My feed is full of people showing off what they built with AI. The prompt that wrote their landing page. The app they shipped in a weekend. The workflow that runs their entire content calendar.
It reminds me of an intern coming back from their first week at a web dev agency, thrilled that they learned how to spin up a WordPress site and clear the cache. And that’s not a shot at interns. Interns are supposed to be excited about the basics. They’re learning.
The difference is interns know they’re at the bottom of the learning curve. These folks think they’re at the top.
It’s noise. And the reason it’s noise gets at something most people using AI haven’t figured out yet: AI is a tool, not a solution.
The most interesting part, though, was the comment section. People are sick of AI slop. They’re calling it out. They don’t trust it.
Fifteen minutes of scrolling turned up over a dozen of these ads. AI-generated everything. The biggest claims possible with the least effort behind them.
People are catching on, and they don’t want it. Nobody cares that you prompted an AI into building something. They want to hire someone who can actually solve their problem, or at least move them toward it with confidence. I get the urge to move fast and let AI assist. But when it does the thinking, the executing, and everything in between, you end up with a failed campaign and a product nobody trusts.
Every one of those agencies made the same mistake. They treated the tool like it was the solution.
You don’t win the game of business by outsourcing the thinking. You win by being strategic and thoughtful about how you approach problems and how you solve them. AI can make you faster. It can’t make you right. It’s a tool, not a solution, and the solving is still on you.
And the bar needs to be a hell of a lot higher than “look what I prompted.”
It’s not what you do, it’s how you do it
“Build a website” isn’t a decision. It’s a wish. The real decisions live underneath it. What’s the site built on? What does it need to say? Who are your customers? How do you move them from curious to convinced? Take the platform question alone. WordPress on PHP gives you thousands of plugins, deep customization, and a massive ecosystem that’s been battle tested for twenty years. Or you can have AI generate a clean HTML page that loads fast and looks fine on day one. One gives you room to grow. The other boxes you in, in ways you won’t notice until you need something it can’t do. Neither answer is wrong. What’s wrong is not knowing you made the choice.Every default is a decision someone else made for you
Here’s what actually happens when someone without depth in a field hands a complex project to AI. Their prompt is vague, because they don’t know what they don’t know. And AI doesn’t stop to say “we need to talk about your tech stack before we go any further.” It fills every gap with an assumption and keeps moving. Hosting, architecture, site structure, messaging, tone. Decision after decision, made silently, by a model that will never have to live with the consequences. You didn’t skip those decisions. You let something else make them without telling you. And this runs all the way down to the code itself. Ask any developer: there are usually a few ways to solve the same problem, and every one of them comes with trade-offs. Fast now against flexible later. Simple against maintainable. AI makes those calls for you too, and it doesn’t always make them the way a person who has to live with the codebase would. A working developer put this better than I could. He asked one of the top models to carry some functionality from one page over to another. Any human would have made the components reusable and moved on. The model duplicated a few thousand lines of code instead. It worked, technically. It was also the kind of mess you only catch if you understand what you’re looking at. His breakdown is worth a few minutes of your time: People with real depth see those decision points coming a mile away. They know which ones are reversible and which ones will haunt the project two years from now. That’s what depth buys you. Not better answers. Better questions.The part where the bill shows up
Which brings me to the second thing nobody gloating about their weekend build wants to hear: it probably doesn’t scale. Ask AI to build you a website and there’s a good chance it spins the project up on Vercel. Great platform, developers love it, free to start. Perfect, right? Look closer. The free tier prohibits commercial use, so the moment you’re running a real business, you’re on the Pro plan. That’s $20 a seat on paper, but most Pro users actually land somewhere around $67 a month once usage charges kick in. And here’s the part nobody reads: the bill doesn’t scale with your team. It scales with your traffic. A growing product around 250,000 monthly visitors blows past the included limits, and the overage on requests alone can cost more than the seats do. Push toward a million and a half visitors and bandwidth plus request overages run north of $2,600 a month. Plenty of developers have watched a $20 plan quietly turn into $700 or more. One woke up to a $23,000 bill after a bot attack, because hostile traffic gets billed at the same rate as paying customers. And leaving isn’t free either. Migrating off a platform like that typically burns thousands of dollars in dev time. So is Vercel a bad choice? No. For plenty of projects it’s the right one. The point is that it was a choice, with real trade-offs, and you never made it. You could self-host instead and dodge the overages, but now you own the security patches, the backups, the updates, all the boring stuff that keeps a business online at 2am. There’s no free lunch here. There are only trade-offs. Someone with depth weighs them. Someone without depth never sees them.AI is a tool, not a solution
