LLMs.txt: Complete documentation index for AI agents
Customize the Spotter chat experience

Customize the Spotter chat experience

You can customize the chat panel and conversational experience using the customization controls available in the SpotterEmbedViewConfig and SpotterChatViewConfig interfaces.

Spotter chat interface🔗

The chat experience in the embedded view varies according to the Spotter experience enabled in your Spotter embed.

FeatureDescriptionAvailability

Data model selector

Allows users to select a data model to set the context for their Spotter query. Supports setting All data models to let Spotter automatically determine and select the data source.

✓ Spotter 3
✓ Spotter 2
x Spotter Classic

Toggle to switch between Quick search and Deep analysis mode

Allows users to switch between the Quick search and Deep analysis mode in the Spotter interface.

✓ Spotter 3
✓ Spotter 2
x Spotter Classic

Add files

Allows users to add files from their local directory to provide context for a Spotter query. Available under the + icon in the chat input area.

✓ Spotter 3
x Spotter 2
x Spotter Classic

Connectors context

Allows users to add external connectors and set context for user queries based on the resources from these connectors.

In the new Spotter experience, this option is available under the + icon in the chat panel.

✓ Spotter 3
x Spotter 2
x Spotter Classic

Spotter starter prompts

Preconfigured search questions that guide users by displaying clickable suggestion pills in the Spotter chat interface. Includes the following prompt pills:

  • Quick search

  • Deep Analysis

  • Know your data

✓ Spotter 3
x Spotter 2
x Spotter Classic

Data model selection🔗

To load the Spotter interface with a pre-selected data source, you must specify the data context in the worksheetId attribute. You can specify the GUID of the data model or auto_mode in the worksheetId attribute to enable All data models option so that Spotter can automatically discover and select a relevant data model for user queries.

const spotterEmbed = new SpotterEmbed(document.getElementById('ts-embed'), {
    worksheetId: "auto_mode",
    // ...other embed configuration attributes
});

Quick search and deep analysis mode🔗

With Spotter 2 and Spotter 3 experience enabled in the embedded view, the Spotter page displays a toggle switch to allow users to switch between the quick search and deep analysis modes.

To show, hide, or disable this feature, use the action ID, Action.SpotterChatModeSwitcher.

const spotterEmbed = new SpotterEmbed(document.getElementById('ts-embed'), {
    // ...other embed view configuration options
    hiddenActions: [
        Action.SpotterChatModeSwitcher,
    ],
});

Spotter starter prompts🔗

Spotter starter prompts are preconfigured search questions that guide users when they first interact with Spotter. They provide curated starting points for data exploration by displaying clickable suggestion pills that help users understand what types of questions they can ask about their data. The starter prompts can be found below the prompt panel in the embedded Spotter page.

Starter prompts can be automatically created by ThoughtSpot based on your data model or manually configured by developers to address specific business needs. The prompts are data-model-specific and their content dynamically changes according to the data source content being used by the user for their interaction with Spotter.

Types of starter prompts🔗

When included in the embedded interface, the Spotter page displays the following starter prompts pills:

  • Quick search
    Includes common queries that users frequently ask. These prompts help users get quick answers to standard business questions, such as top products, sales summaries, or regional breakdowns.

  • Deep analysis
    Includes complex analytical questions that require deeper investigation. These prompts guide users toward more sophisticated analysis, such as trend analysis, correlation studies, or root-cause analysis.

  • Know your data
    Users can use this feature to understand the data available to them. This category is useful for onboarding new users or introducing new datasets.

Enabling Starter prompts in the embedded view🔗

To enable starter prompts in the embedded page, set the enableStarterPrompts property in the spotterChatConfig object to true.

const spotterEmbed = new SpotterEmbed(document.getElementById('ts-embed'), {
    // ...other embed view configuration options
    spotterChatConfig: {
        enableStarterPrompts: true,
    },
});

File uploads in Spotter chats🔗

The Add files option is available in the embedded view by default if your embed has Spotter 2 or Spotter 3 experience enabled.

To customize the visibility of the Add files option, use spotterFileUploadEnabled in the spotterChatConfig object.

You can also restrict the types of files users can upload by specifying file format in the spotterFileUploadFileTypes array. If no file format is specified, all supported file types are allowed for uploads.

const spotterEmbed = new SpotterEmbed(document.getElementById('ts-embed'), {
    // ...other embed configuration attributes
    updatedSpotterChatPrompt: true,
    spotterChatConfig: {
       spotterFileUploadEnabled: true,
       spotterFileUploadFileTypes: ['pdf', 'png', 'xlsx'],
    },
});

MCP connectors and resource selection🔗

The ability to select external MCP connectors and resources for Spotter AI analytics is available when Spotter 3 experience is enabled in the embedded view. This integration allows your application users to include both structured and unstructured data in their conversation sessions.

The Manage connectors icon in the chat panel allows users to select external connectors such as Google Drive, Slack, Notion, Confluence, Web search or Jira, as a data sources in their Spotter sessions. This list displays only the connectors that are preconfigured by the ThoughtSpot administrator.

To show, hide, or disable the Manage connectors option, use the Action.SpotterChatConnectors action ID in the hiddenActions or disabledActions array.

const spotterEmbed = new SpotterEmbed(document.getElementById('ts-embed'), {
    // ...other embed view configuration options
    updatedSpotterChatPrompt: true,
    hiddenActions: [
        Action.SpotterChatConnectors
    ],
});
spotterEmbed.render();

Spotter icon customization🔗

To override an icon, you must find the ID of the icon, create an SVG file to replace this icon, and add the SVG hosting URL to your embed customization code. The most common icon to override is the default Spotter icon and its icon ID is rd-icon-spotter.

The following example uses the alternate-spotter-icon.svg file hosted on https://cdn.jsdelivr.net/ to override the Spotter icon.

 init({
     //...
     customizations: {
    // Specify the SVG hosting URL to override the icon, for example Spotter (`rd-icon-spotter`) icon
         iconSpriteUrl: "https://cdn.jsdelivr.net/gh/thoughtspot/custom-css-demo/alternate-spotter-icon.svg"
     }
 });

The following figures show the customized Spotter icon:

Spotter icon customization

Spotter logo and ThoughtSpot branding label🔗

To hide the Spotter logo and branding in the chat interface and tool response, use the following SpotterChatViewConfig object properties:

  • hideToolResponseCardBranding
    When set to true, hides the ThoughtSpot logo and icon in tool response cards. The branding label prefix is controlled separately via toolResponseCardBrandingLabel.

  • toolResponseCardBrandingLabel
    Custom label to replace the ThoughtSpot prefix in tool response cards. Set to an empty string ('') to hide the prefix entirely.

    Example
    import {
        SpotterEmbed,
        SpotterEmbedViewConfig,
        SpotterChatViewConfig
    } from '@thoughtspot/visual-embed-sdk';
    
    const spotterEmbed = new SpotterEmbed(document.getElementById('ts-embed'), {
        // ...other embed view configuration options
        spotterChatConfig: {
            // Hide the default logo and label on tool response cards in Spotter chat UI
            hideToolResponseCardBranding: true,
            // Set a custom label to display as the branding on tool response cards
            toolResponseCardBrandingLabel: 'CompanyName',
        },
    });
© 2026 ThoughtSpot Inc. All Rights Reserved.