Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 5 additions & 5 deletions _labs/agent-builder-m365.md
Original file line number Diff line number Diff line change
Expand Up @@ -183,7 +183,7 @@ Create, configure, and test a web-based Copilot agent that serves as a knowledge
>
> ![alt text](images/correct-copilot-page.png)
>
> - Make sure the **logged in user** is the fictitious one used in the lab. If you need your normal work user account, **select** the name and toggle to the fictitious user account.
> - Make sure the **logged in user** is the fictitious one used in the lab. If you see your normal work user account, **select** the name and toggle to the fictitious user account.
>
> ![alt text](images/logged-in-user.png)

Expand Down Expand Up @@ -245,7 +245,7 @@ Create, configure, and test a web-based Copilot agent that serves as a knowledge

#### Create your learning assistant agent

6. On the left side navigation pane, select **Agents** to open the **Agent Store**, then select **Create agent** (or the **New agent** card)
6. On the left side navigation pane, select **New agent**.


7. Notice that you can explore existing available templates. For this lab, paste the following into the **Describe the agent you want to create** input area at the top of the form (the describe textbox is the primary input — there is no separate tab to select), and then select Send.
Expand Down Expand Up @@ -297,9 +297,9 @@ Create, configure, and test a web-based Copilot agent that serves as a knowledge

13. Fix any issue like max character limit for starter prompt titles.

14. You can test your agent in the test pane. When ready, select **Create** in the upper right corner to finish creating your agent.
14. You can test your agent in the **Preview** pane. When ready, select **Create** in the upper right corner to finish creating your agent.

![Agent Builder test pane](images/agent-builder.png)
![Agent Builder Preview pane](images/agent-builder.png)

#### Share and test your agent

Expand Down Expand Up @@ -408,7 +408,7 @@ Build a sophisticated Sales Admin Assistant that integrates organizational data

5. Return to [Microsoft 365 Copilot](https://m365.cloud.microsoft/chat/?auth=2&home=1).

6. On the left side pane, select **Agents** to open the **Agent Store**, then select **Create agent** (or the **New agent** card).
6. On the left side pane, select **New agent**.

7. In the **Describe the agent you want to create** textbox at the top of the new agent page, paste the following prompt and press Send (or Enter). The describe textbox is the primary input — there is no separate tab to select.

Expand Down
18 changes: 7 additions & 11 deletions _labs/agent-builder-sharepoint.md
Original file line number Diff line number Diff line change
Expand Up @@ -124,17 +124,15 @@ Build a specialized agent with SharePoint integration and advanced AI capabiliti

1. Navigate to [Microsoft 365 Copilot Chat](https://m365.cloud.microsoft/chat/?auth=2&home=1) and go to the **Chat** tab.

2. On the side pane, expand **Agents** and select **Create agent**.
2. On the side pane select **New agent**.

3. When prompted to **describe the agent**, use this specific prompt:
3. In the **describe the agent** chat, use this specific prompt:

```
You are a Sales Admin Assistant. Your job is to help sales managers track revenue and identify trends across product lines. You understand product hierarchies, time periods (e.g. quarters, fiscal years), and sales metrics. Users can ask questions like 'Graph the sales for the last 2 years with a breakdown per product line and quarter'. You always respond in a friendly and professional tone, aiming to be helpful and insightful.
```

4. **Confirm** the suggested agent name when prompted.

5. Go to the **Configure** tab
4. Go to the **Configure** tab.

<!--
![alt text](images/knowledge-files.png)
Expand All @@ -146,12 +144,12 @@ Commented out; screenshots need update due to CPS updates. Last minute change t
-->

6. Under **capabilities**, enable:
- **Code interpreter** (for data analysis and chart generation)
- **Image generator** (for creating visual content)
- **Create documents, charts, and code** (for data analysis and chart generation)
- **Create images** (for creating visual content)

#### Finalize configuration

6. Navigate to the **Configure** tab to review and refine:
6. Review and refine the following:
- Agent name and description
- Instructions
- Starter prompts
Expand Down Expand Up @@ -228,9 +226,7 @@ Design a professional badge for the first place winner of our 2026 sales contest

