Release notes.

Written, illustrated reports, one-off or recurring

A report is a written, illustrated document produced by the AI from your data: summary, structured sections, key figures, tables, concrete examples and recommendations, with its method shown (period, filters, scope, sampling). It is created in four steps from the Reports page, or with one sentence to the Copilot from a dashboard.

• One-off to answer a question of the moment, or recurring, daily, weekly or monthly, generated at the time of your ritual.
• Every run is kept on the Reports page, and the document is shared with a link.
• A report asked of the Copilot takes the scope displayed on the dashboard.

The Copilot opens from every page

The Copilot is present on nearly every page of the platform, and it knows where it is. From a dashboard it reads the displayed widgets and the active filters, from an evaluation it works on the open conversation, from a grid it helps write the criteria. On pages that carry no figures, it stays available but never puts forward a figure.

Composed answers, ready to be shown

Every chart and every quote appears at the point in the text that discusses it.

• Customer quotes come from the original conversation, dated and clickable.
• While it works, the Copilot says what it is doing, step by step. Once the answer is written, those steps fold into an expandable line.
• When the Copilot writes a report, a button appears under its answer to open it, and it is still there when you reopen the conversation.
• The Copilot sees and names filters on a section score or an additional score, and names the score behind an indicator when you ask what a widget measures.

Report a problem or suggest an improvement from the chat

Something does not work, a button with no effect, a page that will not load, an export that never arrives: you tell the Copilot, in the chat. It asks its questions in one go, shows you a summary and passes it on to our teams, with the page, the active filters and the technical context attached automatically. An improvement request goes through the same path. You get a confirmation in your notifications.

A chart over a whole analysis

A chart brings together the different fields of a custom analysis, one bar per field, to see at a glance which ones stand out and compare them, before drilling into the values of each.

• With a time axis, one line per field follows how often each is spotted over the period.
• Clicking a bar opens the field's detail, then its values and their sub-categories, with a breadcrumb to walk back up.
• A field added to the analysis shows up in existing charts.

Charts that read better

The dashboard charts have been reworked to read at a glance.

• The total appears above stacked bars, and each slice's value shows in the pie legend.
• Labels stay horizontal, a widget title can run on two lines, and the plot fills the whole height of its card.
• Bars can be laid out horizontally.

Free-text answers and the customer's voice join the filters

• A "filled in" or "empty" filter applies to an open question, and can be pinned like the others. The verbatim count can be narrowed to one specific question.
• A speaker selector, customer, agent or bot, lets you keep only the speech you care about, in the segments tab as in charts counted per message.

Worth knowing: the verbatim count now ignores empty answers, so figures may go down on existing dashboards.

Versions and a draft for analyses and grids

Changes to a custom analysis or an evaluation grid stay in a private draft until you publish it, and every publication keeps a version.

• An evaluation stays attached to the version that scored it.
• A selector lets you read a past version and restore it into a new draft.
• The name and the scope are set separately, without creating a version.
• When the Copilot writes criteria, it writes into the draft too: only you publish. It accepts editing a grid that has already been used.

Test an analysis on a few conversations before publishing

You pin a small sample of conversations you already have to an analysis, run the version you are writing over them, and compare against what the live version produces. The test runs on the draft, publishes nothing, and leaves the production results untouched. A conversation is added to the tests from its detail panel.

Pause an analysis without losing its results

An analysis can be paused: it stops applying to new conversations and everything it has produced stays visible. It resumes in one click, with nothing to recompute. The analyses you already run cover everyday events, a campaign, an incident or a peak in activity, without needing one for each: pausing is for transient analyses.

Evaluation marks stay in the right hands

The evaluation visibility setting applies to guests. An evaluated person always sees their own evaluations, with the AI explanation and the discussion. Where the setting calls for it, only the marks are hidden, and they are hidden everywhere: detail, PDF export and Copilot.

Import conversations in bulk

An import screen takes a whole batch of conversations, up to fifty items at once.

• Audio recordings, or written conversations as a spreadsheet or JSON file for chat and email, readable straight away with no transcription to wait for.
• A whole folder can be dropped in one go.
• The agent, the date and the attributes are set for the whole batch, row by row, or from a mapping file whose template you can download.
• Before the import, every problem is reported with the row it sits on: duplicate, unsupported format, future date, unknown agent.

Every chart explains itself

Some charts are complex by nature, and others carry filters applied at design time that not every user necessarily knows about. Two ways to understand exactly what a widget shows:

• Ask the Copilot ("ask a question about this widget"): it explains what is counted, how, and over which scope.
• Click the widget's pictogram, which spells out the applied filters.

The origin of any figure can be checked in one click, at any time.

Conversation segments become available

The new Segments tab shows the conversation segments that actually match the searched term or the applied filters. From the number to the verbatim in one move, without leaving the dashboard.

• Each segment shows its source, date and agent.
• Searched words are highlighted.
• The original conversation opens in one click.

Building your views gets faster

Fewer repetitive clicks, and everyone picks their exploration back up where they left it.

• Duplicate a widget in one click to derive your views.
• Let the platform name widgets from their configuration.
• Pin your favorite filters: they stay at hand from one session to the next, personal to each user.

Alerts that hit the mark

Alerts and widgets now cross several dimensions of the same analysis, a topic and a sentiment for example. Being notified when "delivery" appears with a negative sentiment takes a few clicks to set up: you are alerted on the precise combination you care about, not the topic in general.

The interface is available in Arabic

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

More pleasant, more reliable, everywhere

The platform keeps improving, and your feedback feeds directly into it.

