Add an agent node to an agent flow

The agent node lets a flow hand a step over to an AI agent that can reason, call tools, and pull from knowledge sources before returning a result. Use it whenever a step needs judgment, multi-step orchestration, or information from outside the flow itself.

By using the agent node, you can:

• Call an existing agent.
• Give the agent a natural-language instruction, including dynamic content from earlier steps.
• Equip the agent with tools (Model Context Protocol (MCP) servers and connectors) so it can take action.
• Ground the agent in knowledge (SharePoint sites, public websites) so it can answer from your content.
• Use the agent's response in later steps of the flow.

In agent flows, you can either select an existing published agent or create a new published agent from the agent builder experience and then call it from the flow.

Add an agent node

1. In Copilot Studio, go to Flows and open an existing agent flow or create a new one.

   • New flow: You land on the Designer page to configure a trigger.
   • Existing flow: You land on the Overview page, and you have to go to the Designer page.

2. Select the Insert a new action icon at the point in the flow where you want to call an agent.

3. In Add an action, under AI capabilities, select Run an agent. A Run an agent node appears in the flow.

Choose an existing agent for the agent node

In the run and agent editor, under Agent, you can select one of two options:

• Select an existing agent: Select a published agent from a list. The agent runs with whatever instructions, tools, and knowledge it was already configured with.

• Create a new agent: Select New to create a new agent. This opens a new browser tab with the agent builder experience. Build and publish your agent there, then return to the flow designer to select it from the list of existing agents.

If you select an existing agent, skip to Send a message to the agent.

Create a new agent for the agent node

When you select New, you build the agent in the agent builder experience that opens in a new browser tab. In the agent builder, you configure the agent's instructions, knowledge, and capabilities.

When you're done building the agent, publish it, return to the flow designer, and select it from the list of existing agents.

Send a message to an existing agent

When you pick an existing agent, the node shows a Message field. This is how you tell the agent what to do for this run. The agent's instructions, tools, and knowledge are already set on the published agent.

In the Message field, write the prompt for this run. Use the dynamic content picker to insert tokens from earlier steps so the agent reasons over real run-time data.

Request human assistance when unsure

Turn on Request human assistance when unsure if you want the agent to escalate to a human when it can't proceed on its own. When an escalation occurs, the agent sends a request for input by email to the connection owner.

This option is useful when the agent's task might require interpretation that goes beyond its instructions or available knowledge. For example, if an agent is reviewing expense reports and finds an unusual charge that doesn't match any policy, it can flag the report for a person to decide rather than making an incorrect approval.

Use the agent response in your agent flow

When the agent node runs, the agent flow waits for the agent to complete its task. The agent response then becomes available as dynamic content that you can use in any subsequent step.

To use the agent response in a downstream action:

1. Select the action where you want to use the response (for example, a Send an email or Update a row action).

2. In the field where you want to insert the response, open the dynamic content picker.

3. Look for the output from the Run an agent step and select agent response.

Common ways to use the agent response:

• Insert the response into the body of an email or a message.
• Define a condition to branch your agent flow based on what the agent returned.
• Save the response to a variable or data source for later reference.

Related content

• Edit and manage your agent flow in the designer
• Use prompts to make your agent or agent flow perform specific tasks
• Agent flows overview