Cerebro Hosted

Managed long-term memory for your agents.

agent-cerebro gives your agents persistent, searchable memory with semantic recall. Hosted takes the ops off your plate: managed embeddings, team sync, and an API key — no local SQLite file to back up, migrate, or babysit. Early-access pricing TBD.

[ embeddings ]

Managed embeddings

We run the embedding model and vector index. No API keys to wire up, no dimension migrations, no "which model did I index with again?"

[ sync ]

Team sync

One shared memory store across machines and teammates. Your agents read and write the same knowledge instead of siloed local files.

[ api ]

Just an API key

Same store / search interface you already use, pointed at a hosted endpoint. Drop-in for the open-source CLI.

[ ops ]

No SQLite to operate

No local DB file to back up, lock, or corrupt during a deploy. Memory lives server-side with the durability that implies.

$ cat NOTES.md

Straight talk: the hosted backend isn't built yet. The open-source agent-cerebro CLI works today and always will — local-first, no account required.

This page exists to find out if a managed tier is worth building. Join the waitlist and we'll email you once — when early access opens — with pricing and an invite. That's it.

Run agents in production? Skip the memory ops.