---
title: "Release 5.15.0"
description: "A new Client Deduplication workbench untangles merged and duplicated client cards, the queue cascade can now message a client when the connected operator stays silent, every closing dialog status can lead into its own scenario, email replies get a Paste with formatting button, and the chat widget can openly state that an AI is answering."
---

# Release 5.15.0

The headline of this minor release is the **Client Deduplication workbench**: a dedicated tool that finds merged and duplicated client cards and lets you untangle them safely — merge, split off a contact, or move dialogs — with a preview before anything is applied and a full journal of every action. Alongside it, the release closes three long-standing requests in request handling: the queue cascade can now message a client when the connected operator has been silent for too long; every status that closes a dialog can send the client into its own scenario, per bot; and the email reply field gains a **Paste with formatting** button that keeps links, lists, and tables intact. The chat widget can now also state openly that an AI is answering, as European regulation requires.

## Clients

**Added**

- **The Client Deduplication workbench — untangling merged and duplicated cards.** Over time a client base accumulates cards that actually describe the same person (the same contact arrived through different channels or at different times) and, conversely, cards where the history of several different people has piled up. While they stay tangled, the operator does not see the full request history, and some dialogs are effectively hidden behind someone else's card. This is now a dedicated section with three steps:
  - **Finding problem cards.** A background scan goes through the client base and builds the case list itself, grouped by problem type: exact contact duplicates, cases the system will heal on its own, and "stray" dialogs with no valid owner. The per-type counters double as a filter: one click shows only the painful cases, while the ones that need no action are hidden by default.
  - **Three operations on a card.** **Merge** — fold a duplicate into the main card; **Split off a contact** — move a contact together with its dialogs onto its own card; **Move dialogs** — hand individual requests over to the right card without touching contacts. The recommended operation for the case is pre-selected, and each option carries a sentence describing exactly what it will do to these two cards.
  - **A preview before you apply.** The wizard first shows the exact scope — how many contacts, dialogs, and messages will move — and only then asks you to confirm. Nothing changes until you confirm; if the data changed between the preview and the apply, the system shows you the discrepancy instead of quietly going ahead. The final "Result" step shows both cards as they look after the operation.
  - **A journal of every action.** Each operation is written into the workbench journal (what was done, from which card to which, by whom, and when), recorded in the settings audit, and leaves a note in the history of every affected dialog — so any change can be traced after the fact.

  The contact picker shows a person rather than an internal number: name, username, phone, email, and when the contact first appeared, with a search box — on a card with hundreds of contacts you can pick exactly the ones you need.

  > **Important.** Deduplication operations are irreversible, so during the rollout the workbench is available to the instance owner only. The module is enabled on your instance by the ConnectiveOne team on request — it requires a database structure update and an instance restart.

## Scenario Builder

**Added**

- **An automatic message to the client when the connected operator stays silent.** Previously the queue cascade worked only **before** an operator joined: the moment the operator picked up the request, the reminders stopped. If the operator then went quiet for a long time (checking something with a colleague or in another system), the client saw no movement and wrote again. The queue cascade node now has a second phase — **"Operator is silent"**: once an operator connects, a countdown starts, and if there is no **public** reply to the client within the configured time, the client receives the configured text or is sent into the scenario branch you choose. Every public operator reply restarts the countdown. Internal notes, system messages, and the client's own messages do not reset it — the client is messaged exactly when there really was no reply.

  You configure the steps, timing, and text the same way as ordinary queue steps, and you decide whether the last step repeats. If the operator disconnects, the countdown pauses and the next operator starts it afresh; closing the dialog stops the phase. The cascade node must be placed **after** the connect-to-operator action — otherwise the phase will not start for a returning client. The feature is optional and is enabled on the instance on request.

**Fixed**

- **Rolling a scenario back to an older version through the AI assistant no longer deletes sections silently.** If new sections appeared after the version you selected, the rollback now names them explicitly and waits for your confirmation instead of removing them without warning. Likewise, the assistant can no longer "save" an incomplete scenario when the list of live sections could not be retrieved at all — you get a clear error instead of a quiet save.

## Settings: dialog statuses

**Added**

- **A dedicated scenario after closing a dialog — per status and per bot.** Previously, once an operator closed a dialog, the client always returned to the same point in the scenario, identical for every closing status. A status that closes a dialog now has a **"Send the client into a scenario after closing"** block: for each linked bot you enable the transition and pick the entry point from the list of those available in that bot. The operator closes the chat with "Completed (request resolved)" and the client goes one way; with "Client did not respond" — another. A single status used across several bots leads to its own entry point in each of them.

  If no transition is configured, everything works exactly as before. The entry point is chosen from a list, so it cannot be mistyped, and the system will not let you save a status with the transition enabled and the field empty. The feature is optional and is enabled on the instance on request.

## Operator Line

**Added**

- **A "Paste with formatting" button in the email reply.** A plain Ctrl+V in the email reply field pastes plain text — links, lists, and tables from the copied fragment were lost, even though the internal note kept them. There is now a separate button next to the field: it pastes the fragment with links, lists, headings, tables, and emphasis, cleaned up for an email to a client — without the source's fonts, colors, or technical markup. It works with web pages, Google Docs, and Word. Plain Ctrl+V behaves exactly as before, so the familiar clean paste is still there. If the browser did not allow reading the clipboard, the panel prompts you to press Ctrl+V — that single press within 30 seconds also pastes with formatting.

**Fixed**

- **Sent emails no longer gain extra blank lines between paragraphs.** Every paragraph in a sent email used to become a paragraph plus an extra line break, and breaks appeared inside lists and tables — the client saw double spacing and a scattered table. The email and the dialog feed now keep exactly the structure you see in the editor. The same applies to internal notes in email chats.
- **An operator's reply in the dialog feed reads like an email again.** Spacing between paragraphs, list bullets, headings, quotes, and colored links are rendered instead of being flattened into a solid block of text. Incoming client emails are displayed as before.
- **Links to pages of the panel itself (for example, to a knowledge base article) stay usable in an email.** The editor could turn them into relative links, which would not open from the client's mailbox.
- **Group chats are no longer closed by timeout too early.** When several operators are connected to a dialog at once, the "last connected" time could be taken from the oldest of them, and the chat was closed while work in it was still going on. The most recent connection is now used.

## Client chat

**Added**

- **An open disclosure that an AI is answering.** Article 50(1) of the EU AI Act requires that a person be informed they are interacting with artificial intelligence when this is not otherwise obvious; the rule applies from 2 August 2026. The chat widget header can now show a small **"AI"** badge next to the bot name and a short explanatory line beneath it. The badge is visible only while a bot is talking to the client — as soon as a live operator joins, it disappears. No modals and no banners: the disclosure is persistent but unobtrusive. The explanatory text can be replaced with your own. The feature is optional, off by default, and enabled on request for the instances and bots where it is legally applicable.

## Fast Line Pro

**Changed**

- **The Fast Line Pro module can now be switched off on an instance in a controlled way.** Previously, disabling the module looked to the scenario like an ordinary call failure: the action failed with an error and requests were escalated as "bugs", even though the behavior was expected. The scenario now receives an explicit "module is disabled" signal and can handle it in its own branch. For everyone whose module is running, nothing changes — it stays enabled with no action on your side.

## Stability and internal improvements

- The background check for dialogs awaiting a timeout close was made faster: on instