Colrows v1.5 · The semantic execution layer

Compile enterprise intent into governed, deterministic SQL.

Colrows sits between users, AI agents, and your data systems. It resolves business meaning at compile time, proves every join, and runs dialect-perfect SQL inside a policy-shaped subgraph - so what executes is exactly what was authorized.

Start here

The Colrows runtime

Every request follows the same compile-then-execute pipeline. Ambiguity is resolved at compile time, not at runtime - and policy shapes the plan before any database is touched.

Diagram showing the four-stage Colrows pipeline: intent, context resolution, constrained planning, governed execution.
Intent → Context resolution → Constrained planning → Governed execution.

What Colrows does

  • Versioned semantic graphEntities, metrics, events, concepts, constraints, policies and personas - modeled as typed nodes with point-in-time reproducibility.
  • Compile-time governanceRBAC, ABAC, and row/column-level predicates shape the allowed subgraph. Unauthorized plans are never generated.
  • Join path proofEvery multi-entity query is solved as a constrained graph traversal - ambiguous paths fail compilation rather than silently producing wrong numbers.
  • Autonomous maintenanceDrift detection, conflict detection, and schema evolution run continuously. The graph keeps itself current as the enterprise changes.
  • Multi-vector embeddingsEach concept carries definition, usage, and combined vectors - recall stays accurate as language and context evolve.
  • Dialect-perfect SQLOne semantic plan, many backends. Snowflake, Databricks, Postgres, ClickHouse, Trino, Oracle, SQL Server, and more.

Build with Colrows

Further reading