Summary
In this module, you explored how to transform your agent from a static responder into a dynamic, task-oriented assistant by combining agent flows, custom prompts, and orchestration.
You began by understanding the various tools and their uses in Copilot Studio, then built an agent flow for updating a record in the Machine Orders table.
To guide conversations with more precision, you built a custom prompt tool. This exercise demonstrated how to collect inputs from users, run those inputs through a structured prompt, and deliver customized, actionable responses.
Key takeaways
Orchestration mode defines how your agent makes decisions:
- Classic mode relies on explicitly triggered topics.
- Generative mode uses AI to select the best path—topics, tools, or knowledge—in real time.
Tools let agents perform tasks like fetching weather data or sending emails without writing manual input prompts.
Agent flows provide structured automation flows, ideal for instructing the agent to perform consistent, repeatable actions.
Custom prompts enable natural language generation for tailored responses based on user-provided input and prompt configurations.
With these tools, your agent can now perform real-world tasks, guide users through complex workflows, and adapt to a wide range of queries. In the next module, you'll expand your agent's reach by giving it autonomous features that trigger the agent to engage when specific conditions are met.
Next steps
These tools can perform repetitive tasks on behalf of your customers or internal staff. To further explore the tools, try a prompt tool. This feature is the ultimate tool to reword emails, automatically respond to complaints, reformat data, or translate information using a large language model. See more information about prompts in the Use a custom prompt documentation.