What is ParticleDB?
ParticleDB is a database for AI agents and applications. It combines SQL, key-value access, vector search, analytics and functions called Particles in one database engine.
Use it to store application records, retrieve context, track work and query operational data. Your application controls the context an agent receives and the actions it may take.
The building blocks
Section titled “The building blocks”| Need | Start here |
|---|---|
| Records, relationships and transactions | Relational data |
| Keys, expiration and coordination | Key-value |
| Relevant context and embeddings | Search and memory |
| Reports on application records | Analytics |
| Custom logic beside the data | Particles |
| Agent state and application workflows | Agents and workflows |
Choose a starting point
Section titled “Choose a starting point”New installation: follow the quickstart.
Existing application: choose a language client or read the migration guide.
Looking up behavior: use the SQL reference and limits.
Data and execution boundaries
Section titled “Data and execution boundaries”Data in other services needs an explicit import or integration. Storing data in one engine does not automatically authorize an agent to access it. Application workers and external tools also need their own retry and idempotency handling. Security and workflow patterns explain these boundaries.