Magnum BI is an open-source business intelligence tool, currently in alpha. Users can chat with their model and get analysis and dashboards out of it. Analysts can build dashboards with drag-and-drop and never write SQL. Engineers can keep every model, dashboard, connection, member and row filter as plain YAML in git, and apply changes after a Terraform-style plan. The web UI uses the same HTTP API that scripts and the CLI use, so a change gets the same checks, permissions and audit trail whether it comes from a click, a file or a script.
It runs on ScramDB. Each organisation gets its own instance, which holds its dashboards and a cache of query results. That keeps tenants isolated and makes cached answers fast. You switch caching on per model, and cached results respect each person’s row filters.
Key features:
Drag-and-drop explorer with spreadsheet-style formulas, drill-down and period comparison
Data catalog with owners, certified, draft or deprecated status, a shared glossary and usage lineage
Blends that join data from PostgreSQL, BigQuery, DuckDB and ScramDB without moving it
Row-level security, roles and multi-tenant organisations
One-line embedding, public links and QR codes, plus password-protected guest links that expire
Scheduled reports by email, Slack or Teams, and threshold alerts
A chat assistant that answers in plain language, using your own AI provider and key (OpenAI, Anthropic, Mistral or a self-hosted model)
An MCP endpoint and scoped, expiring API tokens for AI agents
Passkeys, authenticator codes and OpenID Connect SSO
Organisation themes, with CSS design-token import and export
Ways to run it: a Docker image, Docker Compose or a Helm chart for teams. There’s also Magnum BI lite, a standalone desktop app for one person that needs no server or accounts and can open CSV and Parquet files.
EU
Giovanni Binda 
No comments or reviews, maybe you want to be first?