8. Select **...** next to your agent name in the side bar navigation, and choose **Edit**

9. Select the Sales agent you created

10. In the **Configure** tab, explore additional **knowledge** capabilities:
9. In the **Configure** tab, explore additional **knowledge** capabilities:
- All **Websites** (or specific websites)
- All **SharePoint** data (or specific files or sites)
- All **Teams** data (or data from specific group chats, meetings, or channels)
Expand Down
52 changes: 26 additions & 26 deletions _labs/ask-me-anything.md
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
---
module: ask-me-anything
title: "Create an 'Ask me anything' agent for your employees"
description: "Empower employees with answers that matter – Build an intelligent agent that connects to your knowledge and data."
description: "Build an agent that answers employee questions using your organization's knowledge and data."
order: 13
duration: 40
difficulty: 200
Expand Down Expand Up @@ -159,7 +159,7 @@ Create a new agent with SharePoint knowledge integration that provides grounded,

### Step-by-step instructions

1. Navigate to the Copilot Studio home page at [copilotstudio.microsoft.com](https://copilotstudio.microsoft.com/)
1. Go to the Copilot Studio home page at [copilotstudio.microsoft.com](https://copilotstudio.microsoft.com/)

2. Go to the Solutions menu (located in the left-hand menu under the ellipsis …)

Expand All @@ -175,14 +175,14 @@ Create a new agent with SharePoint knowledge integration that provides grounded,

8. Go to **Knowledge**

![alt text](images/knowledge-sharepoint.png)
![Copilot Studio "Add knowledge" dialog with SharePoint option highlighted](images/knowledge-sharepoint.png)

9. Select **SharePoint**, and paste the SharePoint Knowledge URL and select **Add**.

> [!IMPORTANT]
> * For configuration, use the provided values in the **Lab Resources** (specific per training).

10. Choose **Add**
10. Select **Add**

> [!TIP]
> - If an error appears after adding the SharePoint URL, click "Browse items" and then "Confirm selection".
Expand Down Expand Up @@ -291,7 +291,7 @@ Make the agent available in Microsoft 365 Copilot for your users.

8. Select **See agent in Microsoft 365**

![alt text](images/m365-copilot-channel.png)
![Copilot Studio agent published to Teams and Microsoft 365 Copilot channel successfully](images/m365-copilot-channel.png)

9. In the **Microsoft 365 Copilot** experience, the agent description will pop up. Select **Add**

Expand Down Expand Up @@ -355,17 +355,17 @@ Let's start by asking for the user country at the beginning of the conversation.

<!-- replacing with instrucftions on how to do this on message received, so it works in M365 channel too

2. Now, in your 'Ask Me Anything' agent, go to the **Topics** tab, and then to the **System topics** tab.
2. In your 'Ask Me Anything' agent, go to the **Topics** tab, and then to the **System topics** tab.

3. Open the **Conversation Start** topic

4. At the beginning of the topic, right after the **Trigger**, add a new node **(+)** and choose **Add a tool**.

5. Go the **Connector** tab, and search for `Microsoft Entra ID Get user`
5. Go to the **Connector** tab, and search for `Microsoft Entra ID Get user`

6. Select the **Get user** connector action and **Submit**.

![alt text](images/add-tool-connector.png)
![Get user Microsoft Entra ID connector tool in Copilot Studio flow editor](images/add-tool-connector.png)

> [!IMPORTANT]
> - If you don't see the **Get user** action, select another of the Microsoft Entra ID actions, such as **Create user** and **Submit**.
Expand All @@ -378,9 +378,9 @@ Let's start by asking for the user country at the beginning of the conversation.

8. Toggle the **End user authentication** to **Agent author authentication**.

9. In the **Inputs** tab, for **User Id or Principal Name**, select `...`, go the the **System** tab, and select **User.Id**.
9. In the **Inputs** tab, for **User Id or Principal Name**, select `...`, go to the **System** tab, and select **User.Id**.

![alt text](images/select-user-id.png)
![Microsoft Copilot Studio Get user action with User.Id system variable selected as input](images/select-user-id.png)

10. **Save**

Expand All @@ -392,7 +392,7 @@ Let's start by asking for the user country at the beginning of the conversation.

4. At the beginning of the topic, right after the **Trigger**, add a new node **(+)** and choose **Add a tool**.

5. Go the **Connector** tab, and search for `Microsoft Entra ID Get user`
5. Go to the **Connector** tab, and search for `Microsoft Entra ID Get user`

6. Select the **Get user** connector action and **Submit**.

Expand All @@ -402,7 +402,7 @@ Let's start by asking for the user country at the beginning of the conversation.

