Getting Started with Spotter

Administrators can choose the default Spotter experience (Classic or Agent) for the organization or allow users to opt in individually. This option, also known as Make Spotter agent default experience for all users, is available under the ThoughtSpot AI section of the admin settings.

Spotter Agent can be enabled instance-wide or for specific users or groups, and data sharing with LLMs is always opt-in, respecting all security and compliance requirements.

If Spotter Agent is enabled for your instance, you can switch between Spotter Classic or Spotter Agent in your user profile.

Organizations can start with Spotter Classic and move to Spotter Agent as their data governance and AI readiness evolve.

In Spotter Classic, no data values are shared with LLMs; all analysis is based on metadata and governed models.

In Spotter Agent, data is only shared with LLMs when explicitly enabled by admins. All data access is governed by user permissions, and ThoughtSpot maintains strict zero-retention and no-training contracts with LLM providers. Bring Your Own LLM Key is also supported for organizations with additional privacy needs.

Choose Spotter Classic if your organization requires strict data privacy, is in early stages of AI adoption, or only needs metadata-level insights.

Choose Spotter Agent if you want to unlock the full power of agentic analytics, including richer insights, autonomous research, and advanced AI-driven features, while maintainin enterprise-grade governance and security.

To edit a Spotter answer, click the Edit button above the Spotter question box. The Edit Answer window opens.

You can add or remove query tokens, change measures and attributes and even drill down by different attributes directly in the answer while in edit mode or even in the conversation itself.

To save your changes, click Done editing. Otherwise, click Discard changes.

You can see sample rows of the underlying Model to get a sense of what columns are in the data set and how the sample rows look by clicking Preview data. Note that Preview data is not available for Models with chasm or fan traps, unless you contact ThoughtSpot support.

You can click tokens in a Spotter answer to edit them for clarity, such as specifying whether a measure should be classified as count or unique count, or changing the column being used as an attribute.

To edit a Spotter answer’s tokens, follow these steps:

You can give feedback on a Spotter answer without interrupting your conversation. This feedback is visible to your administrator in the Spotter Conversations Liveboard. This can be particularly helpful if you give feedback on incorrect answers, as this can highlight places where Spotter’s coaching can be better tailored to your business use cases.

If Spotter provides a correct answer, click the check mark at the bottom of the answer. If you see a thumbs up icon, you can review and save the underlying business terms and search query tokens so Spotter remembers your choices in the future for similar questions.

If Spotter provides an incorrect answer, click the X . If you are using Spotter Agent, you can give further details on what the answer got wrong. If you see a thumbs down icon, you can edit the underlying question, save, and then review the business terms and search query tokens of the answer.

To save a term, you must Approve or Reject it. After you approve or reject at least one business term, you can click Submit to save and share your feedback with Spotter.

Spotter shortcuts allow you to select from existing verified data constructs, like columns, filter values, and search data keywords, when you search. These shortcuts allow you to mix natural language and deterministic inputs when using Spotter. Spotter shortcuts allow you to create more precise questions by referencing actual column names and values from your Models. To enable Spotter shortcuts, contact ThoughtSpot support.

To use Spotter shortcuts, do the following: