A few days ago, I learned that Tailwind laid off 75% of their workforce. Their framework is everywhere now, in every frontend project, every prototype, every AI-generated landing page. More popular than it's ever been. And yet.
I thought very hard about it.
There used to be a journey. You'd hear about a library from a colleague, maybe see it mentioned in a thread. You'd visit the documentation site, bookmark it, return to it dozens of times as you learned.
The creator's voice would seep into your understanding - their design philosophy, their careful examples, the way they anticipated your questions in the FAQ. You'd eventually hit the premium tier, or the support plan, or the hosted service, because by then you knew these people.
You trusted their taste. You'd already spent hours in their world.
That was the contract. Make something useful, document it beautifully, and the people you help will sustain you.
The Prompt That Breaks Everything
Let me be clear: I'm not against AI. I use it myself. For writing project documentation, for generating those repetitive CRUD endpoints, for stubbing out boilerplate — AI saves me hours. When something requires deep thinking, when I need to understand the architecture or solve a genuinely novel problem, I still reach for first principles. But for the mechanical stuff? Of course I use AI.
But here's the thing. Now I open Claude or ChatGPT and type: "give me a Tailwind navbar with dropdown and mobile responsive." Twenty seconds later, I have working code. I never visit tailwindcss.com. I don't know what premium features they offer. I don't even think about the humans who built this thing I'm using.
The AI trained on their documentation - their carefully written examples, their thoughtfully designed API, and now serves it back to me faster than their own site could. I get the value. They get nothing. Not a visit. Not a chance to convert me. Not even the dignity of my attention.
This is what winning looks like in this modern age of programming, apparently. Total adoption. Complete irrelevance.
What We're Losing
I remember when being a developer who created tools meant something specific. You'd spot an inefficiency, a pattern that kept repeating. You'd write a library with discipline, test it, document it meticulously. If you got it right then if you truly understood the problem and people would come.
They'd read your docs, star your repo, join your community. Some would contribute back. A few would become customers.
There was a relationship there. Indirect, maybe, but real.
The best library authors weren't just good programmers. They were good teachers. They understood that documentation was pedagogy, that examples were arguments for a way of thinking. Reading Stripe's API docs or React's philosophy pages, you weren't just learning syntax - you were learning taste, learning how experienced developers thought about problems.
That's what's vanishing. Not the code but the code is more accessible than ever. But the context. The philosophy. The human relationship between maker and user.
The Cruel Mathematics
Here's what makes this hurt: Tailwind's success is real. Millions of developers use it daily. It's shaped how we write CSS. It's shaped how I, ME, write CSS. The framework works beautifully, and the AI that recommends it has been trained on documentation that took years of care to write.
But the people who wrote that documentation can't pay their mortgages with adoption metrics. They can't feed their families with GitHub stars. The new economics are brutal and simple: if users never visit your site, you don't have a business. Just a widely-used artifact of your labor.
I wonder how many other tools are in this position. How many maintainers are watching their download numbers soar while their revenue flatlines. How many documentation sites have become elaborate training data that benefits everyone except the people who created it.
What Gets Built Next?
This is where my mind goes dark. If the lesson of Tailwind is that perfect documentation and wide adoption don't matter — that the economic value flows to whoever controls the interface where developers ask questions — then what gets built next?
Why would you write comprehensive docs if they'll just be scraped and served by AI? Why would you create free tools if you can't convert users you never see? Why would you share your best ideas publicly if they'll be instantly commodified?
Maybe we'll see more closed ecosystems. More paywalled content.
More tools that only work if you're logged in, tracking every interaction. Maybe the next generation of developer tools will be hostile to AI training by design, encrypted and obfuscated and deliberately hard to learn from.
Or maybe — and this is worse, we just won't see as many tools at all. Maybe the best developers will keep their innovations inside their companies, where at least they're paid for them. Maybe we'll look back on the era of open, beautifully documented libraries as a brief historical moment, something that worked until it didn't.
