half-million-token conversation is the most reliable I’ve used in production. ChatGPT gets a little foggy in the middle — it forgets things you told it 40,000 words ago the way my wife forgets where she puts her car cell phone.

Practical translation for marketers (8th grade mentality): you can load your entire brand book, three years of campaign archives, your audience personas, and your competitive positioning deck into one Claude conversation and trust that page 247 of the brand guide is still influencing the output. That’s not a spec sheet. That’s a different kind of work surface.

3. Projects: Where You Actually Stop Starting From Scratch

Both tools now have a feature called Projects — a persistent workspace where you upload files (brand guidelines, voice examples, personas, competitor research) and they stay loaded for every conversation you start inside that workspace.

Both have this. Claude’s implementation is better for marketing work.

Why: Claude’s Projects do a cleaner job handling document context and, critically, they don’t commingle with a separate “memory” feature that quietly drifts over time. ChatGPT’s memory is a useful feature, but it also means the tool is remembering things about you it shouldn’t be inserting into your brand work. I don’t need my CPG brand copy subtly flavored by the fact that ChatGPT remembers I once asked about how to properly treat my lawn because I fired my lawn treatment guy.

In practice: I have a tons of different projects. A personal Brand Project. A Client Project. A X Product Development Project. Each one already knows who I’m writing for, what I sound like, and what I’ve shipped. No re-briefing. No “as an AI language model, I don’t have context on…” Yes you do. I uploaded it. Read it.

4. Reasoning Quality: Strategy Memo vs. Research Briefing

Ask both tools the same strategic question — say, “should a PE-backed industrial development company invest in brand before exit, and why?”

ChatGPT gives you a research briefing. Well-organized, accurate, readable. Perfectly fine. You could put it in a deck.

Claude gives you a strategy memo. It surfaces second-order implications you didn’t ask about. It pushes back if your framing has a hole in it. It’ll tell you when you’re asking the wrong question. It thinks more like a senior strategist than a research assistant.

For positioning work, channel strategy, competitive analysis, or anything where the quality of the thinking shows up in the output — this is the real gap. And it doesn’t show up in any benchmark number. You just feel it.

5. What Claude Doesn’t Do (Being Honest)

Claude does not generate images. Does not generate video. No voice mode. If your workflow involves rapid visual concepting, social graphics, storyboarding, or audio — that’s a ChatGPT lane, full stop. Honestly, I never even tried to ask it to because it does everything else so well.

The workflow I’ve settled into: Claude for all text-based strategy and creative, coding, and CoWork business on my personal machine. Honestly, I’m using Grok and Gemini for visuals. That’s it. That’s the whole routing logic.

And because Anthropic apparently decided 2026 is the year they ship everything, there's also Claude Cowork — a desktop app that turns Claude into an actual agent for knowledge work. You point it at your files and documents, and it handles multi-step tasks across them on its own. Think: "review these six campaign reports and find the underperforming channels," or "go through this folder of PDFs and pull the pricing data into a spreadsheet." It's early, it's a little confusing (even marketing VPs at real companies have publicly admitted they're not sure where it fits yet), but if you spend a lot of your week doing document-heavy work, it's worth an afternoon of your time to figure out. I personally love it.

6. Instruction-Following: Claude Actually Listens

Give Claude a 500-word brief with layered constraints — tone, format, what not to include, persona, word count, forbidden phrases — and it holds all of it throughout a long output. ChatGPT drifts. Especially past the 600-word mark, it starts reverting to its defaults like a dog slowly wandering back to the couch it’s not allowed on.

For templated workflows and repeatable content systems, this instruction fidelity is a legit operational advantage. If you’ve ever written a 1,200-word prompt and watched ChatGPT forget clause three by paragraph two, you already know.

7. Artifacts: The Quiet Differentiator Nobody Talks About

This is the one most marketers haven’t figured out yet, and it’s the most fun.

Claude has a feature called Artifacts — it renders live, editable code and interactive prototypes right inside the conversation. You describe what you want in plain English, and it builds a working version you can see and iterate on. Want a landing page? A calculator? A dashboard mockup? A lead magnet someone can actually click around? You spec it. Claude builds it. You tweak it by telling it what to change.

I’ve built three live production tools this way — an analytics dashboard, a lawn intelligence tool, a GEO audit tool, and a marketing competitive research review roast lab — and every one of them started as a Claude artifact. Each one became a usable asset in my daily routine.

ChatGPT has a similar feature called Canvas, and it’s gotten much better. But for actually shipping something a client or prospect can use, Artifacts is still the better surface.

For a creative or marketer who has ever wished they could build the thing instead of just describing it in a deck — this is the feature nobody is selling you on and you should be using.

The Token Thing: The Part Nobody Explained to You

Okay. If you’ve made it this far, you deserve the one piece of knowledge that changes how efficiently you use either of these tools.

People think “tokens” are a billing detail. They’re actually the physics of how these tools work. Understand tokens and you stop feeling like you’re randomly hitting limits. Ignore them and you will absolutely set money on fire, possibly while wondering out loud why AI is so expensive.

What a Token Actually Is

A token is roughly three-quarters of a word. “Marketing” is 1 token. “I need help with my marketing strategy” is about 8 tokens. Everything counts — what you type in, every file you upload, every previous message in the conversation, and what the AI writes back.

Every interaction is a two-way token transaction. You pay both ways.

Why This Hits You on Claude

Claude’s subscription plans (Pro at $20/mo, Max at $100–$200/mo) don’t give you unlimited chats — they give you a token budget per time window. Bigger inputs = faster burn. Longer conversations = faster burn. Re-uploading the same brand guidelines in every new chat = also faster burn, and also, may I say, tragic.

The single biggest trap new Claude users fall into: they treat it like a chat app. Long rambling questions. Pasting the same context every session. Asking things the conversation already contains. All of it eats tokens. None of it adds value. You are, essentially, buying the same groceries twice.

How to Not Do That (Five Rules)

1. Use Projects. Stop re-pasting. Load your brand guidelines once into a Project. Every conversation in that Project inherits them automatically. You’ll save yourself thousands of tokens a week and roughly 11 minutes of soul-crushing copy-paste.

2. Be specific, not wordy. A sharp 50-word prompt beats a rambling 300-word one almost every time. Claude is really good at inferring intent. You don’t need to over-explain like you’re briefing an intern on their first day. “Write a 150-word LinkedIn post announcing our Q2 leasing results in a confident, no-jargon tone” beats three paragraphs of throat-clearing.

3. Pick the right model for the job. Claude has three models and using the wrong one is like driving a truck to pick up a sandwich:

• Haiku 4.5 — fast and cheap. For summarizing, reformatting, quick drafts, simple rewrites.

• Sonnet 4.6 — the everyday workhorse. 90% of marketing work should live here. Excellent writing, strong reasoning, reasonable token cost.

• Opus 4.7 — the heavy lifter. Complex strategy, long document synthesis, multi-step reasoning. Heads up: Opus 4.7 uses an updated tokenizer that can burn through more tokens than Opus 4.6 did for the same input. Don’t use Opus to write tweets. Your plan limits will hate you.

4. Don’t start over — refine. Stay in the thread. Saying “shorter and punchier” costs almost nothing. Opening a new chat and re-pasting all your context costs hundreds of tokens. Build on what’s there.

5. Watch long conversations. The longer a thread gets, the more tokens the entire history costs on every single response. At some point, start a fresh conversation with a tight summary brief. Think of it as cleaning out the trunk of your car — nothing dramatic, just occasional hygiene.

Claude Opus 4.7: What Just Dropped

For the people who like to know what’s actually new:

• 1 million token context window — the full book, not just chapters

• Adaptive thinking — Claude now scales its reasoning depth to the task. Easy question = less compute. Complex question = more. You don’t have to tell it.

• Much better vision — it can now read images up to 2,576px on the long edge (~3.75 megapixels, more than triple the previous limit). Screenshots, mockups, dense diagrams now come through at real fidelity. Genuinely useful for anyone who sends a lot of “look at this design — what do you think” messages.

• Better long-horizon execution — it can run multi-step tasks more autonomously and check its own work before reporting back. Still not HAL 9000. Still not trying to be.

• New “xhigh” effort setting — a middle tier between high and max reasoning. Mostly relevant for API users building pipelines, not chat users.

The honest caveat: that updated tokenizer I mentioned means the same input costs more tokens in Opus 4.7 than it did in Opus 4.6. If you upgrade to Opus for everything, you’ll hit your limits faster. Use Opus deliberately. Let Sonnet carry the day-to-day.

How I Actually Use Both (Steal This)

Claude handles:

• Brand strategy docs (Sonnet) — reasoning and context retention

• Long-form editorial (Sonnet) — writing quality and brand voice

• Campaign positioning (Opus 4.7) — strategic depth and instruction fidelity

• Social copy and short-form (Sonnet) — fast iteration inside a Project

• Competitive analysis and deep synthesis (Opus 4.7) — long docs, pattern recognition

• Interactive tools, prototypes, calculators (Artifacts) — actually ships working things

• On-brand visual assets, slides, mockups (Claude Design) — design system baked in

• Code and automation scripts — industry-leading at this

• Emotional Support — (Haiku) You think i’m going to bother someone else with my problems when I have this???

ChatGPT handles:

• Image and video generation — DALL-E and Sora are the right lane

• Keyword research and SEO data pulls — better browsing, plugin ecosystem

• Templated content at volume — consistent, predictable, faster

The Bottom Line

ChatGPT isn’t going anywhere and you shouldn’t necesarilly dump it. Its image generation, ecosystem integrations, and marketplace of specialized GPTs solve real problems. If you need one tool and one tool only, ChatGPT at $20/month is still a very capable all-in-one.

But if you’re a professional marketer doing serious brand, strategy, or content work — Claude is the better engine for that lane. The writing is demonstrably better for anything a real human will read. The context window and Projects mean you stop losing your brand brain every new session. The reasoning produces strategy, not summary. And Artifacts quietly lets you build things that used to require a developer and two weeks.

The learning curve is mostly just understanding tokens. Once you stop burning them carelessly, Claude becomes one of the highest-leverage tools in a marketer’s stack — full stop.

Route deliberately. Use ChatGPT for visuals (or Grok or Gemini), volume, and ecosystem plays. Use Claude for anything where the quality of the thinking shows up in the output — strategy, brand, long-form, and whatever you’re planning to ship. For a professional marketer in 2026, that’s most of the job.

Also: stop being scared of it. It’s a tool. Tools don’t replace marketers who can think. They replace marketers who can’t route.

Nicholas Putz is a marketing director, brand strategist, and founder of PureDigital LLC. He has 20+ years of experience building marketing systems at the intersection of brand, technology, and revenue. He writes at nicholasputz.com.

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world. Cheers!

Leave a comment

Share

I opened Claude Design’s research preview the moment I had access — genuinely excited, not performatively so. Anything Anthropic ships these days has earned the benefit of the doubt in my workflow, and a dedicated design surface was the one gap I’d been waiting for them to close. That said, I’ve been a Creative Cloud subscriber since the Macromedia era (this was obviously before their subscription model) and I know Canva’s keyboard shortcuts by muscle memory. The honest question in my head wasn’t will this be any good? It was where does this actually slot into a stack I’ve spent fifteen years building?

The answer showed up in the first five minutes, before I’d even created anything.

It starts with your brand, not a blank canvas

The thing that caught me off guard — genuinely, immediately — was that Claude Design’s onboarding isn’t “pick a template.” It’s “let me learn your design system.” And because Anthropic’s team correctly assumed that most modern websites and apps live in a Git repo, I could connect my personal brand site (nicholasputz.com) directly and let Claude pull the typography, color tokens, spacing scale, button treatments, and component patterns straight from the codebase.

Context for anyone who hasn’t done this: I just spent months building out and tweaking what I felt was finally a proper design system for my personal brand. Codified every color, every heading weight, every radius. It was arduous in the way only branding work can be — a hundred small decisions each pretending to be trivial. The idea that a new tool was going to make me redo all of that from scratch was the single biggest reason I’d historically bounced off AI design tools.

Claude Design didn’t make me redo it. It read the repo and absorbed the system. I tweaked a few tokens, signed off, and we were live.

That alone was a godsend.

