Build · company brain

The brain your business owns

Install the memory layer on inference you already own — or we stand both up together. You hold root. Steward access is named, scoped, and revocable. built on Mnemma

What you hold when we leave

Root, the map, and the morning brief

  • The graph, receipts, ingest, decay, and re-verify runner — at the last shipped version.
  • Admin. Steward access is named, scoped, revocable.
  • The Admin Card, a signed access matrix, and a signed container policy.
  • Cancel-and-keep as contract, not homepage swagger.
  • Morning brief as the first surface: work that never left your servers.

The install

Five phases, one owner

Phase What happens What you hold
0 · Fit Census of systems, data classes, and who holds root today. The written picture of your environment.
1 · Inference Private model standup (or we start from one you already bought). See Private AI. Serving, keys, Admin Card.
2 · Map Classification workshop. Named pack connected. First receipts on the graph. Policy for what may hit the model.
3 · Brief Morning brief live. Coverage you can see. Recount scheduled. A daily surface on your data.
4 · Walk Day-14 walk. Restore drill. Revoke-us rehearsal. Root, without us.

How this attaches

The brain sits on inference you own

If you already have Private AI, we add the map. If you don't, we stand both up as two line items on one engagement. We do not offer a brain on a lab API.

Connectors are named on the quote, not implied. We will not advertise “we connect everything.” Until a named pack is real, this install is founder-led and scoped from the ops audit.

Next step

See whether a brain is the next move — or whether inference is.

To get started, book a fit call. Many teams buy Private AI first and come back for the map.