Call Power Automate Flows from Copilot Studio | Day 103
Learn how to call a Power Automate flow from Copilot Studio, pass a ticket ID, and return a useful response.
- Published
- Reading time
- 3 min read
Quick answer
Create a compatible agent flow, define its inputs and response, add it as an agent tool, and test the conversation and flow together.

Week 15 · Day 103 of 365 in 365 Days of Copilot Studio — view the full series
Before you start
Is this guide for you?
- Best entry point
- 365 Days of Copilot Studio
- Time investment
- 3 min read
- Prerequisite knowledge
- A Copilot Studio agent and access to Power Automate and an authorized test data source.
What you’ll learn
- An agent passes the required input to a compatible flow.
- The flow performs the business action and returns structured results.
- Test valid, missing and invalid inputs and verify user access.
- Plan for errors and supported response time limits.
Before you start
- Prerequisite knowledge
- A Copilot Studio agent and access to Power Automate and an authorized test data source.
On this page (6 sections)
Day 103 of 365 Days of Copilot Studio shows how an agent can call a Power Automate flow to perform a business task. Following Day 102, we will use an IT ticket lookup to follow the request from conversation to response.
What is a cloud flow?
A cloud flow automates tasks across connected applications and services. Depending on its connectors and permissions, it might send an email, create a SharePoint item, post a Teams message or update a Dataverse record. A compatible flow can be exposed to an agent as a tool so that the agent can invoke it when needed.
How the interaction works
- A user asks the agent to check a ticket.
- The agent identifies the request and gathers a ticket ID.
- The flow receives the ID and retrieves authorized ticket information.
- The flow returns the result, and the agent presents it to the user.
Create a flow for the agent
- Open Copilot Studio and select your agent.
- Go to Tools, add a tool and create an agent flow.
- Configure the supported When an agent calls the flow trigger and define a ticket ID input with a clear description.
- Add the actions needed to validate the ID and look up the matching record in your authorized data source.
- Use Respond to the agent to return the relevant status and other necessary fields.
- Save and publish the flow, configure it as an agent tool and test the interaction.
Interface names and available options can change. Check your environment, connector access, applicable licensing and the supported flow requirements for your configuration.
Example: look up an IT ticket
For a request such as “Check ticket INC-1024,” the agent passes INC-1024 to the flow. The flow searches an authorized SharePoint or Dataverse data source and prepares a result containing the ticket ID, status and assigned team. If the returned status is “In Progress,” the agent can report that status to the user.
Use a connection with appropriate permissions and ensure the requester is authorized to see the ticket. Do not assume that knowing a ticket ID grants access. Return only the information needed for the conversation.
Test the complete conversation
- Open the agent test panel and ask a question that requires the flow.
- Verify that the flow receives the expected ticket ID and inspect its run result.
- Check that the returned fields match the record and that the agent displays an accurate response.
- Repeat with a missing ID, invalid ID, nonexistent record, unauthorized user and unavailable data source.
Common mistakes to avoid
- Wrong trigger: use the supported trigger for an agent-called flow.
- Missing inputs: collect and validate required parameters before calling the flow.
- No response: configure a response with the values the agent needs, including a useful failure outcome.
- Permission issues: check authentication, connection identity and access to the underlying records.
- Slow execution: design synchronous responses within the supported time limit; use a separate follow-up for longer work.
Keep the flow focused, test with representative users and monitor failures after release. Day 104 continues with passing inputs and outputs between the agent and flows.
Tagged
AI Agents · Automation · Governance
Frequently asked questions
How does an agent start a flow?
The agent calls a configured flow tool and passes the required inputs to its supported trigger.
How does the result reach the user?
The flow returns outputs using its response action, and the agent uses those values in its answer.
What should I test?
Test valid, missing and invalid inputs, record access, connector failures, and the response shown to the user.
Have a Microsoft 365 topic idea?
Share article suggestions, community session ideas, corrections, or real-world scenarios for future nextM365 learning notes.
Keep learning Microsoft 365
Explore more practical guides for SharePoint, Power Platform, Copilot Studio, migration, automation, governance, and security.
Continue learning
Related tutorials
Related questions
Related comparisons
Next action