A Tour of Copilot Studio

This guide offers a comprehensive walkthrough of Copilot Studio, empowering users to build custom agents with advanced capabilities such as knowledge integration, variable management, and analytics tracking. By exploring features like conversation management, user authentication, and testing tools, users can optimize their agents for improved interaction and user satisfaction. Overall, this guide is essential for anyone looking to leverage Microsoft Copilot Studio effectively.

Damien Bird
|
59 steps
|
11 minutes

Damobird365

**Disclaimer**\ \ This tutorial has been shared as a preview to the complete [Building a Custom Agent in Copilot Studio](https://damobird365.teachable.com/p/damobird365-membership) course. Please make sure you sign up to the new community which includes both this course (launching in January) and a Power Automate Cloud Flow course due to be launched in June 2025. You'll also get access to the community forum and fortnightly calls with Damien where you can learn more about the Power Platform and even share your own builds and learning. \ \ You can check out his [YouTube DamoBird365 (15,000+ subscribers) here](https://www.youtube.com/c/DamoBird365?sub_confirmation=1)

Copilot Studio Walkthrough

Navigate to [https://copilotstudio.microsoft.com](https://copilotstudio.microsoft.com/environments/Default-16c901d1-9763-49b1-961c-6cd701f5d0f7/bots/f57625fa-c5b8-ef11-a72f-000d3ad4aa08/overview)

Here, we have the custom Agent that you will learn to build throughout the course. For now, we'll take a tour and explore how to navigate around Copilot Studio

Click here.

And finally the **Published Status**, if and when an Agent was last Published

Click "Knowledge"

Knowledge

Click "Knowledge"

By clicking on **Add knowledge**, you can attach various data sources that are natively available to our agent including public websites, SharePoint, Dataverse, Upload Files or use Advanced capabilites such as Azure AI Search

Topics, Authoring Canvas and Nodes

Click "Topics"

Click "Conversation Start"

Click here.

You can add nodes to your Topic by clicking on the **+** and then choosing from the following options: - **Send a message** to the user - **Ask a question** to the user and capture the response as a variable - **Use an adaptive card**, which is a platform-agnostic JSON payload for creating custom cards that capture a response from the user - **Add a condition** to determine an outcome based on branching logic

Via **Variable management** you can **set a variable value,** **parse a value** from custom data or **clear all variables**

From **Topic Management**, you can: - Navigate to another Topic - End the current Topic or all Topics - Transfer to an Agent - Go to a specific step in a Topic - End the conversation

From **Call an Action**, you can: - Call or create a Cloud **Flow** or AI Builder **Prompt** - Call a Bot Framework **Skill** or specific Power Platform **Action** from a Standard/Premium or Custom Connector

Under **Advanced**, you can add: - **Generative Answers**: Add this node to configure knowledge with granular knowledge sources - **Send an HTTP Request**: Call an open API endpoint - **Log Custom Telemetry**: For example, log a failure to Azure Application Insights - **Send an Event**: Enables the Agent channel to intercept events - **Send an Activity**: Enables the Agent to respond to various activity types - **Authenticate**: Enable user authentication based on authentication provider settings

Click here.

By clicking on the ellipses (**... More**), you can: - Access analytics for this Topic to review engagement. - Edit the Topic using a YAML **code editor**. - Reset the Topic to its default settings if custom changes have been made, provided this is a system Topic.

Actions

Under the **Actions** section, you can see any of the Power Platform connector actions that have been added specifically to our Agent. In this case, an action to **Get Items** from **SharePoint**

Click "Triggered by agent (preview)"

Activity

On the **Activity** tab (with **Orchestration** enabled), you can see the history of previous conversations either as specific users where they are authenticated or as an Autonomous Agent

You can see the history of previous interactions by clicking on one of these activities and explore the activity map.

Click this icon.

You can see that the Agent used Knowledge sources on SharePoint and Files uploaded to retrieve details about salary grades

You can also expand the **Activity map** to include a conversastion **Transcript**

The Agent was specifically asked the salary range for grade 9 and was able to respond, including a link to its source.

Analytics

Click "Analytics"

You can check the effectiveness of your sessions by clicking **See details** and then based on a user Abandoning, Escalating or ending their session as Resolved, which is noted when the **End of Conversation** system Topic is called and a Satisfaction Score is requested

The success of your Knowledge source use can also be explored clicking **See details** enabling you to visualling check the relevance to the data you have grounded your Agent on

Finally if a user has provided a feedback score upon completing a session, it will be noted here under **User feedback**

Channels

On the **Channels** tab, you can make your Agent accessible to users, whether they are internal employees or external customers. The availability of specific channels may depend on the authentication type or your organization's Data Loss Prevention (DLP) policy, which might restrict access to certain channels. This is the central location where you configure your agent for end-user availability.

Click here.

Click "Custom website"

Click "Dynamics 365 Customer Service"

Test your Agent and additional controls

Click "Test"

Click the "Ask a question or describe what you need" field.

From the test pane, clicking the ellipsis (**...**) provides access to various settings related to testing your Agent. These include options to test autonomous triggers, manage connections for different actions, flag issues with Copilot Studio, and enable debug mode to return additional information while testing your Agent.

At the very top of the pane, you'll find a **Publish** button to make changes made to your Agent live, along with a **Settings** button to access additional configuration options.

Click here.

Additional Settings

Let's take a look at the additional **Settings**

Similar to the Overview tab, you can configure the **Agent name** and change the **icon**

Click here.

You can **moderate the content** to adjust the bot's responses, allowing it to either be more creative or more precise based on the type of answers you're looking to surface. By fine-tuning these settings, you can strike the perfect balance between imagination and accuracy, tailoring the bot’s tone and focus to better match your needs.

Click here.

Click "Security"

By clicking on **Authentication**, you can view the default method using Entra ID, disable it entirely (which should be carefully considered), or integrate with third-party authentication

On the **Authoring Canvas** tab, you can adjust the behavior of how it renders while building

Click "Entities"

Click "Skills"

Click "Voice"

Click "Languages"

On the **Language Understanding** tab, an advanced capability, integrating a Conversational Language Understanding (CLU) model with Microsoft Copilot Studio enhances your Agent by enabling it to better understand and respond to user conversations. CLU is a custom version of Natural Language Understanding (NLU) designed specifically for conversational applications. This integration allows you to map CLU intents to trigger Agent dialogs, making interactions more natural and satisfying, and is especially useful for more complex conversational needs where specialized language or lingo is required.

Under **Component Collections**, reusable agent components such as topics, knowledge, actions, and entities can be shared across multiple agents within your environment. You can also export and import these collections to transfer content between different environments, making it easier to manage and reuse components across various applications.

Lastly, on the **Advanced** tab, you can connect your Agent to **Application Insights** in Azure to capture custom telemetry, access key **Metadata**, and even **view the solution** contained within your Power Platform environment. This is especially useful for those looking to migrate their Agent between Test, Dev, and Prod environments.

If you've enjoyed this tutorial, make sure you sign up to the new community which includes both this course (launching in January) and a Power Automate Cloud Flow course due to be launched in June 2025. You'll also get access to the community forum and fortnightly calls with Damien where you can learn more about the Power Platform and even share your own builds and learning. [Sign up here](https://damobird365.community.teachable.com/)