The 15-minute test

Once the system was dialed in, I needed something to throw at it. I picked a short promotional video for a recent article of mine — a piece on why JAMstack has eaten WordPress’s lunch for modern B2B sites. My assumption was an hour of iteration, minimum, to get to something I’d actually publish.

It took fifteen minutes.

And the output included motion and VFX elements that historically I’ve only been able to pull off inside Adobe After Effects or Apple Motion — the kind of work that used to mean evenings of keyframing or billing a specialist. I described what I wanted. Claude built it. I refined. Done.

I've been doing this long enough to know when something genuinely shifts and when it's just novelty. This is a shift.

Adobe is already losing, and the market knows it

Let me be plain about this, because the market signal is louder than the product launch.

Adobe (ADBE) is not a sleeping giant about to be disrupted. Adobe is a company that is actively being disrupted and has been for a couple of years now — by Canva on the accessibility and collaboration front, by Figma on the product-design front, and now by a new class of AI-native tools like Claude Design on the generation front. The walled-garden strategy that carried them for twenty years — a dense, interconnected suite where every app assumes fluency in the others — is the exact wrong posture for a market where the alternative to learning your suite is “describe what you want in a sentence.”

You can see this in the numbers. You can see it in the product releases that feel reactive rather than visionary. You can see it in the stock reaction: Figma fell roughly 7% on secondary markets and Adobe dropped about 1.5% the day Claude Design launched. That’s not a market pricing in a future threat. That’s a market pricing in a present one.

Canva, to their credit, read the room years ago. They’ve been aggressive on AI features, they restructured their whole content pipeline around generative tooling, and they’re now leaning into the partnership with Anthropic rather than fighting it — Claude Design exports directly into Canva for collaborative editing. That’s the move of a company that understands the tool layer is commoditizing and the value is shifting to the workflow. Adobe is still trying to protect the tool layer.

Let me be explicit about what I’m not saying: designers and VFX artists are not done. Taste, art direction, craft at the highest levels — that’s still human work. But if you’re a designer or motion artist and you’re not fluent in these tools right now, you’re choosing to compete with one hand tied behind your back. Adopt early. Fight later.

For those of us on the marketing side, this is the thing I keep coming back to: being able to produce professional creative by describing it removes the bottleneck that has slowed every marketer I know for nearly two decades. It’s a relief. And it’s a renewed sense of urgency, because whatever you think the rate of change is, it’s faster.

Quick education: how Claude Design’s metering actually works

A practical note, because people keep asking me this and the answer matters for anyone planning to use it seriously.

Claude Design usage is tracked separately from the rest of your Claude subscription. It is not part of your chat allowance. It is not part of your Claude Code allowance. Anthropic gives each user a dedicated weekly allowance for Claude Design that sits alongside — not inside — your other limits, resets every seven days, and is granted per user rather than pooled across the org.

The reason this matters: design generation is genuinely heavier than a text conversation. A prototype with VFX, multiple component states, and an applied design system chews through far more compute than asking Claude to summarize an email thread. Metering it separately means your regular chat workflow doesn’t get starved by a design-heavy afternoon, and your design work doesn’t get throttled because you were deep in a research session that morning. It’s the right architectural choice.

If you blow through the weekly allowance, you can purchase additional capacity, billed at standard API rates. There’s also a one-time credit on the house for new Claude Design users — roughly 20 typical prompts’ worth — which expires July 17. Use it.

For transparent perspective, the video above ate up my entire weekly allowance. :)

If you’re a ChatGPT user considering the jump

I get asked this frequently, so I’ll be direct rather than diplomatic (nobody likes my diplomatic tone anyway, it’s boring).

ChatGPT and Claude are optimized for different things. OpenAI’s product surface is broad — image generation, voice, a huge consumer footprint, a plugin ecosystem aimed at mass market. Anthropic has spent the last two years building what I’d describe as a professional-grade surface: Claude Code for engineering, Projects for sustained knowledge work, Artifacts for building real things in-conversation, and now Claude Design for creative output that integrates directly with how modern teams actually ship software.

The models themselves diverge in ways that matter if you’re doing marketing operations work:

• Long-context reasoning. Claude holds a 200K-token context (500K on some Enterprise models) and uses it coherently. If you dump a 40-page RFP, a brand guide, and a competitive teardown into a single session, Claude stays on-thread. This is the single biggest unlock for marketers managing complex accounts.

• Instruction-following and tone control. In my hands-on work across both platforms, Claude is substantially better at holding a voice — your voice, the client’s voice, a persona’s voice — across a long conversation without drifting to a generic register.

• Tool and codebase integration. Claude Code and Claude Design both plug into your actual Git repos, your actual design systems, your actual infrastructure. That’s a different posture than “here’s a chat window, good luck.”

• Fewer safety-theater refusals on legitimate business work. This one’s anecdotal, but every marketer I’ve compared notes with has the same experience.

Is Claude objectively better at everything? No. ChatGPT is still strong on consumer use cases and has a broader third-party ecosystem today. But for the work I do — brand strategy, web architecture, content systems, and now visual design — Claude is the platform I reach for, and it’s not close.

If you’re considering switching and want to talk it through, email me or DM me on LinkedIn. Happy to walk you through my setup, what I use each tool for, and how to think about the transition without losing your ChatGPT-based workflows mid-flight.

One last thing

There’s a line I think about constantly, attributed to Eric Hoffer:

"In times of change, learners inherit the earth, while the learned find themselves beautifully equipped to deal with a world that no longer exists."

Every few years the industry produces a tool that forces a decision. Photoshop did it. The web did it. Mobile did it. Figma did it. Generative AI is doing it now, on a timeline compressed by an order of magnitude. The people who will be thriving in marketing and design two years from now are not the people with the most credentials or the longest Adobe Creative Cloud tenure. They’re the people who are in the tools right now, this second, breaking things, iterating, paying attention.

Don’t stop exploring. Don’t stop experimenting. Get dirty. Break some things. Learn how to build better. The rate of change isn’t slowing down, and waiting for things to stabilize is just a slower version of being left behind.

Nicholas Putz is a marketing director, brand strategist, and founder of PureDigital LLC. He has 20+ years of experience building marketing systems at the intersection of brand, technology, and revenue. He writes at nicholasputz.com.

If this resonated — share it with someone carrying the same problem. That’s how this community grows.

Connect with me on LinkedIn

Subscribe now

Share

There is a story the web development industry tells itself — and by extension, tells you. It goes like this: WordPress powers 43% of the internet, it’s reliable, it’s familiar, and your agency knows it cold. Why would you want anything else?

That story is self-serving. And it’s costing your business more than you realize.

We are living through a genuine architectural inflection point in how the web is built, how it performs, and how search engines — including AI-powered ones — evaluate it. The businesses that understand this shift will gain a structural advantage over competitors who are still being managed on the same WordPress stack their agency deployed in 2019. The ones who don’t understand it will keep paying monthly retainers for “SEO work” that amounts to little more than blog posts nobody reads.

First, let’s give credit where it’s due

WordPress wasn’t always the liability it’s becoming. In 2005, it democratized publishing. It gave businesses a CMS they could manage without a developer on call. It spawned an ecosystem of plugins, themes, and page builders that genuinely lowered the barrier to having a professional online presence. For its era, it was remarkable.

But the web has changed in ways WordPress’s architecture was never designed to accommodate. The internet WordPress was built for — one where a server renders a page and ships it to a browser on every request — is a fundamentally different animal than the internet we operate in today.

43% of the web runs on WordPress — most of it bloated and slow

~50% of WordPress sites have at least one vulnerable plugin installed

2–5s typical WordPress TTFB vs. sub-200ms for modern static sites

The architecture argument isn’t theoretical

What’s replaced the old server-rendered model is something the industry calls JAMstack — JavaScript, APIs, and Markup. The name is a little nerdy, but the principle is elegant: instead of building a page from scratch every time a visitor arrives, you pre-render your content into static files at build time and serve them from a global CDN. There’s no database to query, no PHP to execute, no plugin stack to slow you down. The page is already built. It’s just waiting to be delivered.

The performance difference isn’t marginal. We’re talking about time-to-first-byte measured in milliseconds versus seconds. Core Web Vitals scores that are genuinely excellent, not optimized-with-a-plugin almost-passing. Security surfaces that are a fraction the size of a WordPress install, because there’s no server-side execution layer to exploit.

Frameworks like Astro, Next.js, Nuxt, and SvelteKit are what serious developers are building with today. Deployment platforms like Vercel and Netlify make shipping these sites trivially easy. And AI-assisted development tools — Cursor, Claude, GitHub Copilot — have collapsed the time and cost required to build them. What once required a full-stack team can increasingly be done by a single developer, or in some cases a technically sophisticated marketer, in a fraction of the time.

The WordPress tax you’re already paying

Here’s the honest accounting of what WordPress costs you, beyond the hosting bill:

WORDPRESS REALITY

• Server-side rendering on every page load

• Plugin conflicts that break without warning

• Security patches on a rolling, stressful schedule

• Performance that requires constant plugin band-aids

• Database dependency that creates single points of failure

• PHP hosting costs that scale with traffic spikes

• Page builders that generate 300KB of DOM garbage

• Vendor lock-in to your agency’s workflow

JAMSTACK REALITY

• Pre-rendered at build time, delivered from CDN edge

• No plugins — composable APIs instead

• Static files have near-zero attack surface

• Sub-200ms TTFB, near-perfect Core Web Vitals

• No database means no database to corrupt or breach

• CDN hosting is free or near-free at most traffic levels

• Clean, semantic HTML that search engines actually parse

• You own the stack, not your agency

SEO in 2026 is not the SEO your agency is selling you

The SEO packages most agencies sell were designed for a search landscape that no longer exists. The model was simple: publish lots of content, build links, optimize meta tags, watch rankings rise. WordPress was purpose-built for this content-farm model. Publish a post, it’s live. Publish five hundred posts, you’re a “content machine.”

That model is collapsing under three simultaneous pressures. First, Google’s algorithm has become sophisticated enough to identify and discount thin, undifferentiated content at scale. Second, AI search — Perplexity, ChatGPT Search, Google’s AI Overviews — doesn’t work the way traditional search does. It synthesizes. It reasons. It looks for authoritative, structured, clearly-sourced information, not keyword density. Third, Core Web Vitals are now genuine ranking signals, not advisory guidelines. A slow WordPress site with great content will lose to a fast JAMstack site with comparable content.

What does this mean practically? The $2,000/month retainer your agency is charging you to publish twice-weekly blog posts is buying you less ranking power every quarter. The content itself may be fine. The platform it’s sitting on is working against you.

Now let’s talk about the agencies

This is where the conversation gets uncomfortable — but it needs to be said.

A significant portion of digital agencies operating today have built their entire business model around WordPress. Their developers know it. Their project management workflows are built around it. Their templating libraries, their staging environments, their client onboarding processes — all WordPress. Switching costs for them are enormous. Switching costs for you are not their problem.

When an agency tells you WordPress is the best platform for your needs, ask yourself: what would it cost them if it weren’t? Because the incentive to tell you otherwise simply doesn’t exist in their business model. They’re not lying, exactly. They’re just answering the question from inside a framework that has a very specific right answer baked in.

The question to ask your agency: “Can you show me the last three sites you built on a modern JAMstack framework, and their Core Web Vitals scores?” If they can’t answer it — or if the answer is “we primarily work in WordPress” — you are paying for their comfort, not your performance.

The agencies worth working with in 2025 are the ones who have already made the transition. Who can speak fluently about Astro or Next.js. Who have an opinion about headless CMS options like Sanity or Contentful. Who are building on deployment infrastructure that auto-scales and auto-secures without your involvement. Those agencies exist. They’re just not the ones running Google Ads for “affordable WordPress websites.”

The AI dimension changes everything

There is a final argument that rarely gets made, but matters enormously right now.

Large language models — the engines behind ChatGPT, Claude, Perplexity, and AI search — are trained on, and retrieve from, the web. But they don’t retrieve the way Google’s crawler does. They favor structured, semantic, well-organized content. They reward clarity over volume. A 1,200-word explainer page with clean schema markup on a fast-loading static site is far more likely to be surfaced in an AI answer than a sprawling WordPress archive with four years of undifferentiated blog content buried under plugin overhead.

The optimization game has changed. The winners won’t be the ones with the most content. They’ll be the ones with the most structured, authoritative, performant presence — the kind that modern frameworks make easy and WordPress makes genuinely difficult to achieve at scale.

