Agent Interface
REST APIgRPC APIPython/Node SDKsMCP Server (community)
What an agent can do
- ✓Create, manage, and delete vector indexes
- ✓Upsert, query, and delete vectors with metadata filtering
- ✓Perform hybrid search (dense + sparse vectors)
- ✓Manage namespaces for multi-tenant isolation
- ✓List and describe index statistics
- ✓Fetch vectors by ID for inspection
What it can't do
- ✗Generate embeddings (bring your own from OpenAI, Cohere, etc.)
- ✗Run SQL-like queries on metadata (filter syntax only)
- ✗Access index configuration parameters after creation
- ✗Export full index data to file (must paginate through list+fetch)
- ✗Modify account or billing settings via API
Auth Model
API Key (environment-scoped)
Builder Notes
Purpose-built vector DB for agent memory and RAG pipelines. Serverless indexes auto-scale. Metadata filtering enables hybrid retrieval. gRPC API offers lower latency for high-throughput agent workloads. Free tier: 1 index, 100K vectors.
Sources
Last verified: 2026-03-15