What AI Hasn't Changed: A Review of the Boring Basics from an Old Guy

It’s been more than ten years since I wrote a marketing blog post.

Since then I co-founded two companies and ran both of them, got cancer, got better, and went quiet for a long while. I came back to writing this summer, and the first thing I wanted to do wasn’t to have a take. It was to take inventory.

Everyone I follow has spent the last year telling me that everything changed.

And plenty did. I can build in a weekend what used to take a funded team a quarter. That part is real, and I’m not going to be the guy who pretends it isn’t.

But here’s the sentence I keep coming back to: AI changed the cost of building. It didn’t change the cost of being wrong.

If anything, it made being wrong more dangerous.

When building was expensive, the expense did you a favor. It made you wait. You had to court an idea for months before you could afford to marry it, and somewhere in those months you’d sober up and notice that the market wasn’t asking for the thing you were building.

Now you can get married on the first date. By Sunday it has a name, a logo, a website, a Stripe account, and six integrations.

And then the real problem starts, which is that you’ve made so much progress that stopping feels like failure.

That’s the trap. Cheap building didn’t remove the sunk cost. It moved it earlier and made it emotional instead of financial. You’re not defending your money anymore. You’re defending your weekend, your excitement, and the version of yourself who told three people about it on Friday.

Moving fast is good. I’m not arguing for slow. But moving fast gives you fewer natural pauses to ask whether you’re moving in the right direction, and those pauses used to be free. You got them whether you wanted them or not.

There’s a word for what happens next. Poker players have been using it for decades.

tilt

noun. From poker.

1. The state of playing badly because of what you’ve already lost. A tilted player isn’t reading the table anymore. He’s trying to get even.

2. In business: the state of being unable to see your own idea clearly, because you’ve already paid for it in hours, money, and pride. Tilt doesn’t feel like tilt. It feels like conviction.

Used in a sentence: “He’d spent nine weekends on it, so when the market said no, he heard not yet. He was tilted.”

Every founder I know has been tilted. Most of them were tilted for months and thought they were being resilient.

So while everyone else catalogs what changed, I want to do the opposite, which is a habit I stole from Jeff Bezos. At AWS re:Invent in 2012 he put it this way:

“I very frequently get the question: ‘What’s gonna change in the next 10 years?’ And that is an interesting question. It’s a very common one. I almost never get the question: ‘What’s not going to change in the next 10 years?’ And I submit to you that that second question is actually the more important of the two because you can build a business strategy around the things that are stable in time.”

Jeff Bezos, AWS re:Invent, 2012

He’d been saying it since a Harvard Business Review interview in 2007, and Amazon’s 2008 shareholder letter names the durable things outright: customers value “low prices, vast selection, and fast, convenient delivery,” and “it is difficult for us to imagine that ten years from now, customers will want higher prices, less selection, or slower delivery.”

That’s the whole trick. You don’t build strategy on the thing that just changed. You build it on the thing that will still be true when the current excitement is embarrassing.

So let’s start with the one that stung a little to write . . .

Engineering Was the Excuse, Not the Obstacle

Vibe coding got so easy I expected an explosion of new startups.

It didn’t happen. A few vibe-coded apps here and there. Not much traction anywhere.

I know a thousand people who told me they’d start a business if they just had the engineering talent. Every local startup meetup is full of them. Ideas guys. Business guys. All looking for a technical co-founder to go build the thing for them. They’d all be millionaires, if only somebody would build their idea.

They have it now. It’s called Claude Code, or Codex, and it’s sitting right there on their desktop.

And yet most of them still don’t have their first paying customer.

Because in order to get any traction at all, you also have to care about a problem deeply enough to learn it. Put your reputation on the line by actually shipping. Support the thing after it’s live. Do the marketing. Build the list. Build the site.

You have to stand in front of your market and make offers.

All the unsexy stuff everyone assumed didn’t matter.

When I posted a version of this, Gideon Shalwick left a comment that I haven’t stopped thinking about. Gideon and I have known each other for well over a decade, back to when we were both figuring out video and audience building the hard way:

“Nailed it. Your advice to me from more than a decade ago still applies. Millions of details to get right, and you got to get them all right.”

Gideon Shalwick, in a LinkedIn comment

That’s the tell, and it’s the reason I’m writing this at all. The advice didn’t expire. The details didn’t get fewer. The tooling got better at exactly one of them.