I was listening to a freelancer talk about this exact idea recently. A client tells him they don’t need to hire anyone anymore. They’ll just use AI on their own time and build the website themselves. Instead of arguing, he offers a challenge: you use AI, I’ll use AI, and we’ll both build your website. Whose is going to be better? And why? We both know how that plays out. Put me in that seat and mine comes out ten times better, and not because I have some secret prompt. It’s because I’ve spent fifteen years learning what a website actually has to do. I know what questions to ask before anything gets built. I know the requirements the client never thinks to mention. I know what good looks like, what to keep, and what to throw out. A novice doesn’t know any of that, so the same tool just amplifies their guesswork. Because that’s all a tool does. It amplifies the person holding it. A nail gun doesn’t make you a carpenter. It makes a carpenter faster. And here’s what makes the gap impossible to fake: this work isn’t binary. Building a website isn’t a math equation with one right answer in the back of the book. Creative work, whether it’s web design, graphic design, or content, is art and science. The science you can systematize. Performance, structure, tracking, the technical bones. The art is judgment. What to say, what to leave out, what feels right for this audience and dead wrong for another. AI can crank through the science all day long. The art only comes from experience, and it’s exactly what’s missing from every flat, samey, AI-built site you’ve ever clicked away from.Treat AI like the intern it is
So how should you approach complex projects with AI? The same way you’d work with a smart intern. You would never let an intern pick your tech stack. You wouldn’t let them define your positioning or decide where the budget goes. But once you’ve picked the direction and mapped how to get there, you’d absolutely let them build the page. And they’d probably do a solid job. That’s the model. You do the thinking on the decisions that carry long-term effects. At minimum, make AI walk you through the pluses and minuses before you commit to anything big. Then hand off the execution. Flip it, let AI run both the thinking and the doing, and you already know what you get, because it’s everywhere: slop. Sounds flat, looks flat, feels flat, and doesn’t do the job it was built to do.The bar needs to be higher
The proof is already piling up. The wave of AI-first entrepreneurs has mostly shipped apps that look and feel like slop, built because someone wanted to build an app, not because someone needed a problem solved. That’s the tell.Moneyball, The Movie
If you want a picture of the opposite, watch Moneyball. The A’s couldn’t touch the Yankees’ payroll, so Billy Beane and his numbers guy stopped playing the game the way everyone else played it. Instead of chasing the biggest names, they used data to find undervalued players who got on base, because getting on base is what actually wins games. Then they executed on that plan, and a team with a fraction of the budget went toe to toe with the richest clubs in baseball. Here’s the part people miss: every team had access to the same stats. The A’s didn’t win because they had better tools. They won because they thought more strategically about how to win the game they were playing. That’s the whole blueprint for this post. The A’s got strategic. They thought through their options, weighed the trade-offs, committed to a direction, and executed with confidence. No shiny objects, no wishful thinking. They stayed rooted in reality and moved forward with the information they had. The data informed every decision, but every decision was theirs. Nothing got decided for them by default.Consumers Are Sick of AI Slop
I was scrolling Instagram the other day and noticed an onslaught of AI slop ads. Web design agency after web design agency, and from the outside it looked like a race to the bottom. Cheaper prices, faster turnaround, anything to perk up a client’s ears. Reading through the ads and landing pages, it was obvious the copy was written by AI. Nothing about it made me confident the team behind the ad knew how to approach a web design project. It felt like a scam. Big promises, seemingly no experience in the field. Here’s the ad:
The most interesting part, though, was the comment section. People are sick of AI slop. They’re calling it out. They don’t trust it.
Fifteen minutes of scrolling turned up over a dozen of these ads. AI-generated everything. The biggest claims possible with the least effort behind them.
People are catching on, and they don’t want it. Nobody cares that you prompted an AI into building something. They want to hire someone who can actually solve their problem, or at least move them toward it with confidence. I get the urge to move fast and let AI assist. But when it does the thinking, the executing, and everything in between, you end up with a failed campaign and a product nobody trusts.
Every one of those agencies made the same mistake. They treated the tool like it was the solution.
You don’t win the game of business by outsourcing the thinking. You win by being strategic and thoughtful about how you approach problems and how you solve them. AI can make you faster. It can’t make you right. It’s a tool, not a solution, and the solving is still on you.
And the bar needs to be a hell of a lot higher than “look what I prompted.” 