Microsoft opens OneLake data to AI agents with new tool
Microsoft has generally released OneLake MCP, a set of tools that lets AI agents securely access data in its OneLake storage system within Fabric. The release includes 19 commands for operations on files and tables, discovering data schemas, and inspecting Fabric workspaces. It integrates directly into the Fabric MCP extension for VS Code, relying on users' existing Azure permissions for access control.
Power BI reports often sat in silos with undocumented data models and slow refreshes, forcing analysts to manually dig through messy connections and Power Query steps without clear visibility. OneLake MCP changes that by enabling AI to automate schema checks and workspace scans, surfacing issues like broken relationships or unused tables that beginners miss – provided your Fabric setup has proper OneLake shortcuts pointing to clean sources. This shifts data maintenance from hours of trial-and-error to agent-driven insights, but exposes weak models faster, pressuring Excel holdouts to build proper ones upfront.
Analysis
AI agents sound fancy, but they'll just spit back errors from your inherited spaghetti reports unless your data model is solid first. This MCP launch proves Fabric rewards clean foundations over hacks. Install the Fabric VS Code extension now, sign in with Azure, and run one workspace inspection command on your boss's dashboard – fix the first red flag it flags.
Source note
Pulse published by Collab365 Spaces. Cite as "Microsoft opens OneLake data to AI agents with new tool", Collab365 Spaces.