Easy to build was never the same thing as worth building. AI tore down the lowest barrier to entry, and what it exposed was all the other work people were avoiding.

Now we get to see who the real players are.

Here’s what’s still standing.

The Market Is Still the Only Thing That Knows

My clients and I have run hundreds of marketing campaigns. There has never been one time, not once, when our assumptions about what the market wanted survived actual contact.

Hundreds of campaigns. Zero surviving assumptions.

And the part that should scare you: this held for the veterans too. People who’d been in their markets for decades, who knew every competitor and every price point and every industry in-joke. No amount of expertise ever accommodated what a few weeks of real interaction revealed.

Jared Krause put it better than I did.

“I always tell founders that when their great ideas encounter reality, it’s like encountering a meat grinder. Something comes out, but it’s not what they put in.”

Jared Krause, in a comment on one of my LinkedIn posts

Here’s the thing about asking, though: people don’t know what they want built. They know exactly two things. Their problems, and what already exists. So feedback shows up in two forms, always. “Can you build what X built, but better,” or an outcome they want. More money. Six pack abs.

The market gives you pain. It does not give you a product specification.

Nobody hands you a spec on a silver platter. They hand you their pain, and turning it into product is your job. Majid Jalilvand said it in one line on LinkedIn: “The feature is our job to figure out. The pain belongs to the customer.”

And when the feedback is genuinely hard to read, when some of it is sad and some of it is impossible and the need runs deeper than you expected . . . that’s not a bad sign. That’s the clearest sign you’ve found a hungry market. (I once had to grab a beer and sink into a chair before I could get through a survey’s worth of it.)

Nothing Warps Your Brain Like Building the Thing First

This is the one that’s actually new, or at least newly dangerous.

Once you start building a product, you get tilted. A set of incentives forms that blinds you to what the market is telling you. You acquire a reason to confirm the idea, because by then you’re in love with the work.

It’s not a character flaw. It’s physics.

Someone pushed back on me about this, and it was the best objection I got all year. Simple Memo, who builds an iOS app solo, argued that shipping is the reconnaissance:

“With no army and no research budget, I cannot build an unfair advantage against a problem I have only imagined; the terrain only shows itself once real users are on it. Sun Tzu still knew the ground before he won. So maybe the split is not ship-fast vs build-first, it is whether each ship is a scouting run with a hypothesis or just a coin flip.”

Simple Memo, in a comment on LinkedIn

The scouting frame is right. But scouting should happen before the build.

Here’s my worry with shipping as recon. By the time you’ve built the thing, you’re already tilted. You’re incentivized to confirm the hypothesis. You might be in love with the building itself. Almost everyone who has built and launched carries biases that warp what they’re able to see in the response.

Intel should come from your market. Surveys. Replies. Every email answered. Watching which questions keep coming back. Preselling before you write a line of code. The market tells you what it wants when you have nothing invested in a particular answer.

That’s the recon run, and it happens while your thinking is still clean.

Preserve clear thinking for as long as possible.

Someone once asked me what my unfair advantage was when I started. I said years of trying to be a professional blogger, surveying the audience, answering every email. True, but not the real answer. The real answer is that I built the audience first and let the product come as a response to the market, instead of jamming a square peg into a round hole like most product-first founders do.

The advantage wasn’t just the audience. It was never having been tilted in the first place.

I never assumed that software was the answer. Or a service, or a course, or anything in particular. I started with a clean slate, and you only get the clean slate once, before you start building.

After that, the emotional investment entrenches. You develop blinders. You start looking for product-market fit by asking Claude to build more, instead of working on the offer, the pain, the guarantees they’re actually looking for, the language they use, and the frameworks that explain their problem back to them better than they can explain it themselves.

That’s the work. Building more is just the thing you do to avoid it.

Order of Operations Still Beats Horsepower

So if you can’t build your way to clarity, here’s the sequence that still works.

Money, then offer, then sales message, then code.

Before you write one line of code, write your damn sales letter first. Four steps, and any one of them can save you six months:

• Write down how much money you want to make per year. One number. Not a range, not a vibe.
• Build your pricing page. Tiers, prices, what’s included. Will people really pay this? Does it add up to your number? No? You just saved six months.
• Write your sales page and record your sales video. If you can’t sell it in a script, the product won’t save you.
• Outline five comparison pages. Yours versus Salesforce. Yours versus Zoho. Do you have a competitive, or ideally unfair, advantage? If not, you should stop.

