| image | https://learn.microsoft.com/en-us/media/learn/credential/badges/applied-skill.svg?branch=main |
|---|---|
| tags | PL-7008, Reference |
| GA | G-DXYJBX6BH8 |
This course teaches you how to build custom agents in Microsoft Copilot Studio. You'll create and configure an agent, design topics and conversation flows, ground the agent with knowledge sources, and add structured automation with agent flows and tools — then publish the agent to a channel such as Microsoft Teams.
:::success Date: 20260831 Course ID: 108948 :::
:::info Post Course Survey: https://aka.ms/pl7008survey :::
Course PL-7008 English version Course PL-7008 Simplified Chinese version
ESI Labs :::success Training key: 3C31243543A9939E ::: :::info Only need to redeem once Valid for 6 months :::
PL-7008 Labs PL-7008 Lab files
Switch between standard harness and GitHub Copilot harness agents
Get started with Microsoft Copilot Studio
Quickstart: Create and deploy an agent with the standard harness
Work with Power Platform environments in Copilot Studio
Key concepts - Publish and deploy your agent
Design agent conversations using topics
Follow topic authoring best practices
Design effective trigger phrases
Natural language understanding (NLU) overview
Use entities and slot filling in agents
Implement slot-filling best practices
Orchestrate agent behavior with generative AI
Build intelligent agents in Microsoft Copilot Studio
Add knowledge to an existing agent
Add Power Platform connectors as knowledge (preview)
Compare Copilot connectors and Power Platform connectors as knowledge sources
Add structured automation to agents in Microsoft Copilot Studio
Use agent flows with your agent
Create an agent flow as a tool
Use connectors in Copilot Studio agents
| No. | Name | Module | Link |
|---|---|---|---|
| 1 | Build autonomous agents in Copilot Studio — your MCPs, your models & multi-agent | Foundations | youtu.be/OmJAtDFcSZY |
| 2 | Build an initial agent with Microsoft Copilot Studio PL-7008 — Episode 1 | M01 | youtu.be/hzN2-K-8PP0 |
| 3 | Deploy an agent to Teams PL-7008 — Episode 6 | M01 | youtu.be/aW2e4_EoDxs |
| 4 | Manage topics in Microsoft Copilot Studio PL-7008 — Episode 2 | M02 | youtu.be/IOXMxuL9NrI |
| 5 | Work with entities and variables in Microsoft Copilot Studio PL-7008 — Episode 3 | M02 | youtu.be/YANyM1hWxlo |
| 6 | Deep dive into Generative Orchestration — Part 1 — Power CAT AI Webinars | M02 | youtu.be/Vvvyxy-zeRg |
| 7 | Add knowledge to your copilots | M03 | youtu.be/DKd5sb-iikE |
| 8 | Agent Flows: Build Agentic Automations You Can Trust | M04 | youtu.be/ewtKKEkNdr4 |
# Create agents in Microsoft Copilot Studio
## Foundations
- What is an agent: reasoning + actions grounded in [Copilot Studio](https://learn.microsoft.com/en-us/microsoft-copilot-studio/fundamentals-what-is-copilot-studio)
- **Standard harness** vs **GitHub Copilot harness**: rule-based, predictable topics vs reasoning-heavy, multi-step planning — see [harnesses overview](https://learn.microsoft.com/en-us/microsoft-copilot-studio/harnesses-overview)
- Experience is *where* you author; harness is *how* the agent runs — no migration path between them; [switch between them](https://learn.microsoft.com/en-us/microsoft-copilot-studio/agents-experience/switch-experiences) at creation time only
- Responsible AI pillars: fairness, reliability & safety, privacy & security, inclusiveness, transparency, accountability — [Responsible AI FAQs](https://learn.microsoft.com/en-us/microsoft-copilot-studio/responsible-ai-overview)
## M01 - Get started with Microsoft Copilot Studio
- [Create and delete agents](https://learn.microsoft.com/en-us/microsoft-copilot-studio/authoring-first-bot?tabs=web); [quickstart](https://learn.microsoft.com/en-us/microsoft-copilot-studio/fundamentals-get-started) builds an agent on the standard harness
