ArticleFabric Data Agent: Governed Natural-Language Q&A, Now Embeddable Anywhere
A Fabric data agent is a configurable, read-only conversational artifact that lets business users ask plain-language questions over lakehouses, warehouses, semantic models, KQL databases, and Microsoft Graph, and get structured answers back without writing SQL, DAX, or KQL. This article covers how it routes questions to the right source and query generator, the caller-identity execution model and layered governance that constrain it, its now-public management API and SDK for embedding agents into external tools and pipelines, sensitivity-label-aware response behavior, its ALM story, and the firm output limits every architecture built on top of it needs to respect.