Tim
07/15/2026, 4:22 PMAnswer Overflow
07/15/2026, 4:22 PMspotter-code
07/15/2026, 4:22 PMworksheetId when possible (single-model flow).
- If you need multi-model behavior, use Spotter 3 Auto mode intentionally, but avoid prompts that force exhaustive listing.
- Reword instruction to “return top relevant models + ask if user wants full inventory” instead of forcing all 15 every time.
If your requirement is strictly “always fetch/show all 15 immediately,” this appears to be a product behavior/limitation rather than an exposed SDK knob in the docs I found.
---
Suggestions:
- If latency is the issue, split experience:
1) Spotter for analysis,
2) Separate “model catalog” UI outside Spotter (pre-fetched once), so users don’t ask Spotter to enumerate all models.
- If you’re on Spotter 3, test with updatedSpotterChatPrompt: true and Auto mode to see whether model discovery UX is better for your users.
- Open a ThoughtSpot support/product request for a configurable “max models shown/fetched per turn” since I could not find this in official embed docs.
---
Docs:
- Embed Spotter: https://developers.thoughtspot.com/docs/embed-spotter
- Embed AI Search and Analytics (Spotter versions/capabilities): https://developers.thoughtspot.com/docs/embed-ai-search-analytics
- Spotter prompting best practices: https://developers.thoughtspot.com/docs/spottercode-prompting-guide#_best_practices_and_recommendations
- Spotter actions reference: https://developers.thoughtspot.com/docs/actions#_spotter