- Agents live in [environments](https://learn.microsoft.com/en-us/microsoft-copilot-studio/environments-first-run-experience); [publish](https://learn.microsoft.com/en-us/microsoft-copilot-studio/publication-fundamentals-publish-channels?tabs=web) to Teams, Power Apps, and other channels
- **No authentication** vs **Authenticate with Microsoft** vs **Authenticate manually** — see [security and governance](https://learn.microsoft.com/en-us/microsoft-copilot-studio/security-and-governance)
- **Labs**: Lab 00 environment setup, Lab 01 create agents
## M02 - Design agent conversations using topics
- Topics are reusable conversation flows; every agent has built-in **system topics**, makers add **custom topics** — [create and edit topics](https://learn.microsoft.com/en-us/microsoft-copilot-studio/authoring-create-edit-topics?tabs=webApp)
- **Generative orchestration** (agent chooses topics/tools/knowledge by name and description, builds a plan) vs **classic orchestration** (one topic selected by trigger phrase, tools called explicitly) — see [generative AI orchestration](https://learn.microsoft.com/en-us/microsoft-copilot-studio/advanced-generative-actions)
- [Trigger phrases](https://learn.microsoft.com/en-us/microsoft-copilot-studio/guidance/trigger-phrases-best-practices) and NLU drive classic-mode topic selection
- Entities (prebuilt, closed list, regex) and variables (topic-scoped vs [global](https://learn.microsoft.com/en-us/microsoft-copilot-studio/authoring-variables-bot?tabs=webApp)) fill [slots](https://learn.microsoft.com/en-us/microsoft-copilot-studio/advanced-entities-slot-filling)
- **Labs**: Lab 02 manage topics
## M03 - Build intelligent agents in Microsoft Copilot Studio
- Knowledge sources: file uploads, SharePoint/OneDrive, public websites, Dataverse, Azure AI Search, and more — [summary](https://learn.microsoft.com/en-us/microsoft-copilot-studio/knowledge-copilot-studio)
- **Specific knowledge sources** vs **general knowledge** (ungrounded model responses) shape how an agent answers
- **Copilot connectors** (indexed, reusable semantic grounding) vs **Power Platform connectors** (real-time API access with no data replication) — [compare the knowledge-source options](https://learn.microsoft.com/en-us/microsoft-copilot-studio/knowledge-graph-vs-power-platform-connectors)
- [Add knowledge to an existing agent](https://learn.microsoft.com/en-us/microsoft-copilot-studio/knowledge-add-existing-copilot); [real-time connector knowledge](https://learn.microsoft.com/en-us/microsoft-copilot-studio/knowledge-real-time-connectors); [generative answers node](https://learn.microsoft.com/en-us/microsoft-copilot-studio/nlu-boost-node)
- **Labs**: Lab 03 manage knowledge
## M04 - Add structured automation to agents in Microsoft Copilot Studio
- **Agent flows** (deterministic, predefined action sequences, like Power Automate cloud flows) vs **connector tools** (direct one-shot calls to a service) — see [agent flows overview](https://learn.microsoft.com/en-us/microsoft-copilot-studio/flows-overview) and [connectors as tools](https://learn.microsoft.com/en-us/microsoft-copilot-studio/advanced-connectors)
- Use the [connector reference](https://learn.microsoft.com/en-us/connectors/connector-reference/) to check connector availability, tier, release status, supported products, and publisher
- Agent flows use dynamic content, expressions, conditions, and loops (apply for each / do until)
- [Create an agent flow as a tool](https://learn.microsoft.com/en-us/microsoft-copilot-studio/advanced-flow-create); [use agent flows with your agent](https://learn.microsoft.com/en-us/microsoft-copilot-studio/advanced-flow)
- **Labs**: Lab 04 use workflows as tools, Lab 05 (optional) new Copilot Studio experience
Microsoft Applied Skills: Build an agent in Microsoft Copilot Studio
- Money Yu
- Mail: Money.Yu@microsoft.com
- LinkedIn: @abc12207