LLM FutureLove this LLM experiment: "Everyone's focused on AI that writes code. You know the usual suspects, Claude Code, Cursor, Copilot, all that. But that felt like missing the bigger picture. So I built something to test a different question: what if you skip code generation entirely? A web server with zero application code. No routes, no controllers, no business logic. Just an HTTP server that asks an LLM "what should I do?" for every request. The goal: prove how far away we really are from that future." And the interesting thing is it actually worked: "But despite all that, forms actually submitted correctly. Data persisted across restarts. The UI was usable. APIs returned valid JSON. User feedback got implemented. The AI invented, without any examples, sensible database schemas with proper types and indexes, parameterized SQL queries that were safe from injection, REST-ish API conventions, responsive Bootstrap layouts, form validation, and error handling for edge cases. All emergent behavio[u]r from giving it three tools and a prompt." It's not fast, it's not cheap, but certainly gives a hint at the future. Will LLMs become computers?