Still inspired after all that? You probably have a real chance. Go build.

And let the sales letter be the spec.

If the economics don’t work on paper, or the thing can’t be made compelling in a script, another integration is not going to rescue it.

All of which is prologue to the one step almost everybody skips, and it’s the only step that actually settles anything . . .

You Are Not in Business Until You’ve Made an Offer

Right now there are thousands of people on revision forty of an app nobody has been asked to buy. New feature. New refactor. New landing page. Another model release, another rebuild.

Prajwal put the whole phenomenon in one post, and about 295,000 people saw it, which tells you how many of us recognized somebody:

Vibe coding is crazy, man.

Met a 17-year-old with 5+ apps already.

Ignores everyone just to work on them.

Pays for both Claude and Codex.

Sleeps thinking about features. Wakes up fixing bugs.

Ships updates nobody asked for. Refreshes analytics every 10 minutes.

All that . . . just to have 6 users.

2 of them are his own accounts.

Prajwal (@0xPrajwal_) on X, August 2026

Every part of that is effort. Real effort, the kind most people never put in. Ships updates nobody asked for. Refreshes analytics every ten minutes. That kid is working harder than most founders I know, and he’s aimed at nothing.

That’s not a business. That’s a hobby with better tooling.

You are not in business until you’ve stood in front of your market and said: I made something for you. I believe it can help. Here it is. Buy it from me.

Without a live offer, you have no idea how the market responds, so every hour of polish is spent guessing. The offer is the instrument that measures everything else. Ship it and every revision after that is informed. Withhold it and revision forty is as blind as revision one.

While we’re here: an offer is not a product with a price tag on it. A product is the thing. An offer is the main benefit, the sub-benefits, the features, the pricing, the packaging, the bonuses, the expansion path, the guarantee, your positioning in the market, how directly you name their pain, and how well you understand what keeps them up at night.

Weak products with strong offers outsell strong products with weak offers. In SaaS the gap is even bigger, because your prospect can’t tell good software from a good demo.

This is also why I’ve stopped using the word monetize. You don’t monetize a business, because it’s already a business. Sam Walton never asked how to monetize his store. Restaurants don’t ask how to monetize food. The only things that ever get monetized are things that were never meant to make money in the first place, and it shows. It’s a diesel engine jerry-rigged to run on gasoline.

In 2009 monetization looked like a mediocre ebook bolted onto a blog. In 2026 it looks like a Stripe link bolted onto revision forty of a vibe-coded side project.

Same move. Better tooling.

Declan Dunn had the kindest version of this in a LinkedIn comment: “Most of those were better hobbies than to monetize into a business.” Yeah. Sometimes hobbies should just stay hobbies. That thing you’ve been building because you love it deserves one of two futures. It deserves to become a real business, with the offer and the price and the customer designed in from day one. Or it deserves to be left alone. Pure. Unadulterated. Unmonetized.

So make offers, not revisions.

Build Like Somebody Asked For It

Now let me draw a line, because “make offers” gets misread as “ship whatever you have as fast as you can.” That’s not it either.

Here’s the failure mode I keep seeing. Somebody builds first, and then they’re never done building, because every time they open the app they have another idea. They’re vibe coding on the train. They’re vibe coding between meetings. And what they’re building is whatever happened to pop into their head on the way there.

Or a competitor ships something, writes a post about it, and now it’s on the roadmap. Me too. I’ll take one of those.

That’s not product development. That’s improv.

Be more deliberate than that. Build in response to direct asks. Specific requests, from specific customers, about a specific problem you’ve heard more than once.

If you can’t name the person who asked for it, you’re building from your own reflection again. And we just covered where that road goes.

Distribution Is Still the Whole Ballgame

Quick old man moment before I use this word forty times.

When did everyone start saying “distribution” and “GTM” instead of plain old “marketing” and “sales”? Did the YC crowd decide those words were beneath them? Marketing has been marketing since the 1900s. Sales has been sales since we’ve had things to sell. Two words, everybody knows what they mean. Now every founder has a GTM motion and a distribution strategy.

Anyway. Distribution.

Here’s the cheapest test I know, and it costs nothing but your pride. If you don’t have the marketing chops to build a devoted audience or newsletter or following from scratch, then you for sure don’t have the marketing chops to sell a paid product or service.

