Microsoft Fabric adds one-line AI tools for classifying and summarizing data at scale
Reviewed by Helen Jones
Microsoft updated its Fabric documentation for AI Functions that apply large language model transformations with a single line of code. The functions cover sentiment analysis, classification, extraction, summarization, translation, grammar fixes, embeddings, similarity checks, and free-form responses. They work on pandas and PySpark DataFrames in notebooks, through T-SQL in warehouses and SQL analytics endpoints, and inside Dataflow Gen2 via Fabric AI Prompt. Multimodal inputs now include images, PDFs, and text files. The default model is gpt-5-mini with a 400,000-token context and high concurrency that processes up to 200 rows by default. Access needs the tenant Copilot switch turned on and paid Fabric capacity of F2 or higher. Some surfaces remain in preview, English is the primary focus, and the service does not log prompts or outputs.
Until now, most reporting teams handled free-text columns the hard way: export to Excel, manual tagging or formulas, then re-import. That kept enrichment outside the scheduled refresh path and made every new source of comments, notes, or tickets another fragile step only one person understood. These functions move the same enrichment work into the Fabric pipeline itself, so classification or extraction can run with the rest of the dataflow or warehouse query. The numbers that land in Power BI still need the same reconciliation against source totals and business definitions. The change is operational convenience, not automatic trust.
Analysis
Treat this as a controlled test opportunity rather than a production shortcut. If your tenant already has Copilot enabled and F2+ capacity, take one free-text column you currently clean by hand, run ai.classify or ai.extract in Dataflow Gen2 on a sample, and compare the results side-by-side with your existing Excel checks before you wire it into any report stakeholders rely on.
Source note
Pulse published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "Microsoft Fabric adds one-line AI tools for classifying and summarizing data at scale", Collab365 Spaces. 2 sources referenced.