Introduction to Copilot Studio

• How Agents are different from Traditional Chatbots
• Power Platform Tools
• Copilot Ecosystem (M365 Copilot, Power Platform, Azure AI)
• Set up Copilot Studio in your tenant – Licensing, Free trail, Azure pay as you go etc.
• Tour to Copilot Studio Interface

Fundamentals of AI

• Large Language Model (LLM) – Tokens, Context-Window, Temperature
• AI Models available in Copilot Studio
• Grounding, Hallucination and Non-Determinism
• RAG – Embeddings & Vector Search fundamentals)
• Responsible AI Principles for Agent Design
• Prompt Engineering – Anatomy of good prompt

Copilot Studio Harness

• Standard, GitHub Copilot and Copilot Chat Harness – Build first Agent
• Agent Name, Description, Instruction, Knowledge, Skills etc.
• Why good Instruction is important

Knowledge Source (RAG)

• Adding knowledge from public websites, SharePoint, uploaded files, and Dataverse
• How generative answers work and how the agent grounds its responses (Grounding)
• Tips for preparing documents so the agent can use them well

Topic & Conversation design

• What topics are, and how trigger phrases compare with generative orchestration
• System topics: Greeting, Escalate, Fallback, and Conversation Start
• Nodes: Message, Question, Condition, Variable, and Redirect
• Adaptive cards
• Principles of good conversation design

Actions, Tools and Connectors

• Adding tools such as prebuilt connectors (Outlook, Teams, SharePoint, Excel)
• Agent flows and Power Automate flows called from an agent
• Passing inputs and outputs between the agent and a flow
• A quick introduction to prompts (AI Builder) and MCP servers

Skills (New Experience)

• What a skill is, and how skills differ from instructions and topics
• Writing a skill in the editor, and uploading .md skill files
• The structure of a good skill: purpose, when to use it, steps, output format, and guardrails
• Building a reusable skill library shared across agents
• Importing GitHub Copilot or Claude Code skills
• Using a skill to guide a conversation, the new-experience replacement for simple topics

Autonomous Agents and Triggers

• The difference between conversational and autonomous agents
• Event triggers, such as a new email, a new file, or a schedule
• Guardrails, human review, and keeping actions safe

Workflows (New Experience)

• • The new visual workflow designer
  • Triggers, including scheduled runs, new email, new file, and item created
  • Actions, conditions, AI steps, and approvals
  • Adding an agent as a node inside a workflow
  • Autonomous agents built as a trigger in a workflow that calls an agent
  • Calling a workflow as a tool from an agent
  • How workflows relate to Power Automate cloud flows and the older agent flows
  • Multi-Agent Orchestration
    Use case: A Service Desk Orchestrator that routes requests to an IT agent, an HR agent, or a Facilities agent.

Publishing and Channels

• Publishing an agent
• Deploying to Microsoft Teams, Microsoft 365 Copilot, and websites
• Authentication options at a basic level
• Sharing the agent and managing access

Analytics, Governance and Best Practices

• The analytics dashboard
• Managing Credits consumed by an Agent
• Basics of the Data Loss Prevention (DLP) policies that admins set
• Solutions and moving an agent from dev to production (overview)
• Responsible AI considerations