Introduction
Agents built in Microsoft Copilot Studio can answer questions, retrieve information, and take action — but hand-authoring a dialog path for every possible question doesn't scale. The generative answers capability lets you connect an agent to knowledge sources and have AI generate responses on the fly, so you can focus on what the agent should know rather than every conversation branch it might follow.
Note
Microsoft Copilot Studio has introduced a new experience with updated features, capabilities, and navigation. This module is based on the classic experience. For more information, see Classic vs. new agent experience.
Scenario
Woodgrove Bank is building a client advisor agent to support relationship managers who field questions throughout the day. Clients ask about investment products, request regulatory compliance information, and submit documents for classification, all through the same agent. The team quickly discovered that scripting dialog flows for hundreds of product and compliance questions was unsustainable. They needed an approach that could draw answers from existing knowledge: structured databases, document libraries, and live data feeds, surfaced in a professional, consistently formatted way.
This scenario runs throughout the module, with each unit introducing a capability the Woodgrove Bank team uses to make their advisor agent more capable and easier to maintain.
What you'll learn
You'll explore the generative answers node and how to connect it to knowledge sources, including custom data via HTTP requests and flows. You'll also learn how to configure custom instructions to shape the AI's responses, and how to author custom prompts with models from the Microsoft Foundry model catalog.
Goal
By the end of this module, you'll be able to configure a generative answers node, connect diverse knowledge sources, shape AI output with custom instructions, and author custom prompts powered by models of your choosing.