Skip to content

An editorial team and a community building with AI

vibecoding.incJoin Discord ↗
← All articles

News · Microsoft · 2 min read

Event date: · Published:

Azure AI Foundry adds agentic mode to Content Understanding in public preview

Microsoft's Content Understanding service in Azure AI Foundry gained a public preview agentic extraction mode that runs a multi-step reasoning loop for complex document extraction tasks.

Microsoft announced on November 9, 2026, that Azure AI Foundry's Content Understanding service now offers an agentic extraction mode in public preview. The feature was detailed by Joe Filcik, Principal Product Manager on the Microsoft Foundry team, in a Microsoft Foundry Blog post, When one pass isn't enough: agentic mode in Azure Content Understanding.

What's new in agentic mode

  • The feature is called agentic mode for Azure Content Understanding, part of Microsoft Foundry, and is currently in public preview.
  • It can be enabled through the new API version 2026-06-01-preview, or by setting workflow to agentic in an analyzer definition.
  • It is also selectable in the Content Understanding Studio UI as Agentic Extraction under Workflow Kind.
  • The mode runs a reasoning loop with tools that navigates document text and tables, inspects visual evidence, performs calculations, and validates its own results.
  • Demonstrated use cases include treasury calculations across multi-page tables with exclusion rules, interpreting economic charts that need contextual reasoning, and filtering and totaling network logs across 19 pages.

What it means for AI builders

For developers building document-intelligence or RAG pipelines on Azure AI Foundry, agentic mode adds a built-in reasoning loop for complex multi-page or multi-table extraction, so teams no longer need to hand-build their own agent loop to handle hard documents such as multi-page financial tables or charts that require contextual interpretation.

Trade-offs and availability

Microsoft states that agentic mode trades higher token usage and latency for improved accuracy on hard extraction tasks, so it is best suited to documents where a single-pass analyzer falls short. The mode is available now in public preview through the API or the Content Understanding Studio UI.

Sources

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.

Azure AI Foundry adds agentic mode to Content Understanding in public preview | vibecoding.inc