Hey HN, Enrico here. Some of you might remember Big-AGI from early chat days. The response here pushed me to quit my job and go all-in. Spent the last year rebuilding it properly (mostly hand-written) instead of shipping fast.

This is the tool I couldn't find: the AI you use to decide what to build, and why, before going to Claude Code or Cursor.

Even though the hype is on AI that 'does' things through heavy prompted experiences, I believe there's a strong need for AI that helps you "think" broader. How? with multiple models in parallel so you can spot hallucinations and lies, and so you can explore the solution space faster. And also with human bandwidth optimization: UI/UX is way underdeveloped for the "wall of text" phenomenon and hardly optimized for your I/O bandwidth besides voice input. And designed to short the generation/verification loop.

So Big-AGI 2 is now out, to be the fastest chat with a madly optimized UI, be local-first (app soon) because we don't like to wait 150ms/click. With Personas support because memories and data shall be with a relatable entity. If the cost of mistakes is very high for your work - wrong arch decisions, debt-inducing bad tech choices, bling spots - do yourself a favor, and try it.

Anyone can Free/unlimited at big-agi.com, or self-host from github.com/enricoros/big-agi (sync is Pro only, $11/mo).

Worth building a thinking tool while everyone builds doing tools? Worth bootstrapping rather than VC-lubing? Genuinely curious.