Release Notes.

Written and illustrated reports, whether one-time or recurring

A report is a written and illustrated document generated by the AI based on your data: it includes a summary, structured sections, key figures, tables, concrete examples, and recommendations, along with details of the methodology used (time period, filters, scope, and sampling). It can be created in four steps from the Reports page, or by entering a phrase into Copilot from a dashboard.

• One-time, to answer a specific question, or recurring—daily, weekly, or monthly—scheduled for the time of your ritual.
• Each run is saved on the Reports page, and the document can be shared via a link.
• A report requested from Copilote reflects the scope displayed on the dashboard.

Copilote opens from any page

The Copilot is present on almost every page of the platform, and it knows where it is. From a dashboard, it reads the displayed widgets and active filters; from an evaluation, it works on the open conversation; from a grid, it helps write the criteria. On pages that do not display numbers, it remains available but does not provide any figures.

Well-crafted answers, ready to be shared

Each graph and each quote appears at the point in the text where it is mentioned.

• Customer quotes are taken from the original conversation, are dated, and are clickable.
• While he's searching, the Copilot explains what he's doing, step by step. Once the answer is written down, these steps fold into a collapsible line.
• When the Co-pilot writes a report, a button appears below its response to open it, and it remains there when you reopen the conversation.
• The Copilot identifies and names the filters on a section score or an additional score, and names the score behind an indicator when you ask it what a widget measures.

Report a problem or suggest an improvement via chat

If something isn't working—a button that doesn't respond, a page that won't load, or an export that isn't completing—just let Copilote know in the chat. It asks all the necessary questions at once, provides you with a summary, and forwards the issue to our teams, automatically attaching the page, active filters, and technical context. Feature requests follow the same process. You'll receive a confirmation in your notifications.

A graph based on a comprehensive analysis

A chart combines the various fields of a custom analysis—one bar per field—so you can see at a glance which ones stand out and compare them with each other before drilling down into the values for each one.

• With a time axis, a curve for each field tracks its frequency over the period.
• Clicking on a bar opens the field details, followed by its values and their subcategories, with a breadcrumb trail to navigate back.
• A field added to the analysis appears in the existing charts.

More readable charts

The charts on the dashboards have been redesigned to be easily understood at a glance.

• The total is displayed above the stacked bars, and the value of each segment appears in the pie chart legend.
• The text remains horizontal, a widget's title can span two lines, and the chart takes up the entire height of its card.
• The bars can be arranged horizontally.

Open-ended responses and customer feedback are processed through the filters

• A "filled" or "empty" filter applies to an open-ended question and can be pinned just like the others. The count of verbatim responses is limited to a specific question.
• A speaker selector—whether for a customer, advisor, or bot—lets you filter out all but the speech that interests you, both in the segments tab and in the message-based charts.

Please note: The verbatim count now excludes blank responses, so the numbers may be lower on existing dashboards.

Versions and a draft for the analyses and worksheets

Changes made to a custom analysis or evaluation grid remain in a private draft until you publish it, and each publication retains a version.

• A rating remains associated with the version it was assigned to.
• A selector allows you to review a previous version and restore it to a new draft.
• The name and scope are configured separately, without creating a version.
• When the Co-Pilot drafts criteria, they also write in the draft; only you can publish it. They agree to revise a template that has already been used.

Test the analysis on a few conversations before publishing

You link a small sample of existing conversations to an analysis, run the version you're currently writing against it, and compare the results with those produced by the version currently in use. The test runs on the draft without publishing anything, and the production results remain unaffected. A conversation is added to the tests from its details panel.

Pause an analysis without losing the results

An analysis can be paused: it stops applying to new conversations, and everything it has generated remains visible. It resumes with a single click, without recalculating anything. Your existing analyses already cover day-to-day events, campaigns, incidents, or spikes in activity, so there’s no need to create a new one for each one: the pause feature is intended for temporary analyses.

Grades are in good hands

The visibility setting for evaluations applies to invitees. An evaluated person can always view their own evaluations, including the AI’s explanation and the discussion. When this setting is enabled, only the scores are hidden, and they are hidden everywhere: in the details view, in the PDF export, and in Copilot.

Import Conversations in Bulk

An import screen can process an entire batch of conversations—up to fifty items at a time.