My Personal Journey
Three or four years ago, at my old company, we were migrating away from AWS Serverless. Our data lived in DynamoDB, and as we moved to NestJS, everything had to be brought into Postgres.
The process was manual, slow, and painful. So I wrote a small internal tool, something that could convert and map DynamoDB records, infer relationships, and reshape the data into something PostgreSQL could actually understand. It wasn’t elegant, but it saved us countless hours.
That experience stayed with me. One night, after work, I went home and started building Piper - a data migration tool designed to bridge non-relational and relational worlds. The idea was simple but ambitious: clear interfaces, live data editing, and the ability to map relationships visually before a single record is migrated.
Overtime, I start to see the rise of AI and I look back at my small project and I wanted to give it a hug and wondered: " This tool of mine is built with passion, there are already many like mine but I don't care, The only thing I care about is, If I released this, It'd just be another AI scraped piece of shit. It will solve a problem or 2 but then again, no one would care about the way I think, the beauty and passion I put into this. "
A Small Hope
I don't have a solution. I'm not even sure there is one. But I think we should at least notice what's happening. Acknowledge what we're losing. Feel something about it.
When you use AI to generate Tailwind code, you're not just getting utility. You're participating in a system that's gradually making it impossible for people like Adam Wathan to sustain the work that made your prompt possible in the first place. That's not your fault. the economics are structural, not personal - something it's worth sitting with.
Maybe that awareness leads somewhere. Maybe some of us will choose to visit the docs anyway, to buy the premium tier we don't strictly need, to support the tools that support us. Maybe companies will figure out new models that work in this reality.
Or maybe this is just what it feels like to watch a profession transform in real time. To see the old contracts dissolve, the old relationships break, while something new, efficient, but colder, colorless - takes their place.
I miss the journey. I miss knowing the people behind the tools I use. I miss the world where building something useful was enough.
But I'm still here, typing prompts into an AI, getting instant answers from documentation I'll never read. We all are.
That's the melancholy part.
Top comments (4)
I appreciate how this article really looks at the two sides of AI honestly in the real world of software development. Conversations like this matter more than they might seem, because they show how the profession notices itself changing instead of just adapting silently. Even naming the loss is a form of agency.
written in one take, raw, and intentionally unrevised. Apologies for any rough edges. Thanks for reading <3
Not me calling infinite repeated random keyboard smash like this “beautiful.”
But ok, they did the thing. 1M apes. 1M years. Random typing. And somehow it manifested, ahead of time:
Then the spell wore off. Humans are back to Netflix and mortgages. Apes are back in the forest peeling bananas. Nature is healing.
Next gen AI will have to find the next trick: lose Tailwind and fix the web. Or just lose the web, because at this point, why do we need such a broken HTML if we have chat?
You're not wrong about the div soup. I've generated enough
<div class="flex items-center justify-between px-4 md:px-6 lg:px-8...">to know exactly what you mean. It does look like monkeys got loose in the codebase.But here's the thing: those monkeys were us. We built Tailwind because we were tired of naming things. We built Perl because we needed to parse text files at 3am. The chaos has a history, even if it looks random now.
Your "post-web" future might be coming - maybe we do just chat our way to interfaces and skip the markup entirely. But then what? The AI still needs to be trained on something. Someone still has to solve the actual problems, write the actual solutions, document the actual patterns.
And if the new model is "humans solve problems → AI extracts solutions → humans never visit the source" then we're not healing, we're just finding new ways to make the people who build things invisible.
The web might be broken. HTML might be obsolete. But at least when I wrote disgusting nested divs, I knew I was the one who wrote them. I owned the mess.
What's the alternative? AI generates interfaces we don't understand, from patterns we didn't learn, solving problems we never struggled with? That's not healing either. That's just... amnesia with better UX.
But yeah, the class names are pretty ridiculous.