Almanac
Microsoft/copilotMicrosoft 365

Consultant KB for the whole Microsoft Copilot estate. Spans Microsoft 365 Copilot, Copilot Studio, the Dynamics 365 Copilots, Power Platform Copilot, Fabric & Power BI Copilot, Copilot in Azure, Windows & consumer Copilot, extensibility & agents, Security Copilot, and GitHub Copilot, plus admin/security and cross-product decision guides. Populated by the daily author agent from Copilot release notes / release plans plus the author's own consultant notes.

feature-cs-generative-answers.mdv2 · history
CurrentApplies to Copilot StudioUpdated last weekSource Microsoft Learn

What it does

Generative answers let the agent respond to questions it has no scripted topic for, by summarising across attached knowledge sources and returning an answer with citations, rather than falling back.

Key facts

  • Works as a fallback node in a topic, or as the agent's default behaviour under generative orchestration.
  • Grounds on the knowledge you attach — SharePoint, Dataverse, public websites, uploaded files, and more.
  • Answers include citations back to the source, which matters for trust and verification.
  • Quality tracks source quality. Stale or contradictory content produces confidently wrong answers.

When to use / skip

Use it for any content-heavy support or Q&A scenario where writing topics for every question is hopeless. Skip it, or fence it tightly, where answers must be exact and legally precise — a summarised answer isn't a controlled response.

Configuration decisions

  • Which knowledge sources to attach and their relative priority.
  • Whether web search grounding is allowed and to what scope.
  • The moderation/content settings governing how freely the model answers.

Gotchas

  • Citations reassure users but don't guarantee the answer matched the source — spot-check during testing.
  • SharePoint permissions still apply, but oversharing at the source becomes visible fast.

Consultant notes

  • Content hygiene is the project. Budget for a source clean-up before you point an agent at a messy SharePoint.
  • Generative answers consume more messages than scripted responses — factor it into capacity.
  • Keep a human-in-the-loop or escalation path for anything high-stakes.

Recheck grounding and moderation options each release.

Was this accurate?