Release 2024-05-13
| Component | Version |
|---|---|
| Web portal (miarecweb) | 2024.5.13.0 |
| Recorder (miarec) | 2024.4.5.0 (not updated) |
| Screen recorded (miarec_screen) | 2024.4.5.0 (not updated) |
| Live Monitoring (miarec_livemon) | 0.1.0.183 (not updated) |
| Screen capture client | 1.1.0.80 (not updated) |
New features
Add AI Insights with Promt designer / AI Playground
Add AI Call Summary
Call summary structure can be customized by customer.
Upgraded Sentiment scoring to Large Language Model
It has a number of advantages:
- It is more accurate as it analyzed a whole conversation as one piece rather than analyzing invidual phrases.
- It provides a short explanation of why the AI choose such a score
- Customers can influence the sentiment scoring by explaining in the promt to AI what "negative" and "positive" means for them.
Add support of Markdown syntax to custome fields, AI insights and call notes
Minor changes
Hide NER panel in call details by default
Its content is useful to a small number of customers.
Add column "Tenant - Total Users (filtered)" in Tenant Details report
Such a column can be used to calculate a number of users in each tenant that match certain criteria, like users with calls for last 30 days, users of certain role, etc.
Display Custom Fields and AI Insights on the Evaluation Report page
Fix issues
Fix issue with the Password Reset and Create Password workflows not accepting the first verification code
Fix issue with change of "N/A" answer to "N/A (0 of X points)" when manually editing an Auto QA report
Fix issue with topic extraction job when a number of speakers is more than 2
Such number of speakers can be detected by the Speaker Diarization model.
Remove assumption of who is an agent and who is a customer when the source audio was mono stream
When audio is processed with the Speaker Diarization model, and multiple speakers are recognized, such speakers are labeled with generic Speaker 1, Speaker 2, Speaker N rather than Agent/Customer.