Skip to content

Canvas Screen in Chat Inference

The Canvas Screen is an advanced visualization feature in chat inference, automatically triggered based on the user query and the agent's response type. It provides a dynamic and interactive way to present structured data, images, and graphical outputs within the chat interface.

Canvas Screen Display

When a user submits a query that results in structured, visual, image-based, or email-related data, the Canvas Screen can be viewed by clicking on the "View Details" option.

Example

  • After selecting the agent type, model, and agent, if the user queries "list all the available products," the Canvas Screen will appear in a tabular format when "View Details" is selected.
  • If the user requests to "send an email summary," the Canvas Screen will display the email content in a dedicated email component.

Supported Output Formats

  • Tabular Format:

    • Displays data in a table for easy viewing and comparison.
    • Example: Listing products, users, or any structured dataset.

  • Graphical Representations:

    • Supports charts and graphs for visualizing trends, analytics, or relationships in the data.
    • Example: Displaying sales trends, performance metrics, or statistical summaries.

  • Image Display:

    • If the agent is capable of image generation or retrieval, the Canvas Screen can display images relevant to the user query.
    • Example: Showing generated images, product photos, or visual search results.

  • Email Component:

    • Renders email content in a dedicated, formatted card for easy reading and interaction.
    • Example: Displaying generated email summaries, notifications, or correspondence.

  • Use-case Specific Card Components:

    • Presents information using custom card layouts tailored to specific use cases.
    • Example: Showing order details, user profiles, or task summaries in visually distinct cards.

Usage Highlights

  • The Canvas Screen enhances the user experience by providing rich, context-aware visualizations directly in the chat workflow.
  • It is especially useful for agents that return complex data, visual analytics, or image-based results.
  • The format (table, graph, image) is chosen automatically based on the agent's response and the nature of the query.