So what should you actually do?

You don’t have to blow up your existing site tomorrow. But you should be asking different questions of whoever manages your web presence. Ask about Core Web Vitals. Ask about their deployment infrastructure. Ask what it would take to move to a headless or static architecture. Ask whether their SEO strategy has been updated for AI search — or whether it’s the same playbook from 2018 with updated pricing.

If the answers are vague, defensive, or heavily WordPress-centric, you have your answer. Not about whether WordPress is bad. About whether your agency is evolving fast enough to keep your business competitive.

The web is being rebuilt. Not dramatically, not all at once — but steadily, by developers who have access to better tools, better frameworks, and AI assistants that make building fast, modern sites faster than ever before. The businesses that understand this and act on it now are buying themselves an advantage that will compound for years.

The ones still paying for WordPress retainers in 2027 will wonder what happened.

Nicholas Putz is a marketing director, brand strategist, and founder of PureDigital LLC. He has 20+ years of experience building marketing systems at the intersection of brand, technology, and revenue. He writes at nicholasputz.com.

If this resonated — share it with someone carrying the same problem. That’s how this community grows.

Share

I want to be careful with how I say this, because I know how it sounds. But I’m going to say it anyway:

I am now building simple, fully functional, production-grade software products — alone — faster than most developer teams I’ve ever worked with.

Not prototypes. Not demos. Products that work, automate, integrate, and compound value over time. I’m doing this as a marketer. And if you’re reading this and you build things for a living — in any capacity — you need to understand what is happening right now.

CONSIDER THIS YOUR WARNING

If you are a web developer, SEO specialist, app builder, or technical marketing hire who has not yet seriously engaged with AI-native building tools, your role is being quietly eaten from beneath you. Not by AI directly — by people like me who understand the strategy and can now execute the engineering.

What changed

For nearly 20 years I have sat at the intersection of marketing and technology. I understand websites, application architecture, APIs, deployment pipelines, SEO infrastructure — not because I could build all of them by myself, but because I understand what they are for and why they matter. That gap — between knowing and building — was where developers lived.

Claude CoWork finally closed that gap for me.

Not because it writes some random code I paste somewhere. Because it lets me think in systems and execute in systems simultaneously. I describe what I want at the detailed strategic layer, and an autonomous agent works through the technical layers — scripting, API wiring, deployment, design — without me stopping to translate.

"The dangerous kind of builder isn't the engineer who learned marketing. It's the marketer who learned how to build."

Two things I built that I couldn’t have built alone before

I won’t give you the architecture. That’s reserved for me and me only. But I’ll share with you two examples and tell you what they do and what they represent.

The affiliate engine. Tested several affiliate marketing sites in the past built on Wordpress, Shopify, etc. Very time consuming. Without total dedication to the production end of things, it can easily falter. So this time I wanted to try something new. I built a fully automated affiliate marketing engine from the ground up. The input is simple: pick a product in my niche, drop it into a spreadsheet. What follows is a complete automated pipeline — API calls, content generation, SEO optimization, page creation, visual design, and live deployment — all without me touching any of it again. Work that used to take hours per product now takes thirty seconds. The system runs while I sleep. I built it once. It scales indefinitely. Completed to its fully functional, white-label form, in about 18 hours total. Deployable in a new white-label format in less than an hour.

The lawn care dashboard. This one started out of frustration. I fired my lawn treatment guy who I used for 13 years and replaced him in a weekend because he really upset me. Instead of complaining, I built myself a tool. An AI-powered dashboard that tells me exactly what to apply to my lawn, when, which products to buy, where to find them cheapest near me — and automatically adds every treatment to my Google Calendar. I completely replaced a professional service. I now know more than that service did. And my cost per month is pennies in API usage and my time in the sun.

• 30s per product through the affiliate engine, down from hours

• ~$0 marginal cost per task in AI tokens versus full-service fees

• 100% owned IP — architecture, logic, and system design stay mine

On Claude versus ChatGPT

I still use ChatGPT. I use it at work where it makes sense as we have an isolated Workspace for it. But I want to be direct about something: Claude.ai by Anthropic has made ChatGPT feel like a previous generation of tool to me, at least at this point until it possibly leapfrogs one day.

It’s not about which one writes a better paragraph. It’s about which one I trust to hold the complexity of a system in its reasoning while I build. Claude does that with a precision and coherence that ChatGPT, in my experience, doesn’t come close to matching at this time. The accuracy is higher. The integrations feel more natural. The agentic workflows are tighter.

When I’m building something that has moving parts including logic, automation, real outputs that go somewhere, I use Claude. That’s not a partnership anymore. That’s a personal preference with evidence behind it.

The business opportunity hiding in plain sight

Here’s the part most people haven’t caught up to yet.

Every system I build has proprietary IP baked into it — the architecture decisions, the prompt engineering, the workflow logic, the integrations. These aren’t generic tools. They’re built for specific outcomes, and the knowledge required to replicate them isn’t in a YouTube tutorial. It’s in my head, sharpened by days of iteration.

That is a business. People will pay for access to that knowledge. They already are.

I’m not just a marketer anymore. I’m not just a consultant. I am building a team of agents that execute on my behalf, compressing weeks of work into hours, and in some cases hours into seconds. And the IP those agents generate compounds over time.

“I’m not building with AI. I’m building a team with AI. The distinction matters enormously.”

What this means for you

If your job exists somewhere in the stack between marketing strategy and technical execution — content, SEO, web builds, app development, campaign infrastructure — you are closer to a reckoning than you probably realize.

This isn’t a hypothetical future. I am doing it today. Other people who think like marketers and understand complex systems are doing it today. The question isn’t whether this disruption is coming. It’s whether you’re on the building side or the disrupted side when it arrives.

My advice is simple: stop waiting for permission to understand the tools. Pick something you’ve always needed but never had the technical help to build. Build it. Break it. Rebuild it. The gap between strategy and execution is narrowing to nothing, and the people who cross it first will own the next decade of this work.

I’m just getting started. And that’s what makes it so exceptionally exciting.

Share

Leave a comment

In January, I wrote about tightening the operating system behind my work.

Less motion. More proof. A weekly decision system as the centerpiece — a single source of truth that shows what changed, what matters, and what to do next.

I wasn’t pitching anything. I was just being honest about what I needed to build.

A few people messaged me after that post. Most of them said some version of the same thing:

“I’ve been trying to build something like that for two years. How are you going to do it?”

The honest answer at the time was: I wasn’t sure yet.

I had the conviction. I had the problem defined. I didn’t yet have the instrument.

The thing nobody tells you about marketing data

You probably check your analytics. You probably have dashboards. You probably have charts.

And you probably still walk into Monday feeling like you have no idea what actually happened last week.

That’s not a data problem. That’s an interpretation problem.

Here’s the gap that nobody in this industry wants to say out loud: most marketing leaders are running on impressions data, gut feel, and whatever their agency told them — because the alternative is spending 90 minutes in GA4 to come out more confused than when they started.

I’ve been that person. I’ve sat across from leadership with a set of metrics that looked fine and felt completely unable to explain why performance didn’t match the effort. The data was there. The insight wasn’t.

GA4 is a warehouse, not an analyst. It stores everything and explains nothing.

And here’s the compounding problem: when you can’t interpret your data fast, you start making decisions on volume. More posts. More campaigns. More spend. Hope that something moves a number somewhere.

That’s not strategy. That’s noise with a budget.

What I actually built

I’m not going to describe the architecture. That’s not the point.

What I’ll tell you is what it does.

Every week, I open one view. It tells me what’s working, what’s breaking, what shifted, and — critically — what it means and what to do next.

Not a chart. Not a metric dump. An interpretation.

Something like: “Organic traffic is down 12% week-over-week, but branded search held flat, which means this is likely a content indexing lag following last week’s site update — not a rankings problem. Watch for recovery over 7–10 days. In the meantime, your highest-converting content cluster is significantly outperforming — consider shifting paid budget to amplify what’s already working.”

That’s a different experience than staring at a line trending downward and wondering if you should panic.

The difference is intelligence applied to data, not just data displayed.

It doesn’t require a data team. It doesn’t require a BI tool. It doesn’t require you to become an analyst in addition to everything else you already are.

Who this is for

I built this because I needed it. But I’ve been a marketing director long enough to know that my problem is not unique.

If you’re carrying a full marketing function — multiple campaigns, multiple channels, leadership reporting, budget accountability — and you’re doing it without a data analyst on staff, you’re making more judgment calls in the dark than you should have to.

You know the data is there. You know it should be telling you something. But extracting the signal from the noise takes time you don’t have and expertise that wasn’t in the job description.

The result? You’re confident in your instincts and uncertain about your proof.

That uncertainty is the thing that makes budget conversations harder, leadership updates vaguer, and strategy decisions slower than they need to be.

The answer isn’t more tools. You have enough tools.

The answer is intelligence — senior-analyst-level interpretation delivered as a service, without the overhead of a hire.

The shift that changes everything

Most marketing directors are one conversation away from being perceived differently by leadership — and they don’t realize it.

The conversation where you stop presenting what happened and start explaining what it means. Where you stop defending spend and start connecting activity to outcomes. Where you stop reacting to questions and start answering them before they’re asked.

That’s the shift from reporting to advising. From showing up with a chart to showing up with a point of view.

It’s not a title change. It’s not a budget change. It’s a data confidence change — and it’s available to anyone who can walk into the room already knowing the story the numbers are telling.

If you’ve ever felt like you were one level below where your judgment actually sits, that gap is almost always a data confidence problem disguised as a positioning problem.

Where I’m taking this

I spent the first two months of 2026 building, testing, and refining this system against real data.

Now I’m making a version of it available to a small exclusive group of marketing directors and B2B leaders who are carrying the same problem I was carrying.

This is not a software subscription. It’s not another dashboard that requires you to learn a new interface. It’s a managed intelligence service — set up for you, interpreted by a practitioner who has run the exact role you’re running, delivered as clear, actionable analysis on a rhythm that matches how you work.

I’m keeping the initial cohort deliberately small. The nature of this work requires actual engagement, not just a login.

If you want to understand what this could look like for your organization — one conversation, no commitment — I’ll show you what your data has been trying to tell you.

Nicholas Putz is a marketing director, brand strategist, and founder of PureDigital LLC. He has 20+ years of experience building marketing systems at the intersection of brand, technology, and revenue. He writes at nicholasputz.com.

If this resonated — share it with someone carrying the same problem. That’s how this community grows.

Share

Leave a comment

My step-son Jack is six years old. He’s curious, funny, obsessed with fire trucks and race cars, and he happens to be autistic. He’s also what’s known as a gestalt language processor — or GLP.

If you haven’t heard that term, you’re not alone. I hadn’t either — not until my wife Kindy explained it to me. She’s the one who first recognized that the way Jack was learning didn’t match what the apps and traditional programs expected. She did the research, connected the dots, and helped me understand what I was seeing. Most of what I know about how Jack’s mind works, I learned from her.

Here’s the short version: most kids learn language analytically — sound by sound, letter by letter, building words from pieces. GLP kids do the opposite. They learn in whole chunks. They pick up entire phrases, absorb complete words as visual shapes, and build understanding from context and immersion. They don’t decode — they recognize.

Once you understand that, you realize why almost every reading app on the market is built for the wrong kid.

We Are Incredibly Fortunate

Before I go any further, I need to say something clearly: Jack is surrounded by extraordinary people.

His teachers and paraprofessionals show up every single day with a level of patience, creativity, and dedication that I genuinely don’t have words for. They work tirelessly to guide, mentor, and shape Jack into the sweet, growing boy he is. They adapt. They learn alongside him. They celebrate his wins and carry him through hard days. The work they do is not easy, and it is not recognized enough.

Our school system — our community — could not possibly be better. The resources available for kids like Jack where we live are exceptional, and we know that’s not the case everywhere. We are lucky, and we don’t take that for granted for a second.

This isn’t about what’s missing in Jack’s school. It’s about what I, as a parent, with a particular set of skills ( Liam ;-) )can do to extend and support the incredible work his team is already doing.

The Race You Didn’t Train For

Here’s the thing nobody tells you when your child is diagnosed, or when you start to notice that they’re learning differently than other kids: you have to become a specialist in something you’ve never studied, overnight, while still living your life.

