Improve Spotter's AI answers
If your organization has enabled Spotter conversational search, you can ask Spotter natural-language questions and get AI-generated answers. There are multiple ways to improve Spotter's AI-generated answers.
Overview of Spotter improvement methods
The following table compares different ways to improve Spotter's accuracy and usefulness. All methods are specific to a data source (a model), so they only apply when a user asks a question for that model. For the two coaching methods, you can apply coaching to similar models by following Migrate Spotter coaching to another model.
- Commonly asked questions
- Commonly used columns
- Ambiguous columns and terms
- Terms unique to your organization (Spotter is already trained on public information)
| Method and use case | Description | Application | Contributors |
|---|---|---|---|
| Spotter optimization: Make sure a model is ready for Spotter |
Explore helps you check for value indexing and column type issues so that Spotter can successfully interpret a model. |
Applies to all questions that use that model. Applies to all users. | Users with permission to edit that model |
| Coach reference questions: Improve answers to similar questions | You ask a question, get an answer, and edit the answer to ensure it's correct. You can explain your reasoning in natural language. Then, when a user asks the same question, Spotter gives them the exact answer that you coached. Spotter also generalizes what it learns in order to choose the right columns for similar questions. For example, the reference question How many claims are open in Ohio? will also help improve How many claims are closed in Maine? | Applies only to similarly phrased questions that use that model. Can apply to all users or just the user who coached it. | Users with coaching access for that model |
| Coach business terms: Map business terms to search tokens | You map business terms to search tokens. Then, when a user asks a question, Spotter doesn't have to guess what their words mean; It knows exactly which data they're asking for. The term must have a consistent meaning because Spotter applies the mapping to all questions asked for that model. | Applies only to questions that use that business term and model. Can apply to all users or just the user who coached it. | Users with coaching access for that model |