• Audio recordings, or written transcripts in spreadsheet or JSON format for chat and email, that can be read immediately without waiting for a transcript.
• An entire file can be submitted all at once.
• The advisor, date, and attributes can be set for the entire batch, line by line, or using a correspondence file—a template for which is available for download.
• Before importing, each issue is flagged along with the corresponding line: duplicate, unsupported format, date in the future, unknown advisor.

Each graph is explained

Some charts are inherently complex, and others have filters applied from the outset—filters that not all users are necessarily aware of. Here are two ways to understand exactly what a widget is showing:

• Ask it a question via Copilot (“Ask a question about this widget”): it explains what is being tracked, how, and for which scope.
• Click the widget icon, which shows the filters that have been applied.

The origin of each number can thus be verified with a single click at any time.

Conversation segments become available

The new "Segments" tab displays the conversation segments that actually match the search term or the applied filters. Go from metrics to verbatim quotes in a single step, without leaving the dashboard.

• Each segment lists its source, date, and agent.
• The search terms are highlighted.
• The original conversation opens with a single click.

Building views is faster

Fewer repetitive clicks, and everyone can pick up where they left off.

• Duplicate a widget with a single click to create variations of it.
• Allow the platform to name the widgets based on their configuration.
• Pin your favorite filters: they'll be right at your fingertips from one session to the next, and are specific to each user.

Alerts That Hit the Mark

Alerts and widgets now combine multiple dimensions of the same analysis—such as a topic and a sentiment. You can set up an alert to be notified when “delivery” appears alongside a negative sentiment in just a few clicks: you’ll be alerted to the specific combination that interests you, not the topic in general.

The interface is available in Arabic

In addition to French and English, the entire interface is available in Arabic, with a right-to-left layout designed for natural reading. Each user selects their language, and the platform remembers it.

More enjoyable, more reliable, everywhere

The platform is constantly improving, and your feedback directly contributes to that improvement.

• Standardized dates, more streamlined lists, and clearer navigation.
• Administrators invite a user with the appropriate spaces and permissions as soon as the invitation is sent.
• A series of fixes based on your feedback: restored quality of anonymized audio, filters that remain unchanged even when an analysis is modified, and improved stability of dashboards with extensive graphics.
• Behind the scenes: ongoing improvements to security and data compartmentalization, and enhanced performance when handling very large volumes of conversations.

The interface is now available in French and English

Internationalization now covers the entire application: alerts, data sources, custom analyses, the agent dashboard, evaluation grids, knowledge bases, Live Assist, and audit logs are displayed in the language of your choice.

• Dates, durations, and numbers automatically follow the format of the selected language.
• The language change takes effect immediately, without reloading the page.

Answers with clickable sources

When Copilote includes direct quotes in its response, a “cited sources” block appears below it, with one card per excerpt: a summary of the quote, the conversation, the attribution, and the date.

• Clicking on a map opens the cited conversation in a window, centered on the relevant excerpt.
• Quotes are directly linked to actual data: Copilote cannot invent a source.
• Sources remain clickable when you reopen an existing Copilote conversation.

A page on fairer and more readable usage

The Usage page has been reorganized into two tabs (“AlloIntelligence” and “Transcription & Anonymization”), featuring a detailed breakdown by map and a new Knowledge Bases map.

• The count now accurately reflects the analysis work performed: each grid element and each custom analysis counts as one analysis.
• When a evaluation is rerun, only the elements that are actually reanalyzed are counted: results that remain unchanged are not counted a second time.

Usage Tracking for Live Assist

A new "Live Assist" tab on the Usage page provides insight into real-time support activity.

• Hours of support, number of calls, and number of active support staff during the selected time period.
• Detailed breakdown by assistant and by staff member.

More Reliable AI Evaluations

• The quality of automatic evaluations is improving significantly, with a particularly marked improvement in Arabic conversations.
• When a criterion does not apply to a conversation (no actual exchange, condition not triggered), the item is now marked as "not applicable" instead of being rated at the lowest level.

Agent-Initiated Live Assist Sessions via Email

Phone integrations can now create a Live Assist session by identifying the agent by their email address—which is useful for systems that do not know the agent's internal ID.

