Vibe Coding Is Over: Building Software When AI Writes the Code

We’re sitting down with Robert Douglass, founder of Spec Kitty, to talk about what happens after vibe coding. AI coding agents are getting very good, very fast. But as they write more of the implementation, a new set of problems shows up: How do you keep them aligned with what you actually wanted?How do you manage multiple agents working at once?How do you review work you didn’t write?And how much autonomy should you really give them? Robert built Spec Kitty, an open-source framework for spec-driven development with AI agents, to tackle exactly that. We’ll talk about why he built it, what he learned from letting agents loose on real codebases, where vibe coding breaks down, and what software development might look like when humans spend less time writing code and more time defining intent. We’ll also get into: Why better prompts aren’t enough Specs vs. vibe coding How to coordinate multiple coding agents Where humans still need to stay in the loop What breaks when AI-generated code scales Whether developers will actually read the code AI writes Building an open-source product in a space moving this fast Expect a candid founder conversation, live examples, and questions from the room. Then we’ll hang out over some drinks with other founders, engineers, and people building with AI.