8. Toggle the **End user authentication** to **Maker-provided credentials**.

9. In the **Inputs** tab, for **User Id or Principal Name**, select `...`, go the the **System** tab, and select **User.Id**.
9. In the **Inputs** tab, for **User Id or Principal Name**, select `...`, go to the **System** tab, and select **User.Id**.

![alt text](images/select-user-id.png)

Expand All @@ -414,9 +414,9 @@ Let's start by asking for the user country at the beginning of the conversation.
Hello!
```

15. Navigate to the **Topics** tab, open the **Conversation Starts** topic, and to see what values are returned from the connector, enter the **Variables** pane in the topic command bar. Go the **Test** to see the variable values at runtime (i.e., as you are testing in the conversation test pane).
15. Go to the **Topics** tab, open the **Conversation Starts** topic, and to see what values are returned from the connector, enter the **Variables** pane in the topic command bar. Go the **Test** to see the variable values at runtime (i.e., as you are testing in the conversation test pane).

![alt text](images/variables-pane.png)
![Variables panel open with Test tab selected, showing Topic variable GetUser of type record](images/variables-pane.png)

16. Select **Value** under the **GetUser** record, and **expand** it to see if the expected value was returned.

Expand Down Expand Up @@ -468,7 +468,7 @@ Let's start by asking for the user country at the beginning of the conversation.
> - Use the formula provided in the **Lab Resources** (specific per training).
> - Notice how the formula concatenates the SharePoint URL with the office location, effectively using context to target the correct source and location for the content.

![alt text](images/sharepoint-formulas.png)
![Copilot Studio topic editor configuring SharePoint data source with a dynamic formula using office location variable](images/sharepoint-formulas.png)

27. Select **Insert** once done, then **Save** your topic

Expand Down Expand Up @@ -496,7 +496,7 @@ Let's start by asking for the user country at the beginning of the conversation.

32. We need to set the variable value in-between the parenthesis. **Type** `/` and select **Power Fx**.

![alt text](images/instructions-formula.png)
![Power Fx option highlighted in Instructions panel dropdown menu](images/instructions-formula.png)

33. **Type** `Global.GetUser.officeLocation` and select **Insert**

Expand Down Expand Up @@ -554,7 +554,7 @@ This use case demonstrates **agent scoping** principles:

2. Go to the **Agents** tab, and **+ Add an agent**

![alt text](images/add-agent.png)
![Choose how to extend your agent dialog with 'Create an agent' option selected](images/add-agent.png)

3. Update
- Its **Name** to: `IT Support Agent`
Expand All @@ -575,7 +575,7 @@ This use case demonstrates **agent scoping** principles:

5. Select **ServiceNow**

6. Choose the **ServiceNow** connection under **Your connections**, and **Next**
6. Select the **ServiceNow** connection under **Your connections**, and **Next**

8. Select **Knowledge** as the knowledge source.

Expand Down Expand Up @@ -603,9 +603,9 @@ This use case demonstrates **agent scoping** principles:
> [!IMPORTANT]
> In production scenarios, you may want to use the user context when making the connection to ServiceNow. Here, your context (as the author) is used by end-users of your agent when searching for incidents.

8. For **Record Type**, under **Fill using**, choose **Custom value** and choose `Incident`
8. For **Record Type**, under **Fill using**, select **Custom value** and select `Incident`

9. Select **+ Add input** and choose **Query**
9. Select **+ Add input** and select **Query**

10. Next to **Dynamically fill with AI**, select **Customize** and use this for **Description**:

Expand All @@ -619,7 +619,7 @@ This use case demonstrates **agent scoping** principles:

11. Select **+ Add input** and choose **Limit**

12. Under **Fill using**, choose **Custom value** and set **1**
12. Under **Fill using**, select **Custom value** and set **1**

13. Select **Save**

Expand All @@ -639,7 +639,7 @@ This use case demonstrates **agent scoping** principles:
> - Creates user-friendly responses with proper formatting
> - Combines ServiceNow data with other knowledge sources when needed

![alt text](images/multi-agent-servicenow.png)
![Copilot Studio activity map showing IT Support Agent handling USB troubleshooting and ServiceNow ticket query](images/multi-agent-servicenow.png)

29. **Observe** how your agent now demonstrates scoped capabilities:
