Your goals, decisions, and open promises live in one memory it rereads every session. Three times a day it wakes on its own to brief you, move the work, and tell you what changed.
You paste the same context every morning because the model has no idea what happened yesterday. AI Cofounder opens already aware — the goal, the last decision, the promise you made, and the thing quietly going stale.
Not notifications. Each wake has a job, a stopping point, and a bias toward finishing reversible work instead of asking you about it.
Goals, decisions, tasks, and working history live as plain text you can open, correct, and keep. No mystery blob — you can read the memory, and so can it.
Every session it rereads the tree and writes back what changed. Correct it once and the correction sticks — for good, not for the length of a context window.
Impact, confidence, and whether the thing can actually be done right now — checked before anything runs.
Your Claude or ChatGPT account is the surface. Your connected email, calendar, and docs are the hands.
Your rules live in your account, where only you can change them. It acts freely on private work that can be undone. Four things always stop for your yes.
Purchases, payments, and financial commitments.
Posts, publishing, and work that represents you outwardly.
Reaching someone who has not contacted you first.
Strategy, pricing, or a commitment already made.
AI Cofounder runs inside the paid Claude or ChatGPT account you already have. No second workspace to remember, no extra inbox, no dashboard that becomes another thing to keep up with.
Start with a real conversation. We map the work, the tools, and the rules your cofounder runs by — then set it up inside your own account.
Terrell Gentry builds AI systems for working founders. 6omb is small on purpose. Replies come from a person, usually the one who wrote the code.