AI has made building software fast enough to go from zero to feature-complete in hours—but the harder part remains: knowing what to build and having the intuition to cut what doesn’t belong.
Claude (Anthropic AI) announced that computer use is now available in Claude Code. Claude can open your apps, click through your UI, and test what it built, right from the CLI. Now in research preview on Pro and Max plans.
https://x.com/claudeai/status/2038663014098899416
Linear Head of Product Nan Yu shared thoughts on the current state of AI development.
https://x.com/thenanyu/status/2038602288957202941
https://x.com/thenanyu/status/2038599289081872893
Source: AI & I by Every — “Building Is the Easy Part Now | Mike Krieger on What AI Changed”
https://youtube.com/watch?v=KRv9GpJYrUA
Instagram co-founder and Anthropic product leader Mike Krieger has a refreshingly honest take on how AI has changed product development—and what hasn’t gotten easier.
The building part is solved. As Krieger put it, with today’s AI coding tools, “you can get it to go zero, not just zero to one, but zero to end pretty quickly over the matter of hours.” He rebuilt his old project Bourbon in just two hours with Claude—and it was feature-complete, even adding filters the original didn’t have.
But knowing what to build is harder than ever. The paradox: as building gets faster, the temptation to overbuild grows. Krieger admits that because you can add everything, you do—”we overbuilt for V1 before we even got to early access.” The result is a “matrix of functionality that was actually quite hard to test.”
The intuition problem. Krieger draws a powerful metaphor: building software used to be like growing a tree indoors without wind. The slow, iterative process forced you to make decisions one at a time, building intuition about what worked. Now you can grow the entire tree at once—but it lacks the “strength” that comes from exposure to real users at each step.
His solution: embrace rewrites. Since AI makes rebuilding fast (days, not months), Anthropic now freely tears down overcomplicated V1s and starts over. The lean startup principles still apply—they just operate at a different timescale.
The agent-native future. Krieger is bullish on “agent-native” products—software where agents can do anything a user can do. He points to Claude Code as the canonical example: “It’s an agent. It can do anything on your computer that you can do. It’s customizable and flexible and extensible.”
| *Built with follow-builders | Track the builders, not the influencers* |