Integrate and extend agents in Copilot Studio
Enterprise Knowledge and Search Grounding
CoreChoose and govern Copilot connectors, Power Platform connectors, and Azure AI Search as enterprise grounding sources.
Aligned to the AB-620 live Skills measured outline; visible page date April 21, 2026; product behavior verified September 19, 2026.
Why this matters
Grounded answers are only as trustworthy as source selection, identity trimming, indexing freshness, and citation behavior. Enterprise content must remain constrained to what the requesting user may access.
Must Know
- Connect to Copilot connectors: Use a configured Copilot connector when enterprise content must be indexed for Microsoft 365 discovery and grounding with its permission model.
- Connect to Microsoft Power Platform connectors: Use the appropriate connector and connection to access supported service data or operations under Power Platform governance.
- Connect to Azure AI Search: Configure the approved Azure AI Search service and index, identity or key, searchable fields, and retrieval behavior as agent knowledge.
- Grounding retrieves evidence for an answer; it does not replace a tool when the requirement is to perform an operation or computation.
Compare and Distinguish
- Copilot connectors bring indexed external content into the Microsoft 365 search and grounding ecosystem; Power Platform connectors provide service operations and data access through connections.
- Azure AI Search supplies a controlled search index and retrieval configuration; adding a website or SharePoint source uses a different ingestion and permission model.
- Knowledge retrieves evidence for answers; tools perform actions or computations.
Scenario examples
- Scenario: A pilot succeeds but production users have different permissions. Think: trace identity and authorization through every source, tool, and delegated agent.
- Scenario: A component works in development but fails after deployment. Think: inspect solution dependencies, connections, environment configuration, and channel-specific behavior.
- Scenario: A response is fluent but the operational result is uncertain. Think: separate generation from retrieval, tool execution, telemetry, and verification.
Exam traps
- Using a broadly privileged connection that bypasses the intended user access boundary.
- Treating successful indexing as proof that retrieval quality and citations meet the use case.
- Adding overlapping sources without testing which source wins for conflicting content.
Key takeaways
- Select a grounding source by content location, permission model, retrieval needs, and governance boundary rather than treating every connector as interchangeable.
- Make the identity, data, lifecycle, and failure boundaries explicit for connect to enterprise knowledge sources.
- Validate the behavior with representative evidence before broad deployment.
How it works
- Copilot Studio uses the configured descriptions, instructions, conversation state, and orchestration mode to select the relevant topic, knowledge, tool, flow, or connected agent.
- Runtime identity and environment configuration determine whether the selected capability can access its dependency and return a governed result.
Objects and administrative surfaces
- Copilot Studio Knowledge page.
- Microsoft 365 admin surfaces for Copilot connectors and indexed content.
- Azure portal or Foundry surfaces for Azure AI Search indexes, authentication, and retrieval configuration.
When to use it
- Use this topic when the requirement maps to connect to enterprise knowledge sources.
- Prefer the narrowest supported surface that satisfies the required behavior and governance boundary.
Security and governance implications
- Apply least privilege to maker, runtime, connector, source, and deployment identities.
- Classify data and connectors, preserve audit evidence, and avoid exposing secrets or sensitive payloads in prompts and telemetry.
Troubleshooting signals
- Reproduce with the same version, channel, identity, inputs, and environment before changing configuration.
- Use activity traces, flow runs, tool errors, evaluation results, or telemetry to locate the failing layer.
More detail
- Use a Copilot connector to index external enterprise content for Microsoft 365 discovery and grounding; use a Power Platform connection when the agent needs supported service data or an operation.
- Azure AI Search provides a controlled retrieval layer. Configure the approved service’s target index, decide which fields participate and how queries behave, and use the intended key or identity for access.
- Grounding still follows its permission model: confirm which identity reaches the source and whether the intended user is allowed to receive the retrieved content.
- Do not confuse evidence retrieval with execution. Answer-supporting content belongs in knowledge; calculations or system changes belong behind a tool.
- Test representative queries, citations, permission differences, and no-result cases so a fluent response is not mistaken for successfully grounded evidence.
Ready for the quiz?
- Which requirement determines the correct surface for connect to enterprise knowledge sources?
- Which identity performs the operation, and where is authorization enforced?
- What evidence distinguishes a safe successful result from a fluent but incomplete one?
Related objectives
- D2.1.S1 — Connect to Copilot connectors
- D2.1.S2 — Connect to Microsoft Power Platform connectors
- D2.1.S3 — Connect to Azure AI Search