How Datafor works
- Connect. Datafor queries your databases and warehouses directly: Snowflake, BigQuery, Redshift, Azure Synapse, SQL Server, PostgreSQL, MySQL, Oracle, ClickHouse, Doris, SAP HANA, and more. CSV and Excel files can be uploaded as file datasets.
- Model once. An analysis model describes tables, relationships, hierarchies, measures, time intelligence, and business terms. Measures can be bound to certified metrics in the Metrics Library.
- Govern centrally. Row policies, column hiding, and access control lists are defined once and enforced everywhere the model is used.
- Deliver anywhere. The same model powers interactive dashboards, embedded reports in your product, the REST API, the AI Assistant, and external AI clients over MCP.
Model once
Define metrics, dimensions, relationships, and business logic in one shared semantic model. When a definition changes, every dashboard, embedded report, and AI answer that uses it changes with it. Semantic modeling →
Govern centrally
Apply permissions, row-level security, and delivery controls consistently across every user and surface. Test what any user can see before you publish a policy. Governance and security →
Reuse everywhere
Power dashboards, embedded experiences, APIs, and AI-driven analysis from the same governed foundation, instead of rebuilding logic for each channel. Embedded analytics →
Trust AI outputs
Keep AI grounded in approved definitions, access rules, and traceable execution paths. The AI proposes a structured query, and Datafor validates and runs it under the asker’s permissions. AI Agent and MCP →
Deploy your way
Datafor runs in your own environment as a Docker image or installer package today. A hosted edition, Datafor Cloud, is in early access. Request access →