We’re excited to share the latest updates in Build 3126, focused on performance optimization, translation accuracy, AI feedback integration, and user interface consistency. These changes aim to enhance system reliability, improve support workflows, and streamline reporting processes across tenants.
Let’s dive into what’s new:
We’ve introduced a new report caching system to optimize performance for high-traffic API report queries, especially those using consistent filters like CSV format and static date ranges.
Why It Is Useful:
Reduces memory usage and load times for large reports requested via API. Admins can also view and manage cached reports, and the feature is fully toggleable for testing and rollback\
Users can now rate AI responses directly from the Prompt Sandbox using thumbs up/down feedback. The system tracks additional metadata like response generation time and integrates into the existing feedback dashboard.
Why It Is Useful:
Improves visibility into how the AI is performing in testing environments and ensures that quality signals are captured even outside of live sessions.
We’ve localized various elements of the AI Feedback experience—including labels and error messages—into German, Spanish, French, and Portuguese.
Why It Is Useful:
Expands accessibility and usability for international teams, aligning with our localization goals across all user-facing features.
Made translation logic more robust when working with non-standard or unexpected locale formats (e.g., “Basque” or “Catalan” display names instead of IETF codes). Includes fallback matching by language name, improved error handling, and fixes for nullability annotations in the translation service.
Why It Is Useful:
Prevents translation failures when guest locale data is inconsistent, especially in multilingual environments. Enhances cross-language communication between guests and agents.
To prevent unintended interactions, we've added logic that blocks seekers from sending messages while ticket listings are still loading.
Why It Is Useful:
This ensures clean ticket creation, avoids system confusion, and improves the quality of AI and support interactions.
We added support where seekers could begin a session using an image upload, resulting in errors.
Why It Is Useful:
Chat sessions now accept valid image inputs, reducing the risk of misfires and system-level exceptions.
A common issue with adaptive card image sizing in desktop Outlook has been fixed. Images now maintain their intended size and formatting.
Why It Is Useful:
Improves the professional appearance of outgoing emails and ensures layout consistency across platforms.
The ticket assignment caption has been updated to “Assigned to Agent List” for greater clarity. This update includes minor backend refactoring and a migration script refresh.
Why It Is Useful:
Standardizes text formatting and letter casing to match other field labels, improving visual consistency and overall UI polish.
Improved error management during tenant ID uniqueness checks when updating default web clients.
Why It Is Useful:
Prevents failures in edge cases and improves system stability during tenant configuration changes.