I think, just from a purely build-step point of view, it's been evident that tools like Vite, Bun, etc. have achieved all they meaningfully can. If I was the creator of these tools, I've move on too. Good luck and thanks for everything.
lol. any of them - literally just ONE - could have a full blown UI so that you wouldn't have to build projects using a command line like it's 1985. or maybe one of them could just invest in packaging custom html elements, instead of assuming I'm going to use one of a handful of unnecessary "component" libraries, or assuming that I won't be using components at all.
there's plenty of places for these tools to go, but none of them have any appetite to go there. likely because people already have something that's so "good enough" that they don't even bother looking for what "could be better". obviously exacerbated by the management class of development outfits deciding that developers shouldn't actually touch the codebase anymore, in lieu of LLMs doing the actual lifting, so they're building out all kinds of chicanerous nonsense to satisfy "agents". and that doesn't necessarily make things more difficult for devs, but that seems to be the trend. forcing your LLM to comply with tortured and arcane concatenations of character-perfect strings is so much easier than having it navigate anything like a filthy human. so the practical result is less accommodating stuff for humans and more accommodating stuff for robots.
all of which is to say: I disagree. I think there's things they could meaningfully achieve for humans. And I think they are deeply uninterested in doing those things.
> or maybe one of them could just invest in packaging custom html elements, instead of assuming I'm going to use one of a handful of unnecessary "component" libraries
... are these not the same thing? I suppose from a technical standpoint they'd differ, but they achieve the same result: reusable, modular building blocks for creating interfaces.
If Vite, Bun and uv were just "make builds faster" projects, then maybe the returns are diminishing. But the acquisitions by Cloudflare, Anthropic and OpenAI suggest this layer is becoming more strategic, not less.
These tools sit in the software supply chain: dependency resolution, project structure, tests, builds, runtimes, deployment paths and increasingly AI-agent execution loops. They define the default path for building software, and they are where AI-generated code gets tested against real dependencies, builds, tests and deployment constraints.
So I don’t think they’ve achieved all they meaningfully can. The value is shifting from raw build speed to control over the workflow layer where software is assembled.
This was my first thought when reading the title. I've spent the last month using chawan browser, which has a Gemini layer. And it's generally just a lot of fun to mix old web 1.0 layers with a new one.
I thought it was a cool essay because I've been using Gemini (the one mentioned in the article, not Google) in my terminal browser lately. And it has been a lot of fun!
But yes the whole "re-explain by negation" writing style does come across as AI-generated.
I've had this stubborn idea in my head, for a long time, that the petrol station up on Knight's Hill in South London is placed at the absolute nexus of London traffic to collect the best petrol rates across the entire city at any given time.
But yet to explore how I'd validate this idea. Your site helps remind me to have a dig at it.
reply