I met Jack when he was about six months old. I’ve been his father figure ever since. I’ve watched him grow from a baby into this amazing kid, and every stage has brought new joys and new surprises. But some of those surprises came with steep learning curves — and no instruction manual.

And I need to be honest about something: when I describe the work of learning about autism, GLP, and how to support Jack — I’m mostly describing what Kindy has done.

She is the true powerhouse behind everything that makes Jack’s world work. She’s the one who has spent countless hours researching, reading studies, talking to specialists, and translating clinical language into things we can actually do at home. She’s the one who sat in the waiting rooms. She’s the one who found the right therapists. She’s the one who understood what gestalt language processing meant before anyone explained it to us, because she watched her son closely enough to see it.

I came along for the ride. I learned from her. Everything I’ve been able to try and build for Jack flows from the foundation Kindy laid — the understanding she fought to gain while simultaneously being his mom every single day.

When you’re a parent navigating autism, GLP, sensory processing, IEPs, and the alphabet soup of therapies and assessments, it feels like a race against the clock. You’re trying to learn fast enough to help. You’re absorbing terminology you’ve never seen, reading research you were never trained to interpret, sitting in meetings where professionals use frameworks you’re hearing for the first time — and then going home and trying to apply it all in real time. With dinner to make. With a job to do. With a life that doesn’t pause while you figure it out.

And the fear of failure is constant. Not the small kind. The kind where you lie awake wondering if you’re doing enough. If you’re doing the right things. If you missed something you should have caught earlier.

Kindy and I carry that together, and I’ll say this honestly: compared to many families, we have it exceptionally easy. Jack’s support system at school is remarkable. He’s healthy. He’s happy. We have resources many families don’t. I know that. And yet it’s still overwhelming.

That’s the part I think people outside of this world don’t fully understand. And that’s okay. Even when things are going well — even when you’re fortunate — the cognitive load of learning your child’s needs while meeting them in real time can be staggering.

The Resource Gap Is Real

When Jack started his IEP at school, I went looking for tools that matched how he actually learns. Games that emphasize whole-word recognition. Apps that present complete, contextual phrases instead of isolated phonics drills. Anything that would let him explore and absorb language the way his brain naturally works, with the things he enjoys.

I found almost nothing.

There are thousands of phonics apps. Hundreds of letter-tracing games. Dozens of “learn to read” programs that break every word into its smallest possible pieces — the exact opposite of what a GLP learner needs. The few resources that did exist for gestalt processors were PDFs, therapy worksheets, or academic papers. Nothing interactive. Nothing that felt like a game. Nothing Jack would actually want to play. He’s incredibly gifted at the games he plays. Many parents would gasp at the fact we let him play some of the games he plays, but seeing how advanced he is at such a young age and his ability to actually play them, gives me hope for his abilities to learn complex subjects down the road.

For a while, I did what most parents in this situation do: I said “there just aren’t resources for this” and moved on, well, not entirely. That’s not my nature..

I never accepted that. There is a solution for literally everything.

Building What Doesn’t Exist

I’m not a game developer. I’m a full-stack marketing professional and technologist. I know how to build things on the web, I know how to architect complex strategies, and more importantly, I know how to learn fast using the most advanced tools available today.

So I started building. Using AI tools as a development and learning partner — not to replace my thinking, but to become a force multiplier — I began creating free, browser-based games designed specifically for gestalt language processors. These games are being updated constantly as I get time. Iteratively. Nothing is perfect, and this is a learning experience for me as well, to see what Jack wants and needs, and adjust as needed.

The first game concept was Word World City. It’s an explorable city scene where Jack can tap on anything — a fire truck, a building, a cloud — and the whole word appears. No sounding out. No phonics breakdown. Just the word “FIRE TRUCK” in big, friendly letters, spoken aloud, connected to the thing he’s looking at. That’s the GLP way: see the whole word, connect it to meaning, hear it in context. This has since expanded into several other mini-games such as Construction Site and Fire House.

Then came Firefighter Hero, because Jack asked for it. He wanted to put out fires. So we built a game where burning household items show their word first — “TELEVISION,” “BLANKET,” “TOASTER” — and he sprays them with water to reveal the object. Every word is seen, heard, and experienced in a complete, thematic context.

Then Word Match Challenge — a card-flipping memory game with 11 themed categories, from Zoo Animals to Construction Site. Flip a word card, flip an image card, find the match. The whole-word recognition builds through playful repetition, not drilling.

Every game is free. Every game runs in a browser. Every game is designed around the same core principles: whole-word recognition over phonetic decoding, immersive thematic learning over isolated drills, and exploration-driven curiosity over pressure and time limits.

Now, admittedly, these aren’t perfect. They are intended to be simple. I will continue to learn, iterate, and find the positive reinforcement and reward mechanisms these kids need to find joy, empowerment, and success through playing these games.

What AI Actually Made Possible

I want to be clear about what the AI tools did and didn’t do here.

They didn’t replace my understanding of Jack. They didn’t generate the pedagogy. They didn’t know what a gestalt language processor needs. That knowledge came from Kindy first — from the years she spent learning Jack’s world — and then from his speech therapist, from research, from watching him struggle with the wrong tools and light up with the right approach.

What AI did was collapse the timeline between having an idea and holding working games he could try.

When I understood that Jack needed word-first visual learning from Kindy, I could describe that interaction pattern and have working code the same day. When Jack said he wanted a firefighter game, we had a playable prototype that evening. When I realized the games needed better UX mobile support for his iPad, the fixes happened in a conversation, not a development sprint.

But the deeper value wasn’t just speed. It was access to understanding.

I think about what it would have been like to navigate all of this without these tools. Before AI, if I wanted to understand gestalt language processing at the level I require (which is a lot), my options were limited: find the right classes at a community college, pack in seminars and webinars on top of an already maxed-out schedule, or spend hours watching videos hoping to find the ones that actually applied to Jack’s specific situation.

Now I can ask questions in plain language and get clear, contextualized answers. I can say “explain how whole-word recognition differs from phonetic decoding for a GLP learner” and get an explanation I can actually use — not a textbook chapter, not a 45-minute lecture, but a clear answer that meets me where I am, when I need it. And then I can say “now help me build a game that uses that approach” and actually do it.

That’s not a small thing. For parents like Kindy and me, who are simultaneously learning about all three of our boys’ needs and trying to meet them while holding down careers and keeping life together — these tools have been a lifesaver. Not an exaggeration. A genuine, tangible shift in what’s possible for our family.

Facing every new surprise in Jack’s development without these tools would have been a different story. Kindy would have kept fighting regardless — that’s who she is — but it would have been harder and lonelier. For me, these tools give me a way to take the understanding she built and turn it into something tangible the way I know how. Instead of feeling like I was just trying to keep up, I could finally contribute. We went from overwhelmed to equipped. Not experts — but equipped.

What I’ve Learned About Jack (And About Learning)

Building these games has taught me more about how Jacks mind works than any report or assessment ever did.

I learned that he doesn’t need things slowed down — he needs them presented whole. He can recognize “EXCAVATOR” as a complete word shape faster than most adults would expect, because he’s not trying to sound it out. He’s seeing it.

I learned that context is everything. The word “FIRE TRUCK” means more when it’s sitting on a road next to a fire station with a dalmatian out front than when it’s floating on a white flashcard.

I learned that exploration beats instruction. Jack doesn’t want to be told “tap the fire truck.” He wants to discover it. The moment of tapping something and hearing the word is a reward in itself.

And I learned that when you build something for one kid, I could be building something that helps a lot of families.

This Is Only the Beginning

I genuinely believe we are living at the start of something transformative for neurodivergent kids and their families.

For the first time in history, a parent who has no background in special education, no clinical training — just a deep knowledge of their own child, and a formidable understanding of the tech stack — can sit down and build tools tailored to exactly how their kid learns. Not generic tools. Not “close enough” tools. Tools built around the specific way their child’s mind works. This goes for us too.

Never in human history has this been possible before.

And it’s not just about building games. It’s about parents finally having a way to bridge the gap between what they observe at home and what the professionals are doing at school. It’s about being able to walk into an IEP meeting not just with concerns, but with context. Not just asking for help, but contributing meaningfully to the plan.

For neurodivergent kids — GLP learners, kids with ADHD, dyslexia, sensory processing differences, and every other way a brain can be beautifully wired — this could open doors that didn’t exist a year ago. Personalized learning tools. Games that adapt to how they think. Resources that meet them where they are instead of forcing them into a mold that was never built for them.

If you’re a parent in this world, I want you to hear this: you are not helpless. You are not behind. The tools exist now, and they are more accessible than you think. You don’t need to be a formal developer. You don’t need a background in education or technology. You need to know your kid(s) — and you already do.

Take the leap. Ask the questions. Build the thing. It doesn’t have to be perfect. It just has to be built for your child. More importantly, use the tools to empower YOURSELF to be that force multiplier your family deserves.

All of the games I’ve created so far are free and open at glpgames.com. No subscriptions, no ads, no data collection. Just games built for kids like Jack. If they help your kid(s), that’s the whole point. If they don’t, but they inspire you to build something that does — even better.

The era of “there are limited/no resources for this” is over. We just have to build.

If you want to get started somewhere, feel free to download Perplexity using my code
pplx.ai/nicholaspu44750

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world. Cheers!

Subscribe now

Leave a comment

Share

If you are anything like me, you’ve probably looked at your utility bill recently and thought, “What on earth is going on?”

I knew my electricity and natural gas costs were up significantly, but I also knew my actual household usage wasn’t astronomical in comparison to past years. I got fed up, so I sat down and did a massive forensic deep-dive into my last 13 months of billing history with We Energies and Anthropic AI’s Claude.

I expected to find the usual suspects: aggressive base rate hikes (which did happen—the energy charge jumped over 5.5% into 2026) and frustrating new fees like the “Fuel Cost Adjustment” that just started hitting our bills in January.

But what Claude actually found was a quiet, parasitic tax drain that the state has been taking from us every single month. And the best part? You legally do not owe it, and you can get your money back.

Here is the rundown of what happened, how I figured it out, and the exact instructions you need to follow to stop the bleeding.

Disclaimer: I am a marketing professional, not a CPA, tax professional, or attorney. The information in this article is for educational and informational purposes based on my own experience auditing my utility bills. Please consult with a qualified tax professional regarding your specific situation and verify all information with the Wisconsin Department of Revenue.

The “Aha!” Moment Hidden in the Fine Print

While Claude combed through my line items, tt noticed a recurring charge: a 5% “WI State Tax” applied to both my electric and natural gas totals every month. At peak usage, this was costing me $15 to $23 a month in taxes alone.

Then, Claude looked at the “Messages” section buried on Page 2 of my October 2025 bill. There, hidden in plain sight, was this sentence:

“Per Wisconsin state law, a primary residence is exempt from sales tax. If this is your primary residence, please contact us at customerservice@we-energies.com.”

What Actually Happened?

Historically in Wisconsin, residential energy bills were only tax-exempt during the winter heating months (November through April). But a new state law was passed that expanded this exemption: As of October 1, 2025, electricity and natural gas sold for residential consumption is 100% exempt from state sales tax year-round.

So why are we still paying it?

Because the utility companies’ billing systems are not tied to municipal property tax records. They know your house is residential, but they do not automatically know if it is your primary permanent residence or a secondary vacation cabin.The law states the tax exemption only applies to permanent primary residences.

Because they don’t know, they default to charging you the 5% tax. They rely entirely on you to reach out and explicitly declare your exempt status. If you don’t catch it, the state just keeps parasitically taking that 5% from your wallet month after month. For my household, that is easily $50 to $80+ a year gone for absolutely no reason. That’s a Friday night of Chick-Fil-A for us!

How to Fix Your Account Right Now

You need to take five minutes today to plug this leak. Here is exactly what you need to do to stop the state from taking taxes you don’t owe:

Step 1: Stop Future Taxes You have to declare your primary residence status to your utility provider.

• Email We-Energies immediately at customerservice@we-energies.com.

• Include your name, account number, and service address.

• State clearly: “This service address is my permanent primary residence. I am claiming my year-round sales tax exemption under Wisconsin state law.”

• Note: They may reply and ask you to fill out a quick form called the Wisconsin Department of Revenue Form S-211 (Wisconsin Sales and Use Tax Exemption Certificate) to make it official.

