Exploring AI Prompts in Power Automate

[N__A]

The Power Platform empowers makers with prebuilt AI prompts for generating AI-driven responses in Power Automate flows and Power Apps, which can be utilized in common business scenarios such as summarizing, classifying, extracting data, translating, assessing sentiment, or drafting a reply.

How is AI Builder prompts in cloud flows different from Copilot?

The AI Builder prompts lets you use the GPT-3.5 Turbo model powered by Azure OpenAI Service directly in your Power Automate flows, by adding a prompt as an action, provide the text you want to analyze, and the AI generates the desired output. To learn more about prebuilt prompts, go to Available prebuilt prompts

Copilot in cloud flows is designed to help you create and edit Power Automate flows just by describing them in everyday language, providing helpful guidance along the way.

Now let’s explore AI Builder prompts in cloud flows in action

1. Sign in to Power Automate
2. Select +Create > Instant cloud flow.

3. Name the flow, select “Manually trigger a flow”, and then click “Create”.

4. In the designer, select “Manually trigger a flow”. This will open the Parameters tab on the left. Then, select “+Add an input”

5. Select Text, and then enter a name.

6. In the designer, select “+Insert a new step” after Manually trigger a flow, and then select “Add an action”

7. On the left pane in the Search field, enter GPT and then select the action “Create text with GPT using a prompt”.

8. On the left pane in the Prompt, choose one of the available prebuilt prompts that is suitable for your business scenario, or select ‘+New custom prompt’ to create your own custom prompt.

9. In the Input Text field, select “Text” from the Dynamic content list to pass the input text from the manual trigger to the prompt.

10. The ‘Create text with GPT using a prompt’ action generates a flow variable called ‘Text,’ which represents the output of the prompt generated by the GPT model. This variable can be utilized in subsequent actions, such as sending a Teams message or an email that incorporates the ‘Text’ variable.

11. After saving the flow, you will receive a warning stating, “The ‘Create text with GPT using a prompt’ action doesn’t have a content approval action after it.” This warning suggests adding a ‘Start and wait for an approval of text’ action following the ‘Create text with GPT using a prompt’ action. This step ensures that a human review the GPT-generated output to verify its quality.

12. To test your flow, select “Test” in the top right toolbar, choose “Manually,” and then click “Test.”

13. A “Run Flow” dialog will display where you enter the input parameter and then click “Run Flow.”

14. After the flow runs successfully, the GPT-generated output will be generated and sent.

I hope you found this blog helpful. Please don’t hesitate to share any comments or questions you may have 😊