Encrypted index
Embeddings, IDs, and metadata stored only as ciphertext — per-record IVs, authenticated. Compromise the disk, read nothing.
Records AES-256-GCM
Index nodes ciphertext-only
/ V0.16.1
The Encrypted Vector Database for Enterprise AI
CyborgDB keeps inference data end-to-end encrypted — at rest, in transit, and in-use. Build secure enterprise AI on the only encrypted vector store.
TRUSTED BY
> client.query( "patient SSN 123-45-6789" → "RlcjognV5Ikx2QaP" )
QUERY> embed( [0.127, -0.384, 0.552, ...] )
VECTOR> encrypt( dims=1536, norm=L2 )
SEALED> store( Dx: Hypertension )
INDEXED> match( top_k=10, ε=0.02 )
SEARCH> retrieve( match: Dx: Hypertension )
RESULT> decrypt( auth: tenant_a9f2 )
REVEALLATENCY 8ms p95 encrypted query
THROUGHPUT 0.0k/s vectors indexed
KEY BYOK / HYOK never leaves tenant
Encrypted at every stage.
Even at search time.
Already encrypted on the wire
Payloads are AEAD-encrypted before they ever touch the network. TLS rides on top — but it isn't what's keeping the data safe.
Payload AES-256-GCM
Transport TLS
Search over ciphertext
Forward-secure ANN search runs on encrypted index nodes. The server never holds plaintext embeddings or full index keys.
Search tokens scoped
Index traversal ciphertext
Keys
Client-side custody. BYOK or HYOK. Nothing in the server-side pipeline can decrypt without your key.
Ship AI securely.
SOC 2 Type II
CLEARED
HIPAA
CLEARED
GDPR
CLEARED
ISO 27001
CLEARED
Unblock regulated AI
Ship GenAI without a six-month security review.
End-to-end encryption and cryptographic access control walk into every audit with the box already ticked.
Weeks, not quarters Time-to-prod in regulated environments
YOUR STACK
+ postgres redis s3
proxy
Reuse your stack
Turn the databases you already run into a secure vector store.
Proxy mode sits in front of Postgres, Redis, or S3. No migration, no net-new infrastructure.
Zero migration Sidecar deploy, single Docker image
Multi-tenant AI
One shared index + per-tenant keys = cryptographic isolation.
Each tenant's data is encrypted with their own key. Boundaries enforced in math, not application logic.
N tenants, 1 index Crypto-enforced isolation
Encryption, without the tax.
Security doesn’t have to mean compromising performance. CyborgDB keeps pace with unencrypted vector databases — and beats most of them.
CyborgDB encrypted
Qdrant
Weaviate
Milvus
Elasticsearch
pgvector
DATASET wiki-all-1M · 768 dims · 1M vectors · top-k = 10
Built for sensitive workloads
When the data you're searching can't leave your control.
Secure RAG,
at NVIDIA scale.
The Cyborg–NVIDIA Enterprise RAG Blueprint is a reference architecture for deploying secure, scalable AI applications. Cyborg is one of just eight companies invited by NVIDIA to author an Enterprise Blueprint.
It combines best-in-class NVIDIA NIMs with CyborgDB to deliver encrypted-in-use retrieval for AI knowledge bases.
- 15× faster multimodal PDF extraction
- 50% fewer incorrect answers
- Zero plaintext exposure of vector data
Reference Architecture build.nvidia.com / cyborg
Deploy in < 1 hour.
One container, one config, one import change. Keep your stack, reuse your data plane, ship to prod the same afternoon.
0m15m30m45m60m
Step 01 ~2 min
Install
shell
# Pull CyborgDB image
docker pull cyborg/cyborgdb:latest
# Launch CyborgDB service
docker run -p 8000:8000 cyborg/cyborgdb
Step 02 ~10 min
Configure
cyborg.yaml
backend:
kind: postgres
dsn: ${POSTGRES_URL}
keys:
kms: aws
key_id: alias/cyborg-prod
audit:
sink: s3://acme-audit/cyborg/
Step 03 ~45 min
Ship
from cyborgdb import Client
# key never leaves your app
index_key = client.generate_key()
index = client.create_index(index_name="patient-notes", index_key=index_key)
# search runs on ciphertext
index.upsert(items)
results = index.query(query_vectors=q, top_k=10)
Shipped Encrypted vector search running in your VPC, backed by your existing infra, keys in your KMS.
Encrypt everything.
Search anything.
Free for up to 1M vectors. Compliance-ready from the first query.