Agentic RAG
Chroma
Open-source embedding database for AI applications.
Best for local-first RAG prototypes and teams that want a low-friction vector store while validating retrieval quality.
Use Chroma to move quickly while retrieval requirements are still changing. Revisit scale and tenancy before production growth.
Best for
- local RAG
- fast prototypes
- embedding collections
Not ideal for
- large multi-tenant production search without an ops plan
- cases requiring managed enterprise controls immediately
Core concepts
collectionsembeddingsmetadataqueries
Minimal implementation shape
Create a collection, store chunks with metadata, query by embedding, and inspect whether the returned context actually answers the task.