Connect in minutes
Add Stripe, HubSpot, Postgres, or parquet files with one command. 100+ sources supported.
Introduction
Dinobase is the SQL query layer for agent stacks. It connects 100+ business data sources and makes them queryable via a unified DuckDB interface — agents write SQL, cross-source JOINs work natively.
Agent stacks built on per-source tool calls have a structural gap: agents can’t JOIN across APIs, have no semantic context to interpret field values, and receive paginated JSON that fills context windows. Dinobase fills that gap. In benchmarks across 11 LLMs: 91% accuracy vs 35%, 3x faster, 16x cheaper per correct answer.
Why Dinobase
Section titled “Why Dinobase”Cross-source queries
JOIN Stripe customers with HubSpot deals with product analytics. Questions that required a data team now have instant answers.
Semantic layer
After every sync, a Claude agent automatically annotates your data: table descriptions, column docs, PII flags, and relationship graphs. Agents know what every field means before writing a query.
Accurate answers
Semantic context + complete data = 91% query accuracy vs 35% for bare MCP tools, across 10+ LLMs.
Next steps
Section titled “Next steps”- Getting Started — install, connect your first source, run your first query
- Semantic Layer — auto-annotation, relationship graphs, PII detection
- Connecting Sources — all 100+ supported source types
- MCP Integration — use Dinobase with Claude, Cursor, and other agents