This Week’s Highlight

Chat-based training sessions now run on the same live conversation engine as voice sessions. Responses stream in progressively, conversations can branch based on what the learner says, and session tracking works exactly the same way. If you have been using text mode, the experience is noticeably more dynamic starting this week.

Also This Week

Learners get a clear briefing before every training starts

Before entering a session, learners now see a structured overview of the scenario, their role, the intensity level, and what the scoring focuses on. If a tutorial video is attached to the training, learners are prompted to watch it first, and the product remembers whether they have watched it.

Evaluations now run in the background and process faster

Multiple evaluations run at the same time instead of queuing one after another. When you trigger an evaluation, the product returns immediately and a notification arrives when your score is ready. Organizations working with long transcripts can also select a higher-capacity model to prevent failures on large calls.

Voice training conversations flow more naturally

The AI persona in voice sessions no longer gets cut off when a learner starts speaking. Turn-taking now waits for the persona to finish before accepting the next input, so the conversation feels like a real exchange rather than a series of interruptions.

Sign in with Apple is now available

Users can sign in to Insight7 using their Apple ID alongside existing sign-in options.

Fixed & Improved

  • Persona details set in a coaching program (name, job title, and related attributes) now appear correctly in the generated roleplay scenario. Previously they were collected but silently ignored.
  • Sessions that completed without a score now show a Re-evaluate option on the training detail page, with live status that updates when scoring finishes.
  • Incomplete sessions are now labeled “Abandoned” in the session list. The meaning is the same; only the label has changed.
  • The coaching program a training belongs to is now shown on the training detail page.
  • Removing a participant from a coaching program now cancels only their assignments in that program, not assignments in other programs.
  • In-app notifications for new training assignments were silently failing and have been restored.
  • Generated trainings are now checked against your curriculum topic and team context before being assigned, improving accuracy.
  • A crash when opening the create-program modal has been fixed.
  • Research summaries, action items, and next steps are now generated with a more grounded approach, improving accuracy and reducing errors.
  • A new option lets organizations score every evaluation criterion on a call regardless of whether it directly applies, useful for screening-style evaluations.
  • The onboarding experience across Call Performance, AI Coaching, and Qualitative Analysis has been redesigned. The sidebar stays hidden during onboarding and appears at each path’s key value moment.
  • Call signals now use organization context and speaker role information for better-grounded insights.
  • Pidgin English is now available as a language option for voice personas in training sessions.
  • Learning journey plan assignments no longer fail when your preferred template isn’t available; the system now falls back to an alternative automatically.

Log in to your Insight7 workspace to try the new chat training experience, or explore AI Roleplay & Skills Practice to see the full range of what’s new.