Step 2: Get Your Refund (The Claw-Back) Stopping future taxes is great, but what about the money they already wrongfully collected? You are legally entitled to recover the sales tax paid in error, and the statute of limitations generally allows you to look back up to four years.

Because you paid the tax to We Energies and not directly to the state, you cannot just ask the utility for a cash refund. You have to file a “Buyer’s Claim for Refund” directly with the Wisconsin Department of Revenue.

• The Easy Way (Online): You can file this electronically using the WI Department of Revenue’s “My Tax Account” portal. You will need to tally up the taxes you paid in error from your old bills and input them.

• The Old School Way (Paper): If you prefer paper, you need to fill out Form S-220 and Form S-220a (Schedule P). You fill out Sections 1 and 3 of the S-220a, send it to We Energies so they can sign off that you actually paid the tax, and then mail the whole packet to the Department of Revenue in Madison.

We are already battling rising base rates, volatile natural gas markets, and new PSCW-approved fuel surcharges. We shouldn’t be giving away an extra 5% out of ignorance.

Check your bills, send the email, and get your money back. And please, forward this to your Wisconsin friends and neighbors so they can stop overpaying, too!

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world. Cheers!

Leave a comment

Share

In June 2023, I wrote an article on LinkedIn titled “The Importance of Prioritizing Design Software in Your Marketing Department: A Closer Look at Adobe vs Canva.” In it, I made the case—passionately, with 30 years of Adobe fluency behind me—that Canva was a “fast-food style” design tool and that marketing departments should lead with Adobe and treat Canva as a supplement for “non-critical item deliverables.”

I stand here today to tell you: I was wrong. Not entirely, but substantially. And the market agrees with me.

When I wrote that article, Adobe’s stock was trading near $480 a share. As I write this, it sits at $246. That’s a 46% decline over five years, while the S&P 500 gained 79%, the Nasdaq surged 71%, and Apple climbed 119% over the same period. Adobe didn’t just underperform—it collapsed relative to the market it once helped define.

This article is a correction, a reflection, and hopefully, a useful lens for anyone leading a marketing department in 2026. Let’s face it, you probably had more foresight on this than I did.

The Moment That Changed My Mind

Recently, I opened Canva for the first time in a year and a half. I needed a 4K video ad, the kind of deliverable I’d normally build in Adobe Premiere Pro, a tool I’ve used professionally for over a decade. I expected to be underwhelmed.

I wasn’t in any way.

The experience was shockingly intuitive. I produced a polished final product faster than I would have in Premiere, all in a browser, with nothing installed on my local machine. No render queue. No media cache corruption. No codec nightmares. Just... a finished video ad, ready to deploy.

That single session broke something loose in my brain. It forced me to reconcile the reality of where these tools actually are today versus the loyalty I’d built up over three decades.

The Numbers Don’t Lie: Adobe’s Freefall

Let’s talk about what’s happened since my original article.

Adobe’s Stock: A Cautionary Tale

Adobe traded above $635 at its peak in late 2021. Today it’s at $246. That’s a loss of over $170 billion in market capitalization (technically, that’s conservative). In December 2024 alone, shares plunged 14% in a single day after management issued disappointing revenue guidance for fiscal 2025. Multiple investment banks, Deutsche Bank, Wells Fargo, Barclays, UBS, all cut their price targets. The central question haunting investors: can Adobe actually monetize AI, or is it getting disrupted by it?

The Failed Figma Acquisition

In September 2022, Adobe attempted to acquire Figma for $20 billion, an admission that its own collaborative design tools couldn’t compete. EU and UK antitrust regulators blocked the deal. Adobe paid a $1 billion breakup fee for the privilege of losing.

Figma then went public in July 2025, soaring 250% on its first day of trading and briefly reaching a market cap of nearly $68 billion, more than three times what Adobe had offered to pay. Figma crossed $1 billion in annual revenue in 2025 with 40% year-over-year growth. The company Adobe tried to absorb is now valued at multiples of the acquisition price, proving the market sees more innovation potential in Figma alone than in Adobe’s acquisition strategy.

The FTC Lawsuit

In June 2024, the FTC sued Adobe, alleging deceptive “dark pattern” subscription practices, specifically that the company trapped consumers in annual contracts disguised as monthly plans and made cancellation intentionally difficult. The lawsuit cited hidden early-termination fees that could consume up to 50% of remaining payments. A separate class action lawsuit was filed in 2025 making similar allegations. These aren’t the hallmarks of a company that’s winning on product merit. These are the hallmarks of a company that knows its users want to leave.

Meanwhile, Canva Built an Incredible Empire

While Adobe was defending regulatory actions and watching its stock crater, Canva was executing one of the most impressive product expansions in the design software industry.

Canva’s Growth Story

Canva now serves over 260 million monthly active users across 190 countries. Revenue has skyrocketed from roughly $1 billion in 2021 to an estimated $3.5 billion in 2025. More than 95% of Fortune 500 companies use Canva, and the platform’s enterprise segment (companies with 25 or more seats) hit $500 million in annual recurring revenue with 100% year-over-year growth. Over 60 million students and teachers use Canva for Education, for free. This isn’t just a “template tool for small businesses” anymore. This is enterprise infrastructure.

Strategic Acquisitions That Actually Work

Unlike Adobe’s failed bid for Figma, Canva has executed a series of smart, strategic acquisitions that directly strengthen its product ecosystem:

Affinity (2024, ~$380M): The professional-grade design suite, Affinity Photo, Designer, and Publisher, that many designers were already using as their Adobe escape hatch. Canva acquired it and then made the all-new Affinity completely free for everyone. Professional vector editing, photo manipulation, and page layout—free. Perpetual licenses preserved. No subscription trap. This single move demolished one of my core 2023 arguments: that Canva couldn’t serve professional designers.

Leonardo.AI (2024, ~$370M - unconfirmed): A generative AI platform with its own foundation model (Phoenix) and over 19 million users. Leonardo’s technology now powers Canva’s Magic Studio, which has been used more than 7 billion times since launch. The platform reports 800 million AI tool uses per month—a 700% year-over-year increase.

Flourish (2022): Interactive data visualization, now built into the platform.

Pixabay & Pexels (2019): Free stock photography libraries giving users a massive built-in asset ecosystem.

Each of these acquisitions followed the same pattern: buy something excellent, integrate it thoughtfully, make it more accessible, and keep it affordable. It’s the exact opposite of Adobe’s playbook of acquiring competitors to eliminate them.

Where I Was Right, And Where That No Longer Matters

In my 2023 article, I made several specific claims about Canva’s limitations. Let me revisit each one with humility:

“Canva lacks advanced features and customizability.” With the Affinity acquisition, Canva now owns professional-grade vector editing, advanced photo retouching, and full page layout tools. The “customization gap” I warned about has been filled, and the tools are free.

“Adobe is the industry standard for professional design.” This remains partially true in specialized fields such as high-end VFX, Hollywood color grading, complex print production. But for marketing departments? The “industry standard” argument is weakening fast. When 95% of the Fortune 500 uses Canva, when Figma has dethroned Adobe XD, when DaVinci Resolve is replacing Premiere Pro in YouTube studios and film houses alike, the standard is shifting rapidly.

“Adobe’s suite integrates seamlessly.” Canva’s all-in-one browser platform now handles presentations, video editing, social media design, print materials, data visualization, AI image and video generation, websites, and brand management, all in a single freaking browser tab. No Creative Cloud sync issues. No update cycles that break your workflow. No scratch disk errors. No freaking spinning wheel of death.

“Canva requires internet access.” This is still true. But in 2026, reliable internet is nearly ubiquitous for knowledge workers, and the advantages of browser-based collaboration far outweigh the rare scenarios of offline work. With Affinity available for local desktop use, Canva’s ecosystem now covers both bases.

“Your files are trapped in Canva’s cloud.” Affinity’s desktop apps let you save files locally. And Canva’s export options have expanded significantly. The file ownership concern is no longer a meaningful differentiator.

The Real Problem: Adobe’s Hubris

What’s happening to Adobe isn’t a mystery. It’s the predictable consequence of a company that chose extraction over innovation that I should have seen years ago had my head not been shoved so far up Adobe’s... (XD):

The subscription model became a trap, not a service. The FTC doesn’t sue companies that make it easy to cancel. Adobe’s subscription practices were designed to maximize retention through friction, not value. When your customers need a federal lawsuit to stop being charged, you’ve lost the moral authority to claim you’re building “for creators.”

AI strategy is unfocused. Adobe Firefly was supposed to be the answer. Instead, it’s absolutely awful in every sense of the word. Firefly faces fierce competition (if you want to even call it that because it’s so bad) from Gemini, Midjourney, DALL-E, Stable Diffusion, Runway, and now Canva’s own Leonardo-powered tools. Analysts have repeatedly highlighted the gap between Firefly’s usage metrics and its actual revenue contribution. Adobe is spending heavily on AI but struggling to show returns, and anything with less than 14 fingers on a single human body.

Acquisition-as-strategy failed miserably. When your biggest strategic move of the decade is a $20 billion acquisition that gets blocked by regulators, and you burn $1 billion on a breakup fee, and then the target company IPOs at a $68 billion valuation, that’s not bad luck. That’s an embarassing failure of organic innovation.

Price increases outpaced value delivery. Adobe hiked its Photography Plan by 50% in early 2025, largely to fund AI features many users didn’t ask for. Long-time users report increasing bugginess, bloated applications, and legacy codebases that struggle to keep pace with modern tools built from the ground up.

The New Competitive Landscape

The design tool market in 2026 looks nothing like it did in 2023 and I’m honestly shocked. Adobe is no longer the only game in town, it’s not even the best game in several towns:

Collaborative design: Figma (now public, $1B+ revenue, 13M+ users)

All-in-one visual communication: Canva (260M users, $3.5B revenue, Fortune 500 enterprise penetration)

Professional photo/vector/layout: Affinity by Canva (free, perpetual licenses, no subscription)

Video editing: DaVinci Resolve (free professional-grade NLE used in Hollywood), CapCut (social-first video)

AI image/video generation: Midjourney, Runway, Leonardo.AI (Canva), plus OpenAI and Google tools

The “walled garden” advantage Adobe once had (the idea that you needed the full Creative Cloud because everything worked together) has been replaced by an open ecosystem of specialized, often free, tools that integrate better than Adobe’s own applications do with each other.

My Updated Recommendation for Marketing Leaders

If I were writing my 2023 article today, the conclusion would be inverted:

Lead with Canva and its ecosystem. Supplement with Adobe only where specialized, irreplaceable workflows demand it.

For most marketing departments, Canva’s platform now handles 90%+ of what you need: social media graphics, presentations, video ads, brand asset management, team collaboration, print materials, website mockups, and AI-powered content generation. The Affinity suite fills the remaining professional design gaps. DaVinci Resolve handles the video work Premiere Pro once monopolized. Figma owns the product design and prototyping space.

Upon much reflection, the only scenarios where Adobe remains essential in my view are highly specialized production environments: complex motion graphics in After Effects, enterprise-scale print production with specific prepress requirements, or deeply embedded workflows where the switching cost is genuinely prohibitive. Those use cases exist but they represent a shrinking slice of what marketing teams actually do every day.

A Humbling Conclusion

I spent 30 years as a diehard Adobe evangelist. I defended the ecosystem through every price increase, every subscription change, every frustrating update cycle. I genuinely believed—and said publicly—that tools like Canva were creating a culture of “disposable, forgettable marketing content.”

What I failed to see was that the world didn’t need a $60/month subscription to create compelling visual content. It needed tools that got out of the way and let people create. Canva understood that from day one. Adobe still doesn’t.

The market has spoken. From $635 to $246. From Figma rejection to Figma IPO. From “industry standard” to FTC defendant. Giants can fall—and they fall fastest when they mistake their installed base for loyalty, their lock-in for value, and their legacy for relevance.

If you’re leading a marketing department and still paying for full Creative Cloud licenses across your team out of inertia or tradition—I’d encourage you to do what I did: open Canva, build something real, and then ask yourself honestly whether the Adobe premium is delivering premium results.

For me, the answer was humbling. But it was also liberating.

This article is a follow-up to my June 2023 piece: “The Importance of Prioritizing Design Software in Your Marketing Department: A Closer Look at Adobe vs Canva”

Original article: linkedin.com/pulse/importance-prioritizing-design-software-your-marketing-nicholas-putz

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world. Cheers!