- **Tool Selection**: Automatically chooses the right ServiceNow tool based on user intent
Expand Down Expand Up @@ -703,7 +703,7 @@ In this section, you'll configure any third-party knowledge to enrich knowledge

5. Add a new node > **Advanced**, **Send HTTP request**

6. For **URL**, select `...`, go the **Environment** tab, and choose the **Custom Knowledge Endpoint** variable you had created.
6. For **URL**, select `...`, go the **Environment** tab, and select the **Custom Knowledge Endpoint** variable you had created.

7. **Method**: `Post`

Expand Down Expand Up @@ -734,7 +734,7 @@ In this section, you'll configure any third-party knowledge to enrich knowledge

13. Under **Save response as**, select **Create a new variable**, select the variable name that was created (e.g., **Var1**) and rename it `KnowledgeResults`.

![alt text](images/http-request.png)
![HTTP Request action configured with POST method, JSON body using System.Activity.Text, saving results as KnowledgeResults table](images/http-request.png)

14. Add a new node, **Variable management** > **Set a variable value**

Expand Down Expand Up @@ -768,7 +768,7 @@ In this section, you'll configure any third-party knowledge to enrich knowledge
> Notice how the knowledge step returned results that contain citations to a fictitious knowledge base.


![alt text](images/custom-knowledge-source.png)
![Knowledge node response showing vendor update steps with 3 referenced sources highlighted](images/custom-knowledge-source.png)

---

Expand Down Expand Up @@ -855,7 +855,7 @@ In this section, you'll learn how to upload files and run files through a prompt

14. **Save** your prompt

![alt text](images/prompt.png)
![Copilot Studio AI prompt node "Meeting AI Notes" with GPT-4o model returning JSON meeting data](images/prompt.png)

14. **Save** the topic.

Expand Down
10 changes: 5 additions & 5 deletions _labs/autonomous-account-news.md
Original file line number Diff line number Diff line change
Expand Up @@ -161,7 +161,7 @@ Set up an autonomous agent with a recurring trigger that automatically activates

1. In the agent's **Overview** tab, scroll to the **Triggers** section.

> [!TIP]
> [!IMPORTANT]
> The triggers section may not be immediately available as it depends on background processes started when the agent is created. This should take no more than a minute to complete.

1. Select **Add trigger**, select **Recurrence**, and then select **Next**
Expand Down Expand Up @@ -215,7 +215,7 @@ Set up an autonomous agent with a recurring trigger that automatically activates

1. Search for `List rows` and select the **List rows from selected environment** action (under **Microsoft Dataverse**, displayed as *List rows from a table in a Power Platform environment*).

1. Choose an existing Dataverse connection or add a new one (select Oauth as authentication type).
1. Select an existing Dataverse connection or add a new one (select OAuth as authentication type).

> [!IMPORTANT]
> If you need to create the connection use OAuth and sign in with your workshop credentials. Your user requires permission to access Opportunity records, which is provided as part of the workshop.
Expand Down Expand Up @@ -671,7 +671,7 @@ Automate the final step: format relevant news into a clean, branded email for ac

1. Select **Insert**.

1. Save the **Conversation start** topic
1. Save the **Conversation start** topic.

1. Now that your agent has a preloaded HTML template, configure an Outlook email tool to deliver the report.

Expand Down Expand Up @@ -708,7 +708,7 @@ Automate the final step: format relevant news into a clean, branded email for ac

1. Now that you've created the report template and configured the email tool, you'll guide the agent to use them as part of its orchestration.

1. Go to your agent and go to the **Overview** tab.
1. Go to your agent and select the **Overview** tab.

1. In the **Instructions** section, add the following steps:

Expand All @@ -728,7 +728,7 @@ Automate the final step: format relevant news into a clean, branded email for ac
1. **Save** your agent's instructions. The instructions should appear as follows:
![Instructions After Topics](images/final-instructions.png)

1. Go to the **Triggers** section and Select **Test trigger** on the `Analyze Opportunities` trigger.
1. Go to the **Triggers** section and select **Test trigger** on the `Analyze Opportunities` trigger.

1. Once the agent finishes executing, verify:
- It retrieved opportunities from the Sales App
Expand Down
Loading
Loading