Extend Dynamics 365 with Copilot Studio agents
Copilot Studio is Microsoft's low-code platform for building AI agents. The value of these agents multiplies when you bring them into the surfaces your users and customers already use. This unit focuses on how you surface and integrate Copilot Studio AI agents within Dynamics 365 apps, not on how to build them.
What Copilot Studio is
Copilot Studio is a low-code tool in the Power Platform for building, testing, and publishing AI agents. These agents can answer questions, trigger workflows, carry on multi-turn conversations, and perform actions in connected systems. You build agents using a combination of natural language topic definitions, generative AI capabilities—including connections to knowledge sources like SharePoint sites, websites, or uploaded documents—and Power Automate flow integrations.
The platform handles most of the underlying AI complexity, so a functional consultant with no AI development background can build a capable agent in a short time. But the real impact comes from putting that agent in the right place, at the right moment in a user's workflow.
Integration versus creation
There's an important distinction between building an agent in Copilot Studio and integrating it into a Dynamics 365 app. Building involves defining topics, connecting knowledge sources, configuring actions, and testing responses. Integration is about deploying the built agent to a channel, the surface where users interact with it.
Copilot Studio lets you deploy agents to many channels simultaneously: Microsoft Teams, websites, Power Pages portals, and Dynamics 365 apps.
This unit focuses on integration: getting a built agent surfaced where it needs to be within Dynamics 365.
How agents are surfaced in Dynamics 365
There are three primary ways to surface a Copilot Studio agent within Dynamics 365:
As an omnichannel AI agent in Customer Service
The most tightly integrated deployment is in Dynamics 365 Customer Service and Contact Center. Copilot Studio agents can serve as the automated first point of contact for incoming customer conversations. When a customer opens a chat session on a support portal or website, the agent handles the conversation initially, collecting information, answering common questions, and resolving issues that don't require a human agent.
When the conversation exceeds what the agent can handle, it uses Copilot Studio's omnichannel handoff capability to escalate to a customer service representative (CSR). The full conversation transcript, captured intent data, and any information the customer provided transfer to the CSR's session automatically. The customer doesn't have to repeat themselves.
As a chat widget embedded in a form or portal
For Dynamics 365 Sales, Field Service, or any Power Pages portal, you can embed a Copilot Studio agent as a chat widget. The widget appears as a floating chat button or an inline chat panel on the page. Users interact with the agent without leaving the record or page they're on.
This approach is useful for providing contextual assistance tied to what the user is currently doing. A sales rep on an opportunity record asks the agent for deal coaching suggestions. A field technician on a work order asks the agent for equipment troubleshooting guidance and gets back relevant knowledge article content without navigating away.
As a Microsoft Teams app
Copilot Studio agents can be published directly to Microsoft Teams and made available in the Teams app catalog. This works well for agents that answer questions about Dynamics 365 data when users' primary workspace is Teams rather than the Dynamics 365 app. A sales manager might ask the agent "What deals are at risk this quarter?" in a Teams chat without opening Dynamics 365 at all.
Use cases across Dynamics 365 apps
The following table shows how each integration surface maps to a specific app and scenario.
| App | Integration surface | Use case |
|---|---|---|
| Dynamics 365 Customer Service | Omnichannel AI agent | A tier-1 case deflection AI agent that resolves common issues like password resets, order status, or FAQ responses before transferring unresolved conversations with full context to CSRs |
| Dynamics 365 Customer Insights | Teams app | A data Q&A agent that answers questions like "What's the size of our high-value segment in EMEA this month?" by connecting to Customer Insights data through a Dataverse or MCP connection |
| Dynamics 365 Field Service | Chat widget on Work Order form | A field tech assistant that helps technicians troubleshoot asset issues, surface relevant knowledge articles, and log diagnostic notes, all from the work order they're already working on |
Integration steps overview
The integration process follows a consistent pattern across deployment types:
- Build and publish the agent in Copilot Studio. Test it against your knowledge sources and confirm it handles the intended scenarios correctly.
- Configure the channel in Copilot Studio. For omnichannel Customer Service, configure the escalation and handoff settings. For web embedding, copy the code or channel configuration.
- Add the AI agent to Dynamics 365. For omnichannel: set up the AI agent connection in a workstream in Customer Service admin center. For form embedding: add the code snippet to a web resource or configure it as part of the form.
- Test end-to-end in the target environment, including any escalation or handoff paths for customer-facing agents.
Tip
When integrating an agent with Dynamics 365 Customer Service for omnichannel, test the escalation path thoroughly with realistic scenarios. The quality of the handoff—how complete and useful the transcript is when it reaches the human agent—directly affects the customer experience and agent efficiency.
Learn more about configuring Copilot Studio agents in Dynamics 365 Customer Service.
Copilot Studio agents work best when they have real-time access to your Dynamics 365 data. In the next unit, you'll explore Model Context Protocol (MCP), the open standard that lets external AI agents query Dataverse directly, without custom connectors or data exports.