I built Brains because I kept losing context. Every new chat with an AI started from zero. I'd re-explain the project, re-paste decisions I'd already made, re-establish what we were building. The AI was brilliant and completely amnesiac — every session. So I built a persistent knowledge base that connects to any MCP-compatible AI. Agents read from it and write back to it. Decisions made in one conversation show up in the next one, on any device, with any model. The process evolved in a way I didn't expect: Brains became the primary tool I used to build Brains. Every architecture decision, every open loop, every copy draft — stored in the wiki, picked up by Claude in the next session. The AI I was building for was also the AI doing the work. The product validated itself before it launched. This answer? Drafted with help from Claude, using context pulled from Brains. That's the pitch. That's also just what building it looks like.

brains

Persistent memory for your AI. Any model, any device.

Persistent memory for your AI. Any model, any device.