Free is the easiest thing in the world to sell. If you can’t move free, paid is not going to rescue you.

Which is why “my product isn’t perfect enough” is a horrible reason to stall your launch, and “I don’t have a minimum viable following” is a great reason to delay it. One of those is fear wearing a lab coat. The other is a real gap.

But I want to go further than the usual advice here, because “do your marketing” makes it sound like a thing you bolt on afterward.

You should be building the product around the distribution channel.

At Leadpages, that’s exactly what happened, and it’s most of why it worked.

Here’s the real story. I noticed that nothing I published performed like a free landing page template. Nothing came close. Release a new template, people opt in to get a copy. That was the engine, and I could see it working before there was a product.

So the engine came first and the product got designed around it. Templates to build the audience, and affiliates to carry it, because I knew which affiliates I wanted promoting the thing before there was a thing to promote.

Then the accident that made the company. To actually use one of those free templates, you had to be a coder. That was the gap sitting in plain sight in my own funnel. Leadpages existed to close it.

I’d love to tell you I engineered that last part. I didn’t, and it would be revisionist to claim otherwise. But I did build a product around a distribution channel I had already proven, which is why the launch had somewhere to land instead of somewhere to hope.

So before you build, answer these:

• What is your growth engine? Not a guess. An actual mechanism you’ve seen move.
• Do your features map onto it directly, or are they decoration?
• If it’s influencers, which ones, by name? Would they enthusiastically use this?
• Can you build those relationships now, while you still don’t need anything from them?

A warning about the activity that feels like distribution but isn’t: building in public is very often just lazy marketing. Instead of being laser focused on your actual market, you’re doing what’s easy, which is telling your story instead of telling the story of your product to people with demonstrated need. It works if your market is indie hackers. Indie hackers are notoriously cheap, and frankly, a small market. Your customers don’t care about your MRR screenshots. Other founders do, and other founders aren’t buying.

AI can make a product. It cannot make the market care.

The Victorious Strategist Still Wins First and Fights Second

Sun Tzu wrote this 2,500 years ago:

“In war, the victorious strategist only seeks battle after the victory has been won.”

Sun Tzu, The Art of War

That’s the Leadpages story in one line, and it’s the whole argument for doing distribution before product.

For almost everyone, the majority of the work to make a product succeed happens after launch. Grind, iterate, pray like hell. The smarter play inverts it. Spend whatever time it takes building a massive unfair advantage first, then launch as the exploitation of that advantage.

This is why I don’t understand these indie hackers who launch a new product every week. Launching new products weekly is just announcing that you have no unfair advantage and that your startup is a lottery ticket. In my years of doing this, 98% of the people who crush it are exploiting a huge unfair advantage. Whether they tell you or not.

What counts as one? A distribution channel you own. Deep industry experience in obscure markets with esoteric regulatory moats, where B2B spend is high but only a handful of people really know what’s happening behind the scenes.

It doesn’t have to be enormous, by the way. Mine wasn’t. It only has to be enough to give you a reason to win.

Josh Braaten asked the question on LinkedIn that makes this urgent: “Why couldn’t someone better at vibe coding just come along and offer something better?”

They could. That’s the point.

And here’s the answer that’s emerging: when features become cheap, trust becomes expensive.

David Cancel said “product-based differentiation is dead” five years ago, and as a product person I hated it. Vibe coding proved him right.

Watch what you do now when you find a new app. Once you know what it does, you run a checklist. Is this vibe coded? Is this one person? Is this a quick test from someone who’ll be gone in six months? Am I better off just building this myself?

None of those questions are about features. They’re all about trust. Who made this? Have they delivered before? Will they support it? Do serious people use it? Has it been around long enough that the Lindy effect says it’ll still be here next year?

Signing up for a product because it has the features you want is almost entirely over. So invest in brand, which is the least fashionable advice in tech and one of the only kinds that compounds.

Copywriting Got More Valuable, Not Less

Which brings me to the skill I’d bet everything on right now.

Learn the basics of marketing. Learn the greats. Dan Kennedy, Eugene Schwartz, Gary Halbert, David Ogilvy. Read the sales letters that ran as controls for years, the ones that got beaten only by a better version of themselves. That material is sitting right there, cheap, and it’s more relevant than it was five years ago, not less.