Subscribe now

Share

I have been operating in the reality most marketing leaders live in now, wide scope, limited time, and expectations that keep expanding. Over the past few years I have built processes, improved measurement, shipped campaigns, and created internal tools to close gaps where outside help did not fit the real workflow, or budget. The work has been getting done, and the system has held.

2026 is not a reinvention. It is refinement. This is the year I tighten the operating system behind the work, reduce variance, and raise the bar on results. Less activity for activity’s sake, more measurable outcomes tied to growth. The goal is simple, drive revenue with less friction, fewer moving parts, and a clearer standard for what earns time and attention.

That is why I am treating 2026 as a year of focused consolidation, planning, redirection, and revenue.

Not because it sounds disciplined. Because it is required.

Subscribe now

I am responsible for a workload that, in most organizations, would be distributed across a full marketing team. Multiple lines of business. Multiple websites. Multiple campaign streams. Reputation and content. Performance measurement. Coordination across internal teams and outside partners. Budget oversight. Advisory and internal education work on emerging tech.

It… is… a… lot, and pretending it is not a lot does not make it manageable.

For a long time, the hardest part was not effort. It was decision pressure. When everything is a priority, nothing is. So I made a decision that is going to define my year.

I am not trying to do more. I am trying to do fewer things, better, with measurable outcomes.

The real shift, from motion to proof

A lot of marketing looks productive without actually producing outcomes.

You can ship content and still not grow.
You can run campaigns and still not improve conversion.
You can stare at analytics and still not make better decisions.

That is not a talent issue. It is a systems issue.

In 2026 I am prioritizing proof.

Proof that marketing activity connects to revenue.
Proof that messaging is clear enough to convert.
Proof that operations can execute consistently, not heroically.
Proof that work is measurable, repeatable, and scalable.

If I cannot prove it, I am not going to defend it.

My new priorities, firm but focused

I will not share internal specifics publicly, for obvious reasons. But I will share the shape of the decision.

I have set a tighter set of priorities that map directly to growth. That means some projects will get paused, some will get simplified, and some will get cut entirely, even if they are interesting, even if they are popular, even if they once felt important.

This is the uncomfortable part of leadership that people rarely post about.

Saying no is a growth strategy.

The first move, consolidate the foundation

Before you can scale growth, you have to clean up the fundamentals.

If tracking is inconsistent, you cannot trust your insights.
If reporting is fragmented, you cannot make fast decisions.
If your campaign process changes every time, you cannot improve it.
If your website is not built around clear conversion paths, your best marketing leaks.

So my first move this year is consolidation of the foundation.

One measurement approach.
One set of definitions.
One reporting rhythm.
One way of naming and tracking efforts.
One set of conversion actions that matter.

This is not glamorous work. It is the work that makes everything else work.

The strategy, build leverage, not just output

When you do not have a large team, you have two options.

Option one is to grind harder and accept diminishing returns.
Option two is to build leverage.

Leverage is what turns a high workload into a manageable workload.

In 2026, my approach is to build internal systems that handle the repetitive parts of marketing so human attention stays on the parts that require judgment. The goal is not automation for the sake of automation. The goal is reclaiming time, improving consistency, and increasing the quality of decisions.

In practice, that means building three types of leverage.

1. A weekly decision system
   A single source of truth that shows what changed, what matters, and what to do next.

2. A repeatable campaign system
   Templates, checklists, and standard workflows so campaigns are built faster, launched cleaner, and measured consistently.

3. Monitoring and alerts
   Systems that surface problems early, before performance drops and before time gets wasted.

If those three are working, everything else gets easier.

What I am trying to prove by the end of 2026

This year is not about being busy. It is about being able to show results.

By the end of 2026, I want to be able to point to:

A clear growth narrative and positioning that converts.
A streamlined foundation that makes performance measurable and decision making faster.
Repeatable campaigns that improve over time, not restart from scratch.
A content and reputation engine that supports trust and demand.
Less chaos, fewer emergencies, and more consistent revenue impact.

That is the scoreboard.

Not “we posted more.” Not “we tried new tools.” Not “we were everywhere.”

Revenue impact, execution consistency, and measurable improvements.

A note for anyone feeling the same pressure

If you are carrying a workload that feels like three jobs, you do not need a motivational quote. You need constraints.

Here is the filter I am using all year:

Does this increase revenue, reduce cost, improve conversion, protect retention, or reclaim meaningful time.

If the answer is no, it is not a priority. It is a distraction with better branding.

I am not chasing everything in 2026. I am consolidating around what works, redirecting away from what does not, and building systems that let a small team operate like a larger one.

Less noise. More proof.

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world. Cheers!

Leave a comment

Share

In 2025, AI did not enter marketing. It kicked the door in, turned the lights on, and yelled, “I have a dashboard.” Then it sold that dashboard to your CEO, your intern, and your neighbor’s dog.

For many marketers, last year was not “innovation,” it was cognitive overload. A constant feed of launches, trends, hot takes, and LinkedIn inbox pitches from Joe Blow promising a 58 touchpoint AI automation contraption that will eliminate your entire sales staff and magically produce $1,000,058 per month. The pitch always sounds confident. The math never does.

And the software platforms, bless them, kept shipping “solutions” that worked perfectly in a demo and collapsed the moment they met a Tuesday at 2:00 PM. The dashboard looked slick, until you asked who it was built for, how it fits the workflow, and why the person who actually uses it was never asked.

The lesson 2025 screamed at us

I wrote about this in my LinkedIn piece, The Infinite Noise vs. The Lasting Brand: Why brand perception, not funnel tricks, will decide who survives the zero click era. The core point holds, and after 2025 it is even clearer.

Short term funnel hacks are addicted to novelty. Brand is built on trust.

AI is pushing discovery into fewer clicks and more inference. That means the internet’s “word of mouth” is becoming machine readable, and it is becoming the deciding factor. Earned media. Authenticity. Community. Customer satisfaction. Third party validation. These are the assets that compound when everything else is changing weekly.

Think of the AI search and user behavior as becoming smart enough to know when you’re trying to game their system… there’s going to be consequences for that.

The quicksand problem in marketing and SEO

There is a pressure cooker effect happening across SEO and marketing. Competition increases, software gets harsher and more hyper-focused, expectations rise, and everyone starts fighting the next micro battle.

That is where the quicksand shows up.

Companies obsess over hyper granular metrics and tiny optimizations to win a fight that does not change the bottom-line outcome. Everyone is pushing harder for the same result with the same product reality. They keep adding complexity to compete. Then they sink under the weight of it.

The worst part is that it feels like progress. It looks like “more effort.” It shows up as “more activity.” But it often produces the same baseline outcome, just with more stress and now even less clarity and more questions.

I have seen this movie in multiple industries

I am writing this from the multifamily and commercial real estate seat, but I have worked across healthcare, advertising, and consumer packaged goods. Every tech cycle repeats the pattern.

A new capability appears. Vendors rush to paint it onto what they already sell. Everyone claims differentiation. Most of it is the same thing with a new label, a new slide deck, and a new pricing page.

Right now the label is “AI.”

In 2025, many vendors were not building new outcomes. They were layering AI onto old motions and calling it transformation when the end result was no different than before.

2026 will be a tougher year for vendors, and a better year for operators

2026 is shaping up to be more intense for digital product and service vendors. Not because demand is gone, but because the power balance is shifting.

Businesses now have internal AI builders, AI enthusiasts, and practical operators who can create their own solutions using tools like Google AI Studio, n8n, and other automation platforms. They can vibe-code and prototype fast, test in real workflows, and get to the real outcome without paying for a bloated platform that needs API integration dependencies, and an expensive consultant to explain its own navigation.

So vendors are going to do what vendors do under pressure. They will try to sell clients even more complex services that clients think they need.

The winners will not be the noisiest. They will be the ones who tie AI to clear business outcomes, simple implementation, and real adoption.

For marketers, 2026 is a year of reflection, adoption, and for the motivated, building your own solutions with tools that are finally accessible. The capabilities are close to infinite. The differentiator is creative thinking, operational confidence, and the discipline to focus on what matters.

I have found this clarity and have set my goals for 2026. I won’t sit, wait, and hope someone or another company is eventually going to deliver the kind of system, or information I need with the data I have at hand because some API integration does not exist. I have all the tool creation tools and data at my disposal. Frankly, we all do. I’m going to build the tools and dashboards myself, the way I see fit. That’s all there is to it. I challenge you to do the same.

If you want to win in AI mediated discovery, you need evidence, not tricks

I said this in my Substack piece Diamond Ads Are the New Beige Carpet, Resident Experience Is the Next Moat, and I will repeat it because it matters even more now:

“AI systems and modern discovery surfaces rely on signals that are consistent, repeated, and validated by third parties. That means your reputation ecosystem is not a side project, it is core infrastructure.”

If you want to win in AI mediated discovery, and even the next era of SEO discovery, you need:

• Review volume and review quality that reflect reality
• Survey systems that catch issues early, before they hit Google
• Response standards that show accountability, fast
• Resident and customer stories that showcase real lived experiences, not stock photography
• Social proof that is customer generated, not agency polished

This is why reputation tools, lifecycle surveys, earned media, and operational standards matter. Not because “reviews are nice,” but because AI reads them as evidence.

The resource shift that will pay off

Here is the hard truth: If you’re allocating resources away from metric theater and toward the most authentic, consistent, high trust customer experience you can deliver, the future will reward you.

Not with vanity metrics. With durable demand. With referrals. With defensibility. With lower acquisition costs over time. With talent that wants to work with you. With a brand and product line that holds when channels change.

The play is not more tactics. The play is a better experience, clearly communicated, consistently validated by others.

And if you need a closing North Star, Steve Jobs said it best:

“Don’t talk about the product.”

Talk about the outcome. Prove it through the experience. Let the market, and the machines reading the market, do the rest.

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world. Cheers!

Leave a comment

Learn more about me at nicholasputz.com

Those of us in the multifamily industry have trained ourselves to believe that if lead volume dips, the answer is always the same: buy more impressions, upgrade the ILS package, crank paid search, pump paid social, chase the next attribution model like it is going to hand us salvation.

Here is the uncomfortable truth.

When everyone is forced to buy the premium package to compete, premium becomes the minimum. It stops being advantage and turns into rent you pay for the privilege of staying in the game.

That is why I strongly believe the next real competitive layer is not another channel, it is resident experience. And the AI era is about to make that painfully obvious.

The shift is already happening, most teams are just looking in the wrong direction

In the old model, awareness plus frequency plus follow up created leases. In the current model, the discovery journey is splintered across ILS pages, Google business profiles, TikTok, Facebook, Instagram (Meta), and reviews that get screenshot and shared in group chats.

In the next model, AI becomes the front door.

People will ask their phone, their car, their browser, their assistant, “What is the best place to live near my work with a real community and good management?” They will not open ten tabs. They will not study floor plan galleries like it is a research project. They will get an answer.

That answer will be synthesized from signals.

Reviews. Resident sentiment. Service responsiveness. Consistency. Social proof. Community energy. The stories residents tell when your leasing office is not watching.

So if your current strategy is built on buying the loudest megaphone, you are going to hate what happens when the AI summarizes you as “fine.”

Why resident experience becomes the growth engine in an AI first world

AI does not fall in love with your amenity list. It learns patterns from what people say and do.

If your property delivers generic experiences, the internet will describe it with generic language, and AI will repeat that language at scale.

If your property consistently creates moments residents want to talk about, the internet fills with specific stories, specific praise, and specific trust signals. AI will pick up those patterns and recommend you with confidence.

Resident experience is not a feel good initiative. It is a long-term demand strategy that compounds. No different than dollar-cost-averaging your investments.

Retention lifts, referrals increase, negative review volume drops, your costs to compete in advertising decrease, and your overall cost to lease falls because you are no longer paying to overcome distrust. The best part… positive NOI.

The industry mistake, treating experience like decor instead of infrastructure

Most development planning treats experience like an amenity checklist.

Clubhouse. Fitness room. Dog wash. Maybe a pool if the pro-forma is feeling generous.

That is not experience design. That is furniture.