Live Assist · Option to Record and Detection of a Refusal

It is now possible to disable an assistant’s recording (audio and transcript) and to detect during a call that a caller objects to being recorded, in order to stop the assistant. This is a real issue of consent and trust.

User-Friendliness of the Co-Pilot and Live Assist

You remain in control of the scrolling while the co-pilot types: you can read the beginning of a response at your own pace.

Copilot · New Features

The now-unified copilot offers three new features: it applies all the dashboard filters; it provides clickable links to the conversations cited as examples; and it saves and reuses prompts along with a history of exchanges. This last feature is a welcome replacement for a previously popular function: generating dashboard summaries and saving summary templates. More detailed and reusable analyses, with direct access to the conversations cited.

Historical QM/VoC Customers · Continuity Without Regression

Implementation of the following features from the current tool on the new platform: exporting VoC dashboards, generating PDF reports, and copying data from a chart (to retrieve all the chart information, just as on the old platform).

Dashboard · Comparison, Interactivity, and Speed

The ability to compare two time periods—already available for KPIs, lines, and bars—has now been extended to pie charts and lists. The charts are now clickable to filter the dashboard, which significantly improves usability. And query caching significantly reduces load times.

Real-time translation of conversations

A user can translate a conversation into their own language (FR / EN) to read, evaluate, and make use of it, even if the conversation took place in another language.

Copilote · Active Filters, Ad Hoc Analyses, and Relevance

The co-pilot can now see all active filters and can remove a specific one or reset them all at once while retaining the time period; its summary reflects the scope that was actually applied and flags a conflict rather than returning an empty result. CSAT, NPS, and histogram analyses are calculated and displayed correctly; filtering by category once again returns the correct results (and prompts for clarification in case of ambiguous names); and the copilot displays only the actions relevant to the current page.

Analysis & Evaluation · Easy-to-read comparisons and AI categories preserved

In a comparison widget, a missing value from the previous period now displays a true 0 for count/sum metrics (averages, scores, and ratios still display a dash). And editing an analysis no longer resets only the AI categories of the field that was actually modified: the others retain their previous values, without unnecessary regeneration. This ensures clear comparisons and preserves the category history.

Conversations & Agents · Identify the agent at a glance

Whenever a transcript is displayed (rating, conversation list, alert), the agent's tooltip includes their name and a link to their profile; the name also appears in the conversation list on a dashboard.

Live Assist · Expanded Rejection Detection

Refusal detection is being expanded: refusals are recognized based on context (using AI, recording, or analysis), beyond standard phrasing. The safeguard remains in place: a question, a simple mention of AI, or a moment of hesitation will not trigger the system to stop.

Dashboard · New Quality Analyses

Three charts round out the quality monitoring, with new applications:

• Changes in the quality score: Determine whether quality is improving or declining over a period of time.
• Score Distribution: The breakdown of conversations by score range—that is, the proportion of high- and low-performing conversations—beyond just the simple average.
• Top Trends: Identifying weak signals—topics experiencing sharp increases or decreases and newly emerging categories—to spot them early.

In addition, several usability issues with the dashboard have been fixed.

Unified Copilot: QM + dashboard all in one place

Two previously separate co-pilots are now combined: the coaching co-pilot (on the Quality Monitoring side: analyzing an agent’s performance) and the dashboard co-pilot (analyzing a dashboard and conducting qualitative analyses of conversations). Now combined, the co-pilot can be used across a wider range of pages: you can engage in dialogue on both the dashboard and in evaluations to provide coaching, and cross-reference Voice of Customer (customer-focused) and Quality Monitoring (agent-performance-focused) data—which makes perfect sense on a unified platform.

Agent Area

A page dedicated to an agent, which their team leader or supervisor can view to focus on the agents on their team. At this stage, it displays a list of the Live Assistants the agent has access to (and allows them to be launched directly from the page), as well as all calls made by the agent, whether they have been evaluated or not.

Data Accuracy and Segregation

Enhanced isolation of analytics between workspaces, deduplication of the conversation list, and improved reliability of filters and synchronization during bulk imports. Accurate, up-to-date, and segmented data.

Live Assist · More Comprehensive Assistance

Increased the character limit for prompts (up to 5,000) to support more feature-rich wizards, and verified access permissions for assisted calls.