• Harmonized dates, smoother lists, clearer navigation.
• Administrators invite a user with the right workspaces and the right permissions from the start.
• A series of fixes driven by your feedback: anonymized audio quality restored, filters that withstand changes to an analysis, stronger stability for chart-heavy dashboards.
• Behind the scenes: continued strengthening of security and data isolation, and consolidated performance on very large conversation volumes.

The interface now speaks French and English

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

• Dates, durations and numbers automatically follow the format of the selected language.
• Switching languages applies immediately, without reloading the page.

Answers with clickable sources

When the Copilot quotes verbatims in its answer, a “cited sources” block appears below it, with one card per excerpt: quote preview, conversation, attribute and date.

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

A more accurate, more readable Usage page

The Usage page is reorganized into two tabs (“AlloIntelligence” and “Transcription & Anonymization”) with a detailed count per card and a new Knowledge bases card.

• The count now precisely reflects the analysis work performed: each grid item and each custom analysis counts as one analysis.
• Rerunning an evaluation only counts the items actually re-analyzed : unchanged results are not counted a second time.

Usage tracking for Live Assist

A new Live Assist tab on the Usage page gives visibility into real-time assistance activity.

• Assisted hours, number of calls and number of active assistants over the selected period.
• Detailed breakdown by assistant and by agent.

More reliable AI evaluations

• The quality of automatic evaluations improves markedly, with a particularly strong improvement on conversations in Arabic.
• When a criterion does not apply to a conversation (no real exchange, condition not triggered), the item is now marked “not applicable” instead of being scored at the lowest level.

Live Assist sessions by agent email

Telephony integrations can now create a Live Assist session by identifying the agent through their email address : useful for systems that do not know the agent's internal identifier.

Live Assist · recording opt-out and refusal detection

You can now disable recording (audio and transcription) for an assistant, and detect mid-call that a caller refuses to be recorded in order to stop the assistant. A real matter of consent and trust.

Smoother experience for the Copilot and Live Assist

Scrolling stays under your control while the Copilot writes: you can read the beginning of an answer at your own pace.

Copilot · new capabilities

On the now unified Copilot, three advances: it applies all the dashboard's filters; it returns clickable links to the conversations cited as examples; and it saves / reuses prompts with a history of exchanges. This last point advantageously replaces a previously appreciated feature: getting dashboard summaries and saving summary typologies. Finer, reusable analyses, with direct access to the cited conversations.

Historical QM / VoC clients · continuity without regression

Functions from the current tool carried over to the new platform: VoC dashboard export, generation of report PDFs, and copying a chart's data (to retrieve all the chart information as on the previous platform).

Dashboard · comparison, interactivity and speed

Comparison between two periods, already available on KPIs, line charts and bar charts, is now extended to pie charts and lists. Charts have become clickable to filter the dashboard : a notable usability gain. And request caching significantly reduces display times.

On-the-fly conversation translation

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

Copilot · active filters, ad hoc analyses and relevance

The Copilot now sees all active filters and can remove a specific one or reset them all at once while keeping the period; its summary reflects the scope actually applied and flags a contradiction rather than returning an empty result. CSAT, NPS and histogram analyses are computed and displayed correctly, filtering by category returns the right results again (and asks for clarification when a name is ambiguous), and the Copilot only exposes the actions relevant to the current page.

Analysis & evaluation · readable comparisons and preserved AI categories

On a widget in comparison mode, a value missing from the previous period now displays a true 0 for count / sum metrics (averages, scores and ratios keep the dash). And editing an analysis now only resets the AI categories of the field actually modified: the others keep their history, without unnecessary regeneration. Readable comparisons and a preserved category history.

Conversations & agents · identify the agent at a glance

Wherever a transcript is displayed (evaluation, conversation list, alert), the agent's bubble carries their name and a link to their agent space; the name also appears in a dashboard's conversation list.

Live Assist · broader refusal detection

Refusal detection is broader: refusal is recognized by meaning (use of AI, recording or analysis), beyond set phrases. The safeguard remains in place: a question, a simple mention of AI or a hesitation does not trigger the stop.

Dashboard · new quality analyses

Three charts complete quality tracking, with new uses:

• Quality score trend : see whether quality improves or degrades over a period.
• Score distribution : the breakdown of conversations by score range, i.e. the share of high and low performances, beyond the simple average.
• Top trends : surface weak signals: topics rising or falling sharply and newly appeared categories, to spot them early.

Several dashboard usability frictions have also been fixed.

Unified Copilot: QM + dashboard in one place

Two previously separate copilots are now merged: the coaching one (on the Quality Monitoring side: digging into an agent's performance) and the dashboard one (digging into a dashboard and running qualitative analyses on conversations). Unified, the Copilot can be used on a larger number of pages: you can converse on a dashboard as well as on evaluations for coaching, and cross Voice of Customer information (customer-oriented) with Quality Monitoring (agent-performance-oriented) : which makes full sense on a unified platform.

Agent space

A page dedicated to an agent, which their team leader or supervisor can consult to focus on the agents of their team. At this stage, it lists the Live Assistants the agent has access to (and lets you trigger them directly from the page) as well as all the calls made by the agent, whether evaluated or not.

Data accuracy and isolation

Stronger isolation of analyses between workspaces, deduplication of the conversation list, more reliable filters and synchronization during large imports. Accurate, up-to-date, isolated numbers.

Live Assist · richer assistants

Increased character limit for prompts (up to 5,000) for richer assistants, and verification of access permissions for assisted calls.