Because here’s what AI actually changed, and it’s one distinction: editing versus writing.

Staring at a blank screen and creating something from nothing is brutally hard for most people. But show someone something that already exists and suddenly they’re brilliant. They’ll tell you exactly what’s wrong with it, what they don’t like, what they’d improve.

Humans are horrible at zero to one. We’re phenomenal at one to two.

I watched this for years running companies. Everyone wanted to give input on what had already been created. Almost nobody wanted to create from scratch. It’s just not fun.

So AI collapses the hard part, and your brain gets to skip from blank-page mode straight to editing mode. The mode you dread to the mode you’re actually good at. That applies to software, offers, pricing pages, sales letters, and articles like this one.

AI can get you infinitely close to what you need. It cannot quite get you there.

Because what’s left is the core concept. The driving idea. The headline. The thing that makes a stranger stop. AI will hand you a version one of anything in ninety seconds, and version one is competent and forgettable, and so is everyone else’s. Getting from there to something that actually beats the other guy is the entire job, and it’s a judgment call about human psychology that no model is making for you.

The enduring skill was never producing sentences. It’s recognizing the big idea, hearing the private language of a market, developing taste, and knowing what to cut.

And I know exactly how the next part sounds. Old man on the porch, shaking his fist at the mountain.

But it really does seem like the younger generation doesn’t know what it’s like to test twelve headlines against each other. Or to spend eight hours on a single paragraph because the hook on a sales page matters that much. That kind of craft at the word level, the willingness to agonize over one sentence, seems to have gone missing.

Which is strange, because everything became direct response. Every post is a hook. Every video opens with a pattern interrupt. Every landing page has social proof and a guarantee. The whole internet is running plays out of the direct marketing playbook.

Everything became direct sales marketing, and at the same time the art form got lost.

Ami Burstin said the quiet part in a LinkedIn comment: “I guess the real obstacle was always knowing what to build. And how to say it.”

Deep understanding of a market’s psychology, combined with the ability to package a thing so it’s inherently compelling. That’s the last skill to fall, and it isn’t falling this year.

Which is why I think a very specific kind of person is about to have a very good decade . . .

This Is Going to Be the Info Marketer’s Decade

Here’s what I actually came back to say.

When I came up, direct sales marketers went into information products. Ebooks, courses, memberships, seminars. Not because we loved the format, but because it was the only thing you could build without engineers. You could have an idea on Monday and sell it on Friday, and nobody had to raise a VC round.

Meanwhile, every marketing-driven CEO I knew wanted to be in software. That’s where the valuations were. That’s where the leverage was. That’s where a good month compounded into a good decade instead of evaporating when the launch ended.

The wall was never talent. It was operational. To build software you had to hire engineers, manage engineers, retain engineers, and become a company with all the unglamorous machinery that implies.

So the best marketers of my generation mostly stayed put. Guys like Frank Kern, Eben Pagan, Jeff Walker were phenomenal at this craft. Genuinely great. But they were lifestyle-oriented, and they didn’t have the operational appetite to go run a software company, so they didn’t.

That wall is gone.

For the first time, a pure-play marketer who crushes at offers and messaging can build something genuinely scalable. Not an ebook. Not a course. An actual product with actual retention.

So here’s my prediction, and I’ll put my name on it.

The indie hacker is going to get crushed by the direct response marketer who learned to ship software.

Not because the marketer is smarter. Because most people cannot compete with that person on messaging, on hooks, on offers, on positioning, on the sheer force of a personality that’s spent twenty years learning how to make strangers care. Those skills took decades to build and AI does not hand them to you in a tab.

The engineering advantage evaporated. The marketing advantage didn’t.

(To be clear, I’m not predicting Frank Kern is about to launch a SaaS company. I’m saying people cut from that cloth are going to eat, and some of them don’t know it yet.)

If that’s you, and you’re willing to add just a little operational discipline to the practice, a little COO energy to go with the showmanship, this is your decade. Everything you spent years learning got more valuable the moment everything else got free.

For twenty years the marketers wanted what the engineers had.

Now they have it.

The tools got free. The judgment never did.

Talk soon,

Clay

P.S. Most weeks I work these ideas out in public before they end up here, and the best lines in this piece came from people who hung out with me in the comments. If you’ve got a specific thought, reply to one of my posts on X @claycollins. If it’s good, I’ll quote you by name in the next one and link back to you.