AI Is Changing What We Build

There was a time when every other week brought a new JavaScript framework, library, or build tool. React, then Next.js, then Vite, Bun, and Remix.

Each one came from the same place. Someone hit a painful problem and decided it shouldn't exist anymore.

That loop drove the whole ecosystem. Build something, hit a wall, understand it, build a better abstraction, share it. Repeat.

Then AI showed up

Now a lot of developer problems don't need a new tool. They need a prompt.

Broken config? Ask AI. Slow build? Ask AI. Weird bundler behavior or a terrible API? Ask AI.

The problem is still there. It just costs less to live with it.

That changes what we build. Before, when enough people felt the same pain, someone would say:

There has to be a better way.

And they'd build it. Now the reaction is more like:

I'll just ask Claude.

AI may be soaking up the pain that creates frameworks

Nobody builds a framework when everything works. You build one because something keeps hurting.

AI doesn't remove that complexity. It helps you get around it. And that can lower the drive to fix it for good.

Why spend six months on a new abstraction when an AI agent can help thousands of developers work around the old one?

So the real question is: are we solving problems, or just getting better at tolerating them?

Maybe innovation is moving to a new layer

Today's frameworks are already packed with engineering: compilers, bundlers, caching, streaming, server rendering, code splitting, and more. Big performance wins from yet another framework are getting harder to find.

If AI can profile a component, spot bottlenecks, and suggest better caching, the next big speedup might happen inside the stack we already have.

Not good or bad, just different

The old loop was Pain → Tool → Framework → New abstraction. The new loop is Pain → AI → Workaround.

Workarounds are fast. But abstractions create leverage. A workaround fixes one person's problem today. A good framework makes the problem disappear for millions of people for years.

So let's not use AI as an excuse to stop improving the foundations. Sometimes the right move isn't asking AI how to fix something. It's asking:

Why does this problem exist in the first place?

The next great developer tools might come from people who keep asking that question. Not despite AI, but because of it.