Text-to-Insight
Convert natural language business questions into analytical insights.
Overview
The Text-to-Insight tool takes natural language questions and transforms them into analytical answers. Questions about panel metrics (card spend, foot traffic, rankings, growth) run as structured insight queries. Questions about what was said, disclosed, or published in documents — SEC filings, earnings transcripts, news, and bull/bear theses — route to Document Search. It relies on internal AI assumptions and interpretations to make best-guess interpretations of entities, time periods, and filters. This is an analytical tool and consumes MCP tokens.
Input / Output
| Input | Natural language questions |
| Output | Markdown-formatted analytical results with data tables |
Processing Method
- Relies on internal AI assumptions and interpretations
- NER tool makes best-guess interpretations of entities, time periods, and filters
- Less precise but more accessible for human users
- AI fills in gaps when context is incomplete
Example Prompts
- "Show me Walmart Card Spend in 2024"
- "What were the top 10 companies in FL last year?"
- "Which states had the highest growth in ad spend?"
When to Use
Best for:
- Direct human interaction requiring a natural language interface
- Exploratory analysis where precise parameters are unknown
- When the user doesn't have enough context to build structured frameworks
Not suitable for:
- Data discovery questions like "What data is available for Walmart?"
- Schema exploration like "What fields can I filter on?"
For more precise control over query parameters, consider using Framework-to-Insight instead. For questions against filings, transcripts, news, or theses, see Document Search.
Rate Limits
As an analytical tool, text_to_insight has a lower rate limit than discovery tools.
| Per Minute | Per Hour |
|---|---|
| 8 | 90 |
See MCP Rate Limits for how these limits work.