Hands-On: Building Intelligent Agents with IQ, Dataverse and Fabric

Ten minutes before the work: sign in once, meet the company, and check your data is there.

1. Module 0 · What you build
2. Attendee guide
3. Module 0 · Sign in
4. Module 0 · Your names for the day
5. Module 0 · Meet Apex Industrial Group
6. Module 0 · Check your data is ready

An agent that answers from tables alone, checked against SQL, then the same agent with the business's definitions.

1. Module 1 · What you build
2. Lab 1.1 · Build the agent with tables only
3. Lab 1.2 · Ask five questions and check them against SQL
4. Lab 1.3 · Ground the agent and ask again
5. Lab 1.4 · Publish

A GitHub Copilot agent that calls your data agent as a tool and takes its next steps from the procedures.

1. Module 2 · What you build
2. Lab 2.1 · Create the agent
3. Lab 2.2 · Connect your data agent as a tool
4. Lab 2.3 · Add the instructions
5. Lab 2.4 · Add the procedures as knowledge
6. Lab 2.5 · Test, then publish

The same data with a business vocabulary, so a question in Maya's words gets a ranked answer.

1. Module 3 · What you build
2. Lab 3.1 · Create the ontology
3. Lab 3.2 · Add the six entity types
4. Lab 3.3 · Add the six relationships
5. Lab 3.4 · Ask the Ontology agent

What the teams are saying in email and documents, read by an agent that cannot see the data.

1. Module 4 · What you build
2. Lab 4.1 · Put the content in place
3. Lab 4.2 · Check your environment
4. Lab 4.3 · Add Work IQ
5. Lab 4.4 · Add the instructions
6. Lab 4.5 · Check each source

Service cases in Dataverse, queried from your Copilot Studio agent through MCP.

1. Extra · What you build
2. Extra · Use it from your Copilot Studio agent

1. Check your results
2. Answers · Module 1 · Fabric data agent
3. Answers · Module 2 · Copilot Studio agent
4. Answers · Modules 3 and 4 · Fabric IQ and Work IQ
5. Wrap-up: five patterns and feedback
6. All workshop files