Shipped improvements to our Query models

We’ve made significant enhancements to our Query Models for sharper subquery generation, smarter keyword handling, and better performance.

• Sharper Subquery Generation:  No more filtering on generic terms or organization names—say goodbye to irrelevant results like “Feedback from customers” or “Feedback from users.”
• Smarter Keyword Handling: The new model handles specifics more effectively. Example: “What is the total number of support tickets in the last year?” The new model is capbabe of zeroing in on the specificity of support tickets and timeframe.‍
• Query Performance Boost: Enjoy a 2% gain in accuracy without any compromise on overall performance.

Launched Taxonomy Estimator for improved Label processing

We’re excited to roll out our new Taxonomy Estimator for label processing! Previously, whenever a label was edited, merged, or newly created, we reprocessed and re-tagged feedback records based on updated information within an hour (though we occasionally missed this target).

Now, with our latest update:

• Accurate Time Estimates: The Taxonomy Estimator provides iterative time estimates for label processing, giving you a more accurate idea of when your updates will be completed.
• Improved Reliability: If label processing gets stuck or fails, the system automatically restarts to ensure smoother and more reliable operations.

This update sets realistic expectations and reduces frustrations around label processing delays, ultimately enhancing user experience.

‍