Generative AI is shaping the future of work, and Microsoft Copilot is at the forefront, seamlessly integrating with the Microsoft 365 suite. By leveraging AI-driven capabilities, Copilot empowers users to maximize their productivity and fully unlock the potential of their Microsoft 365 subscriptions. This innovation not only transforms workflows but also redefines the way we collaborate and create in the modern workplace.
Creating Power Automate flows with MS Copilot

Generative AI is paving the way for the future, and Microsoft Copilot is leading the charge with its seamless integration into the Microsoft 365 suite. Copilot empowers users to maximise the benefits of their Microsoft 365 subscriptions.
Copilot has many use cases, some of us are using Copilot to summarise our meetings, find relevant documentation, scheduling meetings etc. I’m using Copilot to help rephrase my wording as I write this blog!
In my day-to-day work as a Data and Analytics consultant, I’ve been using Copilot to create Power Automate flows. The “conversation first” experience extends our ability to ask Copilot in natural language to create our automation or flow based on a use case or a scenario e.g., refresh Power BI dataflows and semantic models based on various events. One limitation in Power BI scheduling is that a dataflow or dataset can only be refreshed every 30 minutes. This means that if a semantic model relies on a dataflow, we might have to wait up to 30 minutes to refresh the semantic model after the dataflow is updated. To bypass this restriction, we can use Power Automate flows to trigger the refresh of Power BI datasets and dataflows more regularly.
How to use Copilot in Power Automate?
- Sign-in to Power Automate with your Microsoft 365 account.
- Navigate to home page -> Copilot experience.
- There are two ways to create our flow/automation:
- Describe in detail how automation needs to work in natural language.
- Chat with Copilot to create automation/flow step by step.
- Click on Generate when ready and Copilot begins to depict the structure. Click Next to verify the connections, configure the required settings and rectify if needed.
Best Practices to create Power Automate flow using Copilot.
- When you describe the scenario in Natural language, try to be very specific instead of using generic instruction.
- Provide prompts in When … Do or If…then…else… format.
- Mention connector if possible.
Example: Refresh Power BI dataflow when email arrives.
As mentioned earlier, I am using Power Automate flows to refresh dataflows in Power BI immediately using different events depending on the use case to overcome the limitations of scheduling in Power BI.
The screenshot below shows that I created a Power Automate flow to refresh a dataflow when an email with specific subject from a user arrives. If the dataflow refresh fails, then I will notify the support team.
After that, configure the actions and automation is ready to use!
Conversation first experience
Using the conversation first experience, a developer can “chat” with Copilot to create Power Automate flow in an effective way and Copilot creates the actions as instructed.
I found ‘Chat with Copilot to create automation’ very easy to use. It not perfect, but you can rectify actions if Copilot suggests a wrong action and at the same time, configure the connections/actions.
Total Time Saved
As we are using Power Automate flows in my project to refresh dataflows, dependent datasets, I saved approximately 3 hours in a week by using Microsoft Copilot into creating Power Automate flow.
Task | Traditional Approach | With Copilot | Time Saved |
Creating Power Automate flows | 6 hours | 3 hours | 3 hours |
Conclusion
As Copilot is deeply integrated into Microsoft’s Power Platform, it opens up new possibilities for automation. Using natural language in Copilot in Power Automate helped me discover we I can achieve more, create, optimise, and structure accurate automation quickly. With AI-assisted Power Automate flow generation users can achieve their goals more efficiently with low/no code knowledge.
Author