Skip to main content

Ender Turing Updates — 2026-08-02 to 2026-08-15

Important changes

Report on AI usage in tokens in Gen AI Studio

Category: feature
Importance: high
What changed: The LLM Usage view in Gen AI Studio can now display your AI consumption in tokens, broken down by input, output, and cache.
Customer benefit: This lets you see exactly how many tokens your AI workflows consume, helping you optimize prompts and models for cost-efficiency.
Highlights:

  • Compare token usage across different prompts, models, and workflows using existing filters.
  • Savings from cached responses are now visible, as they are recorded with zero provider tokens.
  • Periods without historical token data are clearly marked as 'Unavailable' instead of being shown as zero.

Filter conversations by time until the next customer callback

Category: feature
Importance: high
What changed: The new 'Interval between conversations' filter allows you to find conversations based on how long it took for the customer to call back.
Customer benefit: This makes it possible to directly measure first-contact resolution by filtering for calls that were not followed by another contact within a specific timeframe.
Highlights:

  • Isolate conversations with no follow-up call at all to build a true 'first-contact resolution' list.
  • Find conversations that triggered a rapid callback to investigate unresolved issues.
  • The single, unified filter handles both the interval from the previous call and to the next one.

Role-based access to team dashboards for agents

Category: improvement
Importance: high
What changed: Access to team dashboards is now granted through a specific user role, rather than being controlled by a single account-wide setting for all agents.
Customer benefit: This gives administrators precise control over which agents can view team-level analytics, improving data privacy and security.
Highlights:

  • A new 'Agent with Team Dashboards' role is available to grant this access.
  • By default, agents can only see their own 'My Dashboard' and 'Training Roadmap'.
  • The previous global setting for agent dashboard access has been retired.

Browser-based recording now supports 15 more platforms

Category: feature
Importance: high
What changed: You can now record and analyze calls from 15 additional contact center and meeting platforms using the WebRTC browser recorder.
Customer benefit: This expands call analysis coverage to more of the tools your teams already use, such as Genesys Cloud, NICE CXone, Five9, and Microsoft Teams.
Highlights:

  • Administrators can enable recording for these new platforms themselves in the WebRTC app settings.
  • Supported platforms include major vendors like Talkdesk, Amazon Connect, Zoom, and Salesforce.

Analyze and filter internal agent-to-agent calls

Category: feature
Importance: medium
What changed: You can now filter, analyze, and report on internal agent-to-agent conversations as a distinct call direction.
Customer benefit: This allows you to separate internal coordination from customer-facing calls in reports, dashboards, and AI analysis for more accurate metrics.
Highlights:

  • Include or exclude internal calls in dashboards, agent views, and conversation lists with a single toggle.
  • Analytics charts for volume, handle time, and silence can now be broken down by internal traffic.
  • Ender Chat now understands and can report on questions specifically about internal calls.

Other changes

Monthly active agent usage now shows full-time and part-time breakdown

Category: feature
Importance: medium
What changed: The Monthly Usage report can now break down active agents into full-time and part-time groups based on how many days they worked.
Customer benefit: This provides a more detailed view of license consumption, showing how much capacity is used by regular versus occasional agents.
Highlights:

  • Agents active on 11 or more days in a month are counted as full-time; those active on 1–10 days are part-time.
  • The breakdown applies to the current month and the full six-month usage history.
  • This feature is enabled on request and does not change how your active-agent limit works.

Consistent output language for AI-generated summaries and evaluations

Category: fix
Importance: medium
What changed: AI-generated content such as call summaries and Auto QA evaluations is now consistently produced in the correct language, regardless of the project's speech recognition settings.
Customer benefit: This ensures AI-generated feedback is always readable and actionable for agents and managers without needing correction or translation.
Highlights:

  • Projects using automatic language detection now reliably receive results in English.
  • Ukrainian bilingual projects now correctly receive results in Ukrainian.

Improved stability in Organization Management

Category: fix
Importance: medium
What changed: We fixed an issue that could cause errors on organization management pages, such as user and settings screens, after a service had been running for a long time.
Customer benefit: This improves the stability of administration tasks, ensuring these pages load reliably at all times.

Read-only EnderGPT Chat access for agents

Category: improvement
Importance: medium
What changed: By default, agents now have read-only access to EnderGPT Chat, where they can view chats shared with them or their team.
Customer benefit: This provides administrators with clearer control over AI usage and costs, as interactive chat creation is now managed via role permissions.
Highlights:

  • Agents can still view scheduled chat results and chats shared directly with them or their team.
  • To grant an agent interactive chat access, an administrator can assign them a role with the appropriate permission.
  • The interface for agents is cleaner, hiding controls for actions they cannot perform.

Improved recovery from failed AI analysis

Category: fix
Importance: medium
What changed: The platform now automatically detects and replaces invalid cached AI responses, preventing repeated failures on the same conversation.
Customer benefit: This ensures Auto QA evaluations and call summaries complete more reliably, reducing gaps in your analytics and reporting.
Highlights:

  • A single corrupted AI response no longer prevents future analysis of the same conversation.
  • This fix helps avoid 'stuck' analyses that previously required manual intervention to resolve.

Clearer error messages for failed EnderGPT answers

Category: fix
Importance: medium
What changed: When an EnderGPT chat answer fails, the message now explains what went wrong and at which processing stage the failure occurred.
Customer benefit: This allows you to immediately see whether a failure is temporary and worth retrying, and helps our support team diagnose issues faster.
Highlights:

  • Failure reasons can include issues like AI provider rate limits or connectivity problems.
  • The message shows whether the failure happened while assessing the question, researching, or writing the answer.

Improved reliability of transcript corrections

Category: improvement
Importance: medium
What changed: We've improved the reliability of our transcript correction system and fixed a bug where a dictionary entry could add incorrect words to a transcript.
Customer benefit: This prevents a potential source of transcript errors and improves the foundation for future accuracy enhancements, such as correcting misheard names.
Highlights:

  • The fix prevents dictionary entries with no alternative spellings from damaging transcript text.
  • This change also hardens how original, unedited transcript text is displayed in the conversation view.

Clearer interface for topic accuracy improvements

Category: improvement
Importance: medium
What changed: The topic accuracy improvement feature is now labeled as Beta, and the list of topic versions is easier to read and manage.
Customer benefit: This helps set clear expectations about the feature's maturity and makes it faster to compare, select, and activate the correct topic version.
Highlights:

  • The action to make a version live is now a clearly labeled 'Activate' button.
  • The layout of the version list is now more consistent across all supported languages.
  • Actions like 'Activate' and 'Delete' now appear on hover to reduce clutter.
Product Release