What's Changed
AI-Powered Result Insights
- Introduced AI profiling for SQL result sets, automatically generating descriptive summaries, statistical insights, and actionable recommendations directly in the SQL Console.
- Added a dedicated Summary tab and Result Set Insights panel to the result table UI, surfacing AI-generated context like data distribution, outliers, correlations, and potential anomalies.
- Launched AI-powered analysis actions, including quick transformations (e.g., “Filter top 10”, “Pivot by category”) and natural-language-driven chart suggestions—accessible via the Copilot Panel and Result Overview.
UX & Performance Improvements
- Stabilized layout behavior: Fixed flickering and misalignment when switching between table/chart/summary tabs; headers and charts now maintain consistent positioning.
- Optimized result caching: Prevented unnecessary invalidation of result caches during metadata-only updates (e.g., column renaming), improving query re-execution speed.
- Refined chart tooling: Unified auto-chart generation logic across result profiles, updated the chart control bar with better bucketing and action grouping, and improved canvas responsiveness.
Fixes
- Resolved an infinite sorting update loop in the virtualized result table that caused UI freezes during interactive sorting.
- Eliminated duplicate AI insight prefetches and tuned loading timing to reduce latency and redundant API calls when rendering results.
Internal
- Unified backend analysis infrastructure: consolidated
ai/result-insightsandanalysis/runendpoints, refactored prompt engineering (result.insights.ts), and introduced reusable types for result-context, actions, and chart profiles. - Added comprehensive test coverage for result-set insights and analysis suggestions (
result-set-insights.test.ts,analysis-suggestions.test.ts).