The problem disappeared

I’m a Tailwind Plus customer. Paid for it. Used it regularly.

Before Claude got good at writing UI code, I needed Catalyst — Tailwind Labs’ component kit — to get consistent results. I’d describe a component to an AI, and what came back was functional but messy. The pieces didn’t hang together. So I’d reach for the paid product. It filled the gap between what AI could generate and what I actually needed.

Then Claude got better. And I stopped reaching for Catalyst.

Not because I lowered my standards. Because what comes back now is actually good. Consistent. Cohesive. It knows the conventions. It’s seen thousands of examples of what “polished” looks like, and it pieces things together that feel right.

The product I paid for became unnecessary. Not because I couldn’t find it — because AI started doing what it did.

That’s the part everyone’s missing in the Tailwind Labs story.

Adam Wathan recently shared what’s happening with unusual candor: six months of runway, docs traffic down 40%, 75% of engineering laid off. The obvious narrative is that AI killed the traffic funnel. Developers used to search “Tailwind modal” and land on the docs. Now they ask Claude. Fewer eyeballs, less revenue.

That’s the symptom. Not the disease.

The disease is that the problem disappeared.

Tailwind Plus sold three things: time savings, design consistency, and reference implementations. All three are now available on demand for the cost of describing what you want. “Building polished UI is hard and time-consuming” — that was the problem worth paying to solve. It evaporated. Not because someone built a better solution. Because the problem itself dissolved for anyone with access to a decent model.

You can compete with a better solution. You can’t compete with a problem that no longer exists.

Meanwhile, shadcn/ui — same ecosystem, free, copy-paste — became the standard for AI-generated component code. The creator works at Vercel. The components don’t need to make money directly. They make the Next.js ecosystem more attractive. The value gets captured elsewhere in the stack.

Tailwind Labs tried to capture value at exactly the layer AI commoditized hardest. shadcn/ui never tried to capture it there at all. Same ecosystem. Radically different outcomes.

More traffic won’t fix a product that solves a problem nobody has anymore. The question isn’t how to get more eyeballs on the docs — it’s what new problem is worth solving next.

The traffic decline is a symptom. The product obsolescence is the disease. And the treatment isn’t better distribution — it’s a different product.