Wren AI, built by Canner since 2024, brings a software-engineering practice into business intelligence that's unusual for this category: its semantic layer, called MDL, and the agent skills that use it are versioned, branched, and reviewed through pull requests the same way application code would be, rather than living as opaque configuration inside a closed platform.
That git-native governance sits underneath an agentic query system: rather than a single-turn chat exchange, Wren AI's agents reason over the governed semantic model across multiple steps in sandboxed environments, enforcing row- and column-level access policies consistently whether a query comes through the web interface, an API, or an embedded integration. More than twenty data source connectors keep it dialect-agnostic across different database engines.
The core platform is open source and free, with commercial features available on top. For an engineering-minded organization that wants BI governance managed with the same rigor as software, version control, code review, branching, applied to how business metrics get defined and queried, Wren AI's git-native approach to semantic governance is a distinctive, structurally different model than a closed commercial BI tool has.
