Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
[This article is prerelease documentation and is subject to change.]
Use sub-agents to let your primary agent delegate a focused intent, such as billing or technical support, to a specialist agent that's built directly inside the same orchestration.
Tip
Before you add a sub-agent, review best practices for working with sub-agents, including how to keep sub-agents mutually exclusive and how to reference them from other agents' instructions.
Prerequisites
- A primary agent created with the Customer Conversation harness. See Create a new agent (Customer Conversation).
Add a sub-agent
Sign in to Copilot Studio.
Open the primary agent you want to add a sub-agent to.
On the Build tab, select Add a sub-agent.
In the Name field, enter a name for the sub-agent (up to 42 characters).
In the Description field, describe the intent this sub-agent handles (up to 1,024 characters). Make the description specific enough that the orchestrator can distinguish it from your other sub-agents and connected agents.
In the Instructions section, describe how the sub-agent should behave, including its role, tone, and any rules or guidelines it must follow (up to 20,000 characters).
Add the skills, tools, and knowledge this sub-agent needs to handle its intent.
If the sub-agent needs data from the primary agent, or needs to return data to it, configure input and output variables. Mark a variable as required if the sub-agent can't run without it - the orchestrator then collects that value before routing to the sub-agent. For output, choose whether the sub-agent's response is returned verbatim (passed back to the user exactly as the sub-agent generated it) or summarized (condensed by the primary agent before it's shown to the user).
Save the sub-agent.
Test routing to a sub-agent
On the agent navigation bar, select the Preview tab.
Send a message that matches the sub-agent's intent.
Verify that the response reflects the sub-agent's instructions, skills, tools, and knowledge.
Send a message outside the sub-agent's intent to confirm the primary agent or another sub-agent still handles it.