Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
Note
This legacy article only applies to classic chatbots created in Power Virtual Agents or in the Microsoft Copilot Studio app for Microsoft Teams.
If you created a classic chatbot by using Power Virtual Agents or an early version of the Copilot Studio app for Microsoft Teams, you can clone and convert it into an agent powered by the standard harness in Copilot Studio. This article explains migration eligibility, the cloning process, configuration considerations, and recommended testing before deployment. Cloning creates a new agent—it doesn't modify your existing classic chatbot.
A clone feature is available when you open a classic bot built using Power Virtual Agents or the Copilot Studio app for Microsoft Teams.
Eligibility
Classic chatbots that use any of these features aren't eligible to be cloned:
- Languages other than English
- Handoff to Omnichannel for Customer Service
- Knowledge Management extensions
Important considerations
This feature clones the selected classic bot, and creates a new agent powered by the standard harness. Be aware of the following important considerations in the cloning process:
- The cloning process is only available in the same language as the source classic bot.
- The cloning process only clones topics built using the Copilot Studio web canvas.
- The cloning process clones entities and associated synonyms and custom entities.
- An agent administrator needs to reconfigure authorization, channels, and security settings in the cloned agent.
- Any Bot Framework Skills from the classic bot need to be reconnected with the cloned agent.
- The connection to Power Automate Flows, built using the Copilot Studio web canvas, are cloned and associated within the corresponding topic in the new agent. However, you should test the connections.
- Consider any custom canvas and other custom architecture components you previously connected to your classic bot. You might need to reconnect or reconfigure these custom components to your new agent and test them. In addition, consider your entire architecture.
- You might need to update any automated deployment pipelines if they use the original classic bot ID or any reference to the classic bot.
- You should update any custom reporting to your new agent deployment. Examples include if you use the custom data pipeline solution or alternative reporting solutions.
Important
This list isn't exhaustive. Ensure you complete the required testing for your agent before deploying it to production.
By migrating a classic chatbot to Copilot Studio unified authoring, you can access features such as generative AI, variable management and Power Fx, code view, events, and more.
When you're ready to migrate your classic bot, go to Clone a classic chatbot.
What's different in the Copilot Studio authoring experience
The following overview isn't an exhaustive list of the differences in the authoring experience in Copilot Studio depending on whether you opened a classic chatbot or an agent powered by the standard harness. Learn more in Quickstart: Create and deploy an agent with the standard harness.
Agent creation
The creation process lets you name the agent, specify the agent's language, and also get started with boosted conversations.
You can also configure the icon, and select the default solution and default schema name. When you select the schema name, it directly uses the prefix from the selected solution.
User interface
User interface changes include:
- Topic creation: You can create a new topic manually or use Copilot.
- Topic triggers: Many types of triggers are available for topics, including triggers related to events, such as Message received.
- Topic nodes: More node types are available.
System topics
When you clone a classic chatbot, you get more system topics for your new agent: for example, Conversation Start and Multiple Topics Matched.
Conversation Start: Previously, classic bot makers had to customize the classic bot with extra code to automatically begin a conversation. This code is no longer required, and the Conversation Start system topic begins the conversation automatically.
Multiple Topics Matched: For conversational experiences, when a user asks a question, there can be more than one topic matched with high confidence. The multiple matches cause a disambiguation experience to occur, often referred to as a Did You Mean experience. The agent asks the user, "Did you mean X, or did you mean Y?" In Copilot Studio, you can tweak the disambiguation experience by using the Multiple Topics Matched system topic.
Pass entire objects by using Power Automate
Copilot Studio is seamlessly integrated with Power Automate to allow agents to use many different connectors to help retrieve and access data in other systems. Classic bots can only exchange text, number, and Boolean (yes/no) objects with Power Automate.
Copilot Studio agents can send more complex objects, such as Dataverse records, or collections of Dataverse records. Your new agent can use Parse value nodes and variables to extract the information from these objects, and format the result as required (for example, in a custom Adaptive Card).
Variables, variable watch window, and testing
A wider selection of variable-related nodes let you create new variables and update them, and utilize system data and variables. You can use Power Fx formulas to manipulate data and perform calculations at runtime. Learn more in Create expressions using Power Fx.
When creating conversational experiences, it's important to be able to test variables within a topic flow. For classic chatbots, the variable watch window is in the topic mini-map. For agents, this feature is available in the variable panel, on the side panel within the authoring canvas, under Test.
Write your agent in YAML
You can author conversational experiences that use both the graphical user interface (GUI) authoring tool and code view. The GUI uses rich responses, adaptive cards, and more. Developers can switch to the code view within a topic, in real time, to directly build or edit the underlying YAML code.
Events
Events introduce extra capabilities within the Copilot Studio web interface to create and manage the conversational experience. Traditionally, trigger phrases are used to detect the intent from a spoken sentence or word and the best matched topic, if found. Events are an alternative way to manage events that are either sent to or received from the user or alternative systems.
Generative AI features
Boosted conversations: You can create a useful agent in seconds by connecting it to a datastore, such as a public website, allowing the agent to generate answers by using that datastore. You can use this feature at agent creation or from the Knowledge page for your agent.
Topic creation with Copilot: Creating agents with AI is simplified by using Copilot in Copilot Studio. You can utilize Copilot when creating a new topic, entering a name and an initial prompt to build the topic, directly within Copilot Studio. You can also use Copilot, to open the Copilot panel on the side of the authoring screen to update a topic by using natural language.
Application Insights integration
Application Insights is integrated by default within Copilot Studio. Application Insights allows administrators to monitor the performance of applications. Copilot Studio allows administrators to both proactively understand how the application is performing, and reactively to review root cause analysis and determine the cause of an incident. Learn more in Telemetry with Application Insights overview.
Clone a classic chatbot
In Copilot Studio, switch to the classic experience.
Open the classic chatbot you want to migrate.
On the Overview page, select Copy this chatbot.
A Chatbots window opens, with these options: Copy and convert this bot or Try out the new studio experience.
In the window that appears, select Copy and convert this bot.
Replace the default name for your new agent if you want.
Select Create.
Important
If your bot isn't eligible for cloning, Copilot Studio prompts you to create a new agent instead.
Wait for Copilot Studio to create the new agent. When it's ready, select Go to your bot to open the new agent in Copilot Studio.
Ensure you completed all your testing, and complied with your organizational policies, before deploying your migrated classic bot to production.