Experience IS architecture. It is the intentional design of how living there feels across the full lifecycle:

• The first inquiry
• The tour
• The application moment
• Move in
• The first maintenance request
• The first community interaction
• The first complaint
• Renewal season
• And what they say after they leave

If you don’t design those moments, the default experience becomes “transactional housing with occasional inconvenience.” That is a very easy experience for AI to summarize, and a very hard one to defend with more ad spend.

More importantly, if left unchecked, this can destroy your business once the LLM’s are trained. There is almost no going back.

The new playbook, build an Experience Operating System

Here is the framework I am building and pushing internally, because the math and the market both point here.

Layer 1: Experience design that is planned before the building opens

Stop asking “What amenities do we have?” and start asking:

• What outcomes do we want residents to feel and say
• What rituals make those outcomes real
• What standards ensure consistency, even on a busy Tuesday
• What recovery process turns a negative moment into renewed trust

Examples that move the needle far more than another treadmill:

• Personalized move in touches that feel human, not automated
• A 30, 60, 90 day cadence of check ins so residents feel seen
• Lifestyle programming that is actually relevant, not forced fun
• Local partnerships that add real value, not coupons nobody uses
• Resident led groups that create belonging, not just attendance

Layer 2: Signal engineering, because AI recommends what it can verify

This is the part multifamily marketers need to wake up to.

AI systems and modern discovery surfaces rely on signals that are consistent, repeated, and validated by third parties. That means your reputation ecosystem is not a side project, it is core infrastructure.

If you want to win in AI mediated discovery, you need:

• Review volume and review quality that reflect reality
• Survey systems that catch issues early, before they hit Google
• Response standards that show accountability, fast
• Resident stories that showcase lived experience, not stock photography
• Social proof that is resident generated, not agency polished

This is why reputation tools, lifecycle surveys, and operational standards matter. Not because “reviews are nice,” but because AI reads them as evidence.

Layer 3: Automation that supports humans, not replaces them

AI should take friction out of resident life and property operations. It should not turn your community into a call center.

Use AI tools to improve:

• Communication speed, acknowledgements, follow ups, status updates
• Maintenance triage, scheduling, and proactive issue detection
• Personalization of community updates based on resident interests
• Content capture and distribution, turning events into shareable moments
• Staff enablement, scripts, knowledge bases, faster resolution

The goal is simple.

Residents should feel like the community is attentive and organized. Staff should feel like they have leverage, not extra work.

Budget reality, stop funding the top of the funnel while starving the product

If you are spending aggressively to acquire residents, but under investing in the experience they receive after move in, you are buying churn.

It is the same logic as any subscription business.

You can pour money into acquisition, but if the product does not deliver, your unit economics collapse, and your reviews become your worst sales rep.

I am advocating for a clear budget shift:

• Reallocate a defined percentage of paid media spend into resident experience programming, service standards, and reputation systems
• Tie that shift to measurable outcomes, renewal lift, referral lift, review sentiment, maintenance response performance
• Treat the experience layer as an asset, not an expense

If your team cannot measure the impact, you are not running a program. You are running a vibe.

The execution standards that separate serious operators from hopeful ones

Experience fails when it stays conceptual. It wins when it becomes operational.

A few standards I believe should be non negotiable for any brand that wants to earn trust:

• Service recovery follow up within 48 hours after any negative experience, with a clear resolution plan
• Move in experience that includes a welcome touch, a walkthrough, and a follow up call within 72 hours
• Maintenance response standards that residents can predict and rely on
• Communication standards that always acknowledge messages quickly, even when full resolution is still in progress

These are not marketing ideas. These are trust mechanics.

And trust is the currency AI will amplify.

What “resident experience” looks like when it is done right

It is not a monthly pizza party. It is a system that makes residents proud to live there, and confident recommending it.

Here are the outcomes that matter most:

• Residents feel genuinely cared for, not managed
• Residents feel connected, not isolated
• Residents feel proud, so they share and defend the brand
• Residents feel listened to, so issues get resolved before they go public
• Residents feel value beyond the unit, so renewal becomes the default

When those outcomes happen consistently, the marketing becomes easier, cheaper, and more effective. Not because you got better at ads, but because the product got better.

The uncomfortable question every multifamily leader needs to answer

Are you building a lead machine, or are you building a community people love?

One of those gets more expensive every year.

The other compounds.

If you want a real edge in a world where AI compresses the discovery journey into a single recommendation, your best strategy is not louder advertising.

It is a better lived experience.

Because in the end, your residents are the brand. AI is just going to repeat what they already said.

If you are a developer, operator, or marketer and you want to pressure test your current approach, here are three questions worth asking yourself this week:

1. What are the five moments that shape resident perception the most, and are they intentionally designed or left to chance

2. What signals would an AI system find about your community today, and would you be happy with the summary

3. If you cut paid media by 20 percent, what experience initiatives would protect occupancy through retention and referrals

That is where the next competitive bar is headed.

And if we get this right, it is also where multifamily gets more human again.

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world. Cheers!

Leave a comment

Share

On Friday nights some people go out and get drunk. I stay up until 3am building things and flushing out my ideas.

I went looking for a simple way to throw a URL at an assistant and see how well it could explain the page back to me. Nothing out there did what I needed in a concise enough manner, so I built my own tool, called GEO Crash Test, and hosted it myself as a working concept.

Here is the idea. Paste a public URL, get a GEO score for AI search readiness, see where assistants are confident and where they are guessing. Then you get a short list of fixes you can actually act on.

The problem I keep running into

Search is shifting from ten links on a page to one answer in a chat. People ask an AI, get a single response, and move on with their day. If models cannot cleanly restate what you offer, your brand never appears inside that answer.

That is not theoretical anymore. It is already happening across assistants, chat search, and the “ask first, click maybe” behavior we are all falling into.

I kept asking the same question in my work as a marketing and technology evangelist:

“If an assistant had to explain this page in one paragraph, would it sound like us, or like vague corporate noise”

Most teams cannot answer that. So I decided to make a tool that would.

Meet GEO Crash Test

Use this tool to check how understandable your page is to AI assistants. Paste a URL, see a GEO score, and get clear fixes.

That is the entire pitch.

What it does

GEO Crash Test currently simulates how an AI assistant tries to read your page, then scores how easy it is to reuse your content in an answer.

You test one URL at a time. That focus keeps the conversation practical. You are not arguing about an entire site strategy. You are looking at one concrete page and deciding what to change next.

The inputs and outputs

• Input one public URL

• Output an overall GEO score, category scores, and three recommended changes worth doing first

Behind the scenes the service fetches your HTML, strips scripts and styling, compresses the content, and runs it through a structured evaluation that mirrors how assistants reason through a page. Scores roll up into an overall GEO score and category scores, plus priority fixes.

It runs on Cloudflare Workers with OpenAI; the frontend is static HTML, CSS, and JavaScript.

For now it is free to use. There is an optional tip jar if you want to help cover the model bill.

Who I built the GEO Crash Test for

Marketing and content teams
You just shipped a new landing page, product page, or thought piece. Before you celebrate, you want to know if assistants can restate the offer without butchering it.

Owners of complex sites
You have portfolio pages, solution pages, FAQ clusters, and you suspect AI is guessing about sectors, locations, or pricing. You can feed in a few critical URLs and see where the cracks are.

Agency and fractional leaders
You need a neutral score when you explain why content, schema, and internal linking need attention. GEO Crash Test gives you a shared language that is not just “because I said so”.

When you should NOT use GEO Crash Test
If the page is behind authentication, contains confidential data, or is still a rough draft, do not test it. Finish the content first, then run the score.

How the score thinks about your page

Every result you see contains the same category set. These categories are designed to match how assistants actually “experience” your content.

• Structure
  Headings, sections, and hierarchy. Does your page read like a clear outline, or as one giant block of marketing fluff

• Clarity
  Who you are, what you do, for whom, and where. Can an assistant answer those basics without guessing

• Coverage
  Common questions real buyers or residents ask. Does this single page carry enough substance to support useful answers

• Schema and signals
  Schema, metadata, FAQs, internal links. Are you giving models clean hooks for facts, or forcing them to reverse engineer everything

• Assistant legibility
  How well your page compresses into a machine friendly summary without losing meaning

• Technical basics
  Titles, descriptions, crawlability, and general performance signals that decide whether your page is visible at all

In the result view you see scores for each category, a short narrative summary, and three priority fixes. That is the piece that matters. GEO Crash Test is designed to drag the conversation back to “what should we change this week” instead of letting it drift into theory.

Screenshots from the lab

You can see this in action in the screenshots here.

1. Input view
   The test page where you drop in a URL, tap “Run GEO Crash Test”, and let the system do the messy work in the background.

   

2. Result view
   A result card with the overall GEO score at the top, category scores in individual cards, and a simple list of priority fixes at the bottom.

   

This is not meant to be a dashboard addiction. It is a quick “pull over, check the gauges, decide what to fix first” moment.

Why I keep building things like this

I live in the intersection of AI, marketing, and real businesses that actually need to lease apartments, move product, or close complex B2B deals.

That means I am constantly asking:

• How do we make AI search and assistants work for brands that are not tech giants

• How do we design content so it feeds both humans and models without becoming jargon soup crap

• How do we give marketing professionals and technologists better tools without turning them into full time prompt engineers

GEO Crash Test is one small answer to those questions.

I want modern marketing to catch up to modern technology, not by copying whatever is trending this week, but by understanding how discovery is actually changing and building systems around that.

So I keep pushing into new use cases. Agentic search. GEO. Machine friendly views of content. JSON outputs that tools and assistants can reuse. The lab section on my site is where these experiments land first.

Where this goes next

Every time someone runs a test, the system gets a little smarter about how different sectors present themselves. Over time there will be anonymized sector context so you can see rough comparisons to peers without handing me your data.

The scoring method, weights, and prompts will keep evolving as AI search shifts. The public categories stay stable so you have a reference point. The guts under the hood will keep changing so the signal stays relevant, as often as I can get to them.

My broader plan is simple.

• Keep shipping practical tools that reveal how AI search actually sees our work

• Give marketers and technologists a cleaner way to talk about that reality with their teams and clients

• Use those lessons to shape better content systems, not just better dashboards

An invitation

If you work in marketing, product, or you own a business, pick one important page and run it through GEO Crash Test.

Notice where the assistant understands you perfectly. Notice where it hesitates or points out gaps. Then change one thing this week based on those findings.

You can try it here: nicholasputz.com/labs/geo-crash-test.

If you want to compare notes, or apply this across a portfolio of sites, there is a contact form on the page. I read emails and I reply relatively quickly.

And if GEO Crash Test helps you and you feel like buying me a coffee or a little crypto, the tip options are there too. Completely optional, just appreciated.

Modern marketing is about to be defined by how clearly we speak to humans and machines at the same time. Tools like this are my way of helping us get there a little faster.

<!— IMPORTANT NOTE —>

One important caveat, not every website stack plays nicely with the kind of requirements the future may require at this time. That’s not to say they won’t. The path to a stronger GEO score will depend heavily on how your site is built and where it lives. Some setups are very simple to tune. Others will demand more technical work than most marketers are used to.

Many open source tools are still catching up on AI ready features. That does not mean they will stay behind forever, it just means you may need a few extra tricks today. On the other side, hosted builders like Squarespace put real limits on what you can do. You do not get true root access to your hosting environment, which makes it difficult to serve knowledge JSON, LLM reference files, and other structured assets that assistants love.

So treat your GEO Crash Test results with this in mind. The score tells you what needs attention. How hard it is to fix will depend on the rails your site is running on.

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world. Cheers!

Leave a comment

If you still think “AI assistants” are some novelty plaything for tech bros and college kids, you’re already six months behind.

These things can now do work.
They can surface inventory, answer nuanced questions, book tours, start applications, capture deposits, and even pre-qualify industrial requests all without ever sending someone to your site.

Traffic still matters, but here’s the truth:
Conversion has left the building.

It’s now happening inside the chat.

We’re restructuring our marketing around that fact, because the next decade of Multifamily and Industrial CRE leasing will be won or lost inside conversational interfaces.

What Changed for CRE

Search used to hand off to websites.
Now assistants complete the journey.

Results are conversational, voice-first, and trust-weighted. The single answer that gets read aloud is the conversion opportunity.

Over time, paid placements will crowd organic mentions.
The winners will be the operators whose:

• Data is readable by assistants.

• Expertise is verifiable by external sources.

• Actions are executable inside the interface.

