Exercise 06: Use generative AI orchestration to interact with your connectors
Goals for this lab
After this lab you will be able to:
- Understand the basics of plugin actions
- Use Copilot Studio to request data from another data source using plugin actions in a basic use case (using the MSN Weather Connector) and return the data in a conversational dialog with a customer or user
Generative AI orchestration
By default, a copilot responds to users by triggering the topic whose trigger phrases match most closely the user’s query, and it fills topic inputs from the conversation context. You can configure your copilot to use generative AI to choose not only from topics you created, but also from actions you added to extend the copilot.
In generative mode, a copilot can fill topic inputs, not only from the conversation context but also by generating questions to prompt the user for values. To learn more about this behavior and how to manage it, see Manage topic inputs and outputs.
Using generative AI to determine how your copilot responds can make the conversation more natural and fluid for the users. When a user sends a message, your copilot selects one or more actions or topics to prepare its response. Multiple factors determine the selection. The most important factor is the description of the topics and actions. Other factors include the name of a topic or actions, any input or output parameters, and their names and descriptions. Descriptions make it possible for your copilot to be more accurate when it associates a user’s intent with actions and topics.
In generative mode, a copilot can select multiple actions or topics at once, to handle multi-intent queries. Once actions and topics are selected, the copilot generates a plan that determines their execution order.
When you test a copilot that uses generative mode in Copilot Studio, you can open the conversation map to follow the execution of the plan.