3 November, 2025

AutoQA

We updated the linked KB articles processing to avoid AI timeouts when entire KB is linked and improved AutoQA's output to more accurately cite used KB articles.

Criteria

We've done a few updates on the refinement with specific scores, including the provided input and output. Also, when you update AutoQA instructions, we set to "expire" any pending AutoQA optimisations to ensure any recent updates are not overwritten from old optimisation results.

Training

Admins can now update AutoQA's result in training evaluations to allow for more flexible and accurate scoring.

We've done further improvements on call simulations and fixed a bug to avoid evaluation criteria not follow the specified order instead of being reshuffled.

Data ingestion

Several fixes and improvements on our Salesforce and Jira integrations to support more data and in a better display format.

Evaluations

We've added sampling rule details on evaluation cards, for evaluations that come from workloads, and a "Sampling rule" drop-down on the Evaluations table.

Other

We've added a starter guide when you first get onboarded in the app or create a new workspace to help you go through and complete all the tasks you need to start receiving evaluations.