Microsoft Foundry: Build an AI Knowledge Assistant with RAG

About this course

Build a complete AI assistant on Microsoft Foundry, starting with nothing but an Azure subscription and ending with an agent your users can open in Microsoft Teams.

Every IT help desk answers the same questions all day. The answers are in the policy documents, and nobody reads them. In this Microsoft Foundry course, you'll build an assistant that has.

Across more than three and a half hours of video instruction and twelve hands-on labs, you and I build a help desk knowledge assistant for a fictional company called Contoso, and we build it the way a real IT team would, with the governance, security, evaluation, and monitoring a production system needs. Nothing is theory for its own sake. Every concept shows up because the project needs it at that moment.

What you'll buildA Microsoft Foundry resource and project, with a deployed model and the quota behind itInstructions and parameters that shape how the model behaves, including output limits and reasoning effortA custom guardrail with a blocklist for information no content classifier could know about

Grounding on Contoso's policy documents, so the agent answers from them, cites them, and admits when it doesn't knowA Foundry IQ knowledge base backed by Azure AI Search, for knowledge at scale

Web search with citations, with routing instructions that control what leaves the Azure boundary

An evaluation run against a real test dataset, and the skill of reading the results like an engineer

Security groups and least-privilege roles for the people who build the agent and the people who use itTracing, a monitoring dashboard, and a budget with a cost alert

The finished agent published to Microsoft Teams and Microsoft 365 Copilot, and used from the Teams client

How the course works

Read the full description on Udemy on Udemy (opens in a new tab)

Prices, discounts and availability are set by Udemy. We may earn a commission when you purchase through links on this site.