AI capabilities allow Agents and Copilots to use built-in AI features such as web search, image generation, and Canvas. These capabilities help AI-powered experiences search public information, create visual outputs, and generate structured content.
For Copilots, you can also configure Conversation starters to help users begin a conversation with suggested prompts.
Before you start
Before configuring AI capabilities:
You must have permission to edit the Agent or Copilot.
The Agent or Copilot must already be created.
Some capabilities depend on the selected AI model.
Canvas is available for Copilot only.
Conversation starters are available for Copilot only.
AI capabilities for Agent and Copilot
AI capabilities are configured in the Agent or Copilot settings.
Available capabilities are different for Agents and Copilots.
Capability | AI Agent | Copilot | Description |
Web search | Supported | Supported | Allows the Agent or Copilot to search public websites. |
Image generation | Supported | Supported | Allows the Agent or Copilot to generate visual mockups, illustrations, and image-based outputs. |
Canvas | Not supported | Supported | Allows Copilot to create structured content, documents, and visual working outputs. |
Open AI capabilities settings
Open the application.
Go to Application settings.
Open AI Agent or Copilot management.
Select the Agent or Copilot you want to configure.
Open the Overview tab.
Locate the AI capabilities section.
In this section, you can turn available AI capabilities on or off.
Configure Web search
Web search allows the Agent or Copilot to search public websites.
To enable Web search:
Go to the AI capabilities section.
Turn on Web search.
Save your changes.
Use Web search when the Agent or Copilot needs to answer questions using public web information.
Notes
Web search only searches public websites.
Internal application data should be provided through Knowledge sources, application resources, or tools.
Web search availability may depend on the selected AI model and platform configuration.
Configure Image generation
Image generation allows the Agent or Copilot to generate visual outputs from user requests.
To enable Image generation:
Go to the AI capabilities section.
Turn on Image generation.
Save your changes.
Image generation can be used for:
Visual mockups
Illustrations
Image-based outputs
Business content drafts
Notes
Image generation availability may depend on the selected AI model.
Generated images should be reviewed before being used in official business materials.
Output quality may vary depending on the prompt and selected model.
Configure Canvas for Copilot
Canvas is available for Copilot only.
Canvas allows Copilot to create structured content, documents, and visual working outputs. When users select Canvas in the Copilot chat, the generated result appears in a workspace on the right side of the page.
To enable Canvas:
Open the Copilot settings.
Go to the AI capabilities section.
Turn on Canvas.
Save your changes.
Publish the Copilot.
Use Canvas in Copilot chat
After Canvas is enabled and the Copilot is published, users can select it from the Copilot chat input area.
Open the Copilot chat.
Click Tools in the message input area.
Select Canvas.
After Canvas is selected, it appears in the input box.
Enter a request and send the message.
Copilot generates the output in the Canvas workspace on the right side of the page.
Users can review the generated content and, when available, copy the code or save the result as a Dashboard.
Notes
Canvas is available for Copilot only.
Canvas is designed for structured and visual outputs, such as reports, summaries, dashboards, and document-style content.
Canvas should be used when the user needs a working output, not just a standard chat response.
Configure Conversation starters for Copilot
Conversation starters are suggested prompts shown when users open a Copilot. They help users understand what they can ask and provide a faster way to begin common tasks.
Conversation starters are available for Copilot only.
Add or edit Conversation starters
Open the Copilot settings.
Go to the Overview tab.
Locate the Conversation starters section.
Click Manage.
The section switches to edit mode.
From here, you can:
Add a new starter
Edit an existing starter
Delete a starter
Reorder starters
Save or cancel changes
You can configure up to 4 conversation starters.
Each starter can contain up to 500 characters.
How Conversation starters appear to users
After Conversation starters are configured and the Copilot is published, users will see them when they open the Copilot conversation window.
Each starter displays up to three lines. If the text is longer, the remaining content is shown with an ellipsis.
Display behavior depends on screen size:
On smaller screens, each row shows one starter.
On larger screens, each row can show up to four starters.
Save and publish changes
After configuring AI capabilities or Conversation starters:
Click Save.
Test the Agent or Copilot.
Publish the Agent or Copilot when ready.
Published changes become available to end users after deployment.
Notes
Enable only the AI capabilities needed for the Agent or Copilot use case.
Canvas and Conversation starters are available for Copilot only.
Use Conversation starters to guide users toward common tasks or supported questions.
Keep Conversation starters clear, practical, and action-oriented.
Review generated images and Canvas outputs before using them in official business materials.
Use Knowledge sources, application resources, or tools when the Agent or Copilot needs access to internal application data.