Vector Database Built for Production AI

CyborgDB is a self hosted encrypted vector database that runs approximate nearest-neighbor search directly on encrypted data, so embeddings stay encrypted in storage and during search.

Only the final candidate results are decrypted, briefly and in memory, after a key check against the customer's own KMS, and per-tenant keys give multi-tenant deployments cryptographic isolation instead of application-layer row filters.

Cyborg has a published RAG blueprint with NVIDIA and technology partnerships with NVIDIA, DDN, NetApp and F5.

The platform is built for RAG pipelines and AI agents that handle regulated data in healthcare, financial services, defense and the public sector.

Cyborg is venture funded and based in New York.

Founder Nicolas Dupont holds 14 US patents, including the encrypted search patents behind CyborgDB.