- Expand query scope — Let the AI query any view in the model - not just topics - for more unconstrained analysis
- Enable analysis validation — Have the AI automatically check its work for errors before returning results
- Manage conversation context — Control how aggressively Claude prunes conversation history to balance cost optimization with context preservation
- Tune model and thinking levels — Supported for Claude models only. Control the tradeoff between speed and accuracy for different task types (analysis, model building, summarization)
- Disable specific AI tools — Restrict the Blobby chat assistant from using specific auxiliary tools like Slack, Google Calendar, GitHub, visualize, or learn from conversation
- Control AI-generated custom fields and CTEs — Restrict whether the AI can generate custom fields and CTEs (common table expressions) based on user permission levels
Syntax
Properties
The parent object containing all AI settings sub-parameters:
| Parameter | Description |
|---|---|
| allow_ctes | Controls whether the AI can generate CTEs (common table expressions) in its queries, based on user permission levels. |
| allow_custom_fields | Controls whether the AI can generate custom fields in its queries, based on user permission levels. |
| analyze_configuration | Controls the model and thinking level for complex analytical tasks, such as query and visualization generation. |
| build_configuration | Controls the model and thinking level for model-building tasks, such as the modeling subagent and topic metadata generation. |
| conversation_prune_length | Controls how aggressively Claude prunes conversation history to balance cost optimization with context preservation. |
| disabled_tools | Disables specific auxiliary tools in the AI chat interface, such as Slack, Google Calendar, GitHub, visualize, or learn from conversation. |
| query_all_views_and_fields | Controls whether the AI can query any view in the model or only views organized within topics. |
| simple_summarize_configuration | Controls the model and thinking level for lightweight tasks, such as summaries, documentation search, and field metadata. |
| validate_analysis | Controls whether the AI automatically validates its analysis for errors before returning results. |
Examples
Expand query scope and validate analysis
Override model categories (provider-agnostic)
Restrict AI capabilities for Restricted Queriers

