Typed, versioned schemas
Evolve domain contracts through immutable revisions and explicit compatibility.
SEMANTIC COMPILER AND EXECUTION LAYER
Schema-first data infrastructure for the Agentic Age
CatDB is an application-facing data layer that lets applications and AI agents build against your domain instead of your storage.
Define Customers, Claims, or Contracts once, then map them to the databases and warehouses that already hold the records. CatDB checks every query against a versioned schema, sends each source the work it can handle, and returns typed results with complete provenance.
CatDB query foundations
Validated mappings
Pin each physical representation to an exact semantic schema version.
Capability-aware plans
Compile each read against source capabilities and policy.
Field provenance
Return the provider that answered every field.
request path / governed execution
One schema between your application and every source.
Applications and agents query the CatDB model. CatDB validates the request against the active schema, plans the reads each source can perform, and returns one typed result with provenance.
callers
Applications · agents · BI tools · MDReader
interfaces
CatDB schema · API · MCP · read-only SQL compatibility
Semantic control plane
Versioned schemas · Mapping versions · Constraints · Provenance rules · Policies
Query data plane
Query planner · Connector runtime · Pushdown execution · Result reconciliation · Provenance capture
backend-family model
PostgreSQL-compatible · Snowflake · DuckDB · Redshift · MongoDB
connected systems / read-only record authority
One semantic model across databases and warehouses.
Connect databases and warehouses to the same typed domain without moving their source records into CatDB.
One semantic modelschema · mapping · policy · provenance
- PostgreSQL-compatible
- Snowflake
- DuckDB
- Redshift
- MongoDB