If your brand can’t be understood, cited, or acted on inside chat you’re effectively invisible.

We used to optimize for clicks.
Now we optimize for completions inside the interface.

The Long Game Principles

Here’s how I think about it as a marketer leading this shift in CRE:

Be patient with compounding.
Authority, reviews, and assistant trust accrete slowly, then suddenly. It’s SEO deja vu. I know, I’m impatient too.

Be persistent with shipping updates.
Ship something every week that improves the machine-human handshake. Progress > perfection.

Build for people and machines equally.
Clean data for assistants. Clear value for humans. Both audiences are judging you.

Treat assistants as a channel, not a gimmick.
This isn’t a sandbox experiment. It’s the new leasing desk and the digital deal room.

Phase 1: Make Your Inventory Machine-Readable (Days 0–90)

If an assistant can’t parse your data, it can’t sell your property. Period.

Start here:

1. Publish a JSON feed.
One for Multifamily, one for Industrial.

• Multifamily fields: community_id, address, neighborhood, unit_types, price_range, availability_date, pet_policy, parking, tour_options, application_url, photos, reviews_summary.

• Industrial fields: asset_id, address, square_feet, clear_height, power, dock_doors, drive_ins, zoning, incentives_notes, site_plan, photos.

2. Expose a simple OpenAPI.
Endpoints like:

• GET /communities?city=Milwaukee&beds=2&max_price=2200&pets=true

• GET /industrial?county=Racine&min_sf=80000&docks=6

Assistants love predictable patterns. Humans love results that make sense.

3. Connect tour scheduling directly.
Calendly, RentCafe, whatever. Just make it one click from chat.
No one wants to fill out a form in 2025.

4. Add assistant-friendly FAQs and schema markup.
For Multifamily: pricing, pet rules, parking, utilities, lease terms.
For Industrial: power specs, loading, labor radius, incentives, zoning.

Assistants quote this data verbatim. Feed them clarity.

5. Test your prompts weekly.
If you ask:

“Show two-bedroom options under $2,200 near the Medical College of Wisconsin with parking and pet fees. Book Friday at 3 pm.”

…and your property doesn’t show up, your data is broken.

Phase 2: Shape and Scale Demand (Days 90–180)

Once you’re visible, you need to be valuable.

1. Build topic hubs that answer entire journeys.
Assistants reward comprehensive clarity.
Create hubs that solve full user intents:

• Multifamily: parking and transit, pet policies, school districts, employer proximity, move-in checklists.

• Industrial: power and utilities, loading access, zoning guides, incentive calculators, labor and commute data, build-to-suit guides.

Each hub should include FAQs, a tool or calculator, and a next step the assistant can act on.

2. Publish verified expertise.
Named bylines from executives, asset managers, and regional leads. Consistent bios. Backlinks to canonical profiles.
Assistants verify authorship before they trust output, you can’t fake it anymore.

3. Prepare for paid assistant placements.
When the APIs open, test small.
Creative = concise offer + trust signal + next step.
Track blended CAC across assistant-paid and organic.
Expect CPMs to be friendly at first, then spike.

Phase 3: Measure the Invisible Wins (Month 6 and Beyond)

This is where marketers really add the value to the bottom line — measuring what nobody else can see yet.

Tours are booked. Deposits placed. RFIs submitted.
All without a website visit.

Here’s how to track it:

1. Capture Assistant-Initiated Actions.
Use server-side events or webhooks to log when assistants trigger bookings, deposits, or inquiries.
Tag them as “chat-native” conversions.

2. Separate Assistant vs. Classic Metrics.
Assistant conversions have different velocity and intent. Don’t bury them under “website goals.”

3. Measure Share of Answer.
Audit the top 20 queries that matter in each market:

“Two-bedroom near [employer]”
“80,000 SF warehouse with 28’ clear height and eight docks.”
Track how often your brand or property is recommended or cited.

4. Track Review Velocity and Language.
Assistants quote review phrasing directly.
Encourage specifics: service response, power reliability, parking convenience, dock flow, cleanliness.

5. Maintain Data Freshness.
Every property feed should carry a “last updated” date.
Assistants devalue stale data like Google does broken links.

6. Build a Weekly Chat Scorecard.
Report like a channel:

• % of tours or inquiries initiated by assistants

• Share of answer for top queries

• Assistant-to-site conversion rate

• Review sentiment and recency

• Data feed freshness

This is your new playbook. Most marketers won’t build it until it’s too late. I know i’m still working on mine. :)

Governance: Who Owns What

• Marketing owns data feeds, schema, and APIs.

• Property management owns pricing, calendars, and availability.

• Asset management owns industrial specs and incentives.

• Analytics owns assistant attribution and dashboards.

Everybody touches the data. Everybody protects the signal.

The Close: The Long Game Nobody’s Playing Yet

This isn’t a hack. It’s a new operating model.

We’ll win by being:

• Easy for assistants to trust.

• Simple for renters and site selectors to act on.

• Relentlessly consistent in how we show up every week.

Patience compounds results. Persistence compounds trust.

And when the chat interface becomes the primary customer journey — which it will — the Multifamily and Industrial brands who already optimized for it won’t just show up first.

They’ll close first.

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world. Cheers!

Leave a comment

The Race Has Evolved

In The Race to Stay Educated, I wrote that curiosity had become the new competitive advantage. The ability to stay informed — not just about your role, but about how technology is reshaping it — was the dividing line between staying relevant and quietly fading into irrelevance.

But the race has changed again.

Knowing about AI is no longer impressive. It’s assumed. The next advantage belongs to those who know how to apply it intentionally — not to automate mindlessly, but to think differently.

The truth is, most professionals are stuck in “AI demo mode.” They test tools, skim headlines, and feel like they’re keeping up. But the people quietly rewriting the rules of productivity aren’t just using AI — they’re building personal systems around it.

That’s the new race.
Not to learn faster — but to apply smarter.

We’re entering an age where “efficiency” no longer means doing more things — it means doing fewer things perfectly aligned with your value. AI is the bridge that makes that possible.

If you’ve felt overwhelmed by new tools, scattered workflows, or the anxiety of being “left behind,” you’re not alone. The solution isn’t chasing every new app or GPT — it’s designing a personal operating system that amplifies your focus, removes friction, and scales your creativity.

The goal of this essay is simple: to help you work like you have a team of ten, without burning out or losing your authenticity.

If The Race to Stay Educated was about preparing your mind,
this one is about preparing your workflow.

Below is the full guide — a practical roadmap for building your AI fluency, based on real frameworks, daily habits, and insights drawn from the Perplexity at Work methodology.

Read more

The multifamily marketing world is already beginning to feel the tremors of what I called in an earlier essay “the click disappearance moment.”

For years, the playbook has been predictable: optimize for SEO, compete on ILS platforms, tweak PPC bids, polish virtual tours, and hope your leads don’t ghost before signing. But that playbook is being rewritten, not by Google, but by AI search itself.

When renters no longer search by typing but ask an intelligent agent to find their next home, the standards that have guided multifamily marketing for decades start to disappear like a finger snap of Thanos. The question isn’t if this will happen, it’s how fast, and whether your brand will recognize the shift before your analytics tell you too late.

From Keywords to Conversations

As I wrote in When the Click Disappears, the funnel we’ve relied on for years isn’t breaking — it’s vanishing.

Renters no longer follow a predictable trail of impressions, clicks, and conversions. They simply ask.

“Find me a two-bedroom apartment with EV charging, under $4,500 (inflation joke), close to good schools, and where pets are welcome.”

The AI won’t serve ten blue links.
It’ll deliver one verified, data-rich, contextually perfect answer.

That single moment — the shift from browsing to asking — is the quiet collapse of the traditional marketing ecosystem.

When AI intermediates discovery, your competition is no longer the property across town.

It’s the language model deciding which listings to surface and which to forget.

Yesterday’s Playbook (and Why It’s Breaking)

For the last decade, multifamily marketing has been a discipline of rhythm and repetition:

• Depend on ILS networks.

• Tinker with SEO.

• Optimize PPC and re-marketing loops.

• Build static landing pages.

• Showcase aesthetics and amenities.

That rhythm has been reliable — almost mechanical. But AI doesn’t browse. It comprehends.

The AI agent doesn’t care if your site ranks #1 for “luxury apartments near me.”
It cares whether your data is credible, structured, and relevant.

If your listings aren’t readable by machines, you’re not invisible to renters — you’re invisible to the systems that interpret their intent.

The AI Compression Effect (and What I’m Seeing Already)

This isn’t theory. It’s already happening.

In my own analytics across multifamily portfolios, I’ve started noticing something strange — traffic patterns that defy traditional logic.

Organic impressions remain stable, but click behavior is splintering. Paid campaigns are delivering erratic results with fewer visible conversions, even when engagement quality feels consistent. Some weeks, performance metrics spike without reason; other times, they dip into unexplainable quiet.

It’s like watching your marketing system suddenly bleed from the nose mid-conversation while it continues to smile and talk to you, insisting it feels just fine.

I’ve been chasing down the causes, changing ad copy, testing CTA variations, analyzing user flows, but it’s becoming a game of cat and mouse. It’s becoming quite obvious though what’s occurring.

The truth is, these anomalies are early signals of AI-mediated discovery.
Users aren’t always arriving through traditional channels anymore. Some are getting partial answers from AI search, never reaching your site. Others are showing up through “invisible” referral paths that analytics platforms can’t yet classify.

We’re witnessing the start of the AI Compression Effect:

• Fewer clicks.

• Fewer trackable visits.

• More unseen decisions made upstream, before a human ever touches your site.

The funnel hasn’t broken — it’s simply moved somewhere you can’t see it.

And that should terrify anyone who still believes performance equals visibility.

What Becomes Obsolete

If you’re still counting impressions as success, it’s time to re-evaluate the scoreboard.

• SEO blogs like “Top 10 Reasons to Move to Dallas”… obsolete.

• PPC landing pages… redundant once discovery becomes conversational.

• Cookie retargeting… incompatible with AI-moderated, privacy-first ecosystems.

• Human leasing scripts… replaced or augmented by AI copilots fluent in your brand’s tone.

The new metric of success isn’t traffic.
It’s interpretability — can AI confidently understand, verify, and recommend your property?

If not, no amount of ad spend can make up for being misunderstood by machines.

The New Multifamily Stack

As I explored in The Next Evolution in Real Estate, the next phase of this shift merges AI fluency with Web3 transparency.

1. AI-Structured Data Layers: Listings, amenities, and reviews formatted for comprehension, not decoration.

2. Decentralized Resident Profiles: Verified renter preferences traveling via blockchain, not cookies.

3. Smart Contracts: Leasing, deposits, and renewals, automated and tamper-proof.

4. Tokenized Loyalty Programs: Resident engagement rewarded through digital value exchange.

5. AI Leasing Copilots: Digital counterparts who represent your brand and guide inquiries intelligently.

This is the next frontier of multifamily — not automation for convenience, but comprehension for credibility.

Preparing for the AI + Web3 Era

If you’re in marketing leadership, stop asking how to get more clicks. Start asking:

“Can AI understand what I stand for?”

Here’s where to start:

1. Audit your property data — structured, accurate, discoverable.

2. Shift KPIs from traffic to trust — measure how AI interprets your brand.

3. Rebuild for transparency — inconsistencies will be punished.

4. Train your brand’s tone — make it recognizable by both humans and algorithms.

5. Educate your teams — leasing fluency now includes prompt literacy and data reasoning.

Because when the interface disappears, your data becomes your reputation.

When the Click Disappears, Comprehension Becomes Currency

The multifamily revolution won’t announce itself. It’s already hidden inside your analytics reports — mislabeled as anomalies.

AI search is removing friction, rewriting behavior, and reshaping what performance even means.

The marketers who cling to old metrics will waste months diagnosing the wrong problem. The ones who embrace this shift and design for AI comprehension will own the next decade of leasing.

When the click disappears, marketing stops being about impressions.
It becomes about interpretation.

Your properties won’t just need to be found.
They’ll need to be understood.

Because the next vacancy isn’t in your building.
It’s in your strategy.

If this sparked something for you, share it with a colleague or drop a comment — I want this to be a dialogue, not a monologue. I’ll keep exploring how AI, marketing, and human perception collide, and I’d love to hear how you’re seeing it unfold in your world. Cheers!

Leave a comment