For solo devs, vibe coders & small teams

Ship RAG and GraphRAG on one engine.

CEREBRAL is the vector + graph database for AI search: one API for vectors and relationships, compressed from the first write. Run it as one Docker image, or as an extension in the Postgres you already have.

Request early access ↓Email compression@crbrl.ai

Designed for up to 8x^*smaller vectors, design target

GraphRAGwithout a second database

MCP serverconnect your agent

Get in touch

Request early access. Join early access.

A few details help us put you in the right batch.

First name

What are you building?

Storing vectors in today (optional)

Hosted vector databasePostgresSelf-hosted vector databaseNot yet

Embedding dimensions (optional)

Chunks / vectors (optional)

Team size (optional)

Preferred way to run it (optional)

GitHub or project link (optional)

I build for clients and would like details of the referral program when it opens.

I agree to CEREBRAL contacting me about this request, and I have read the Privacy Policy. One-click unsubscribe on anything that isn't a direct reply.

Self-serve sign-up opens soon; you'll be in the first wave. Routed to compression@crbrl.ai.

Questions we get early

Is there a free tier?

Builder pricing isn't final, so we won't quote it here. Priority-access members get pricing first.

Do I have to rewrite my app?

CEREBRAL speaks REST, gRPC and native MCP. On Postgres, crbrl-pg runs as an extension inside it.

What does "designed for up to 8x" mean?

A per-vector design target, not yet independently benchmarked. Index and graph edges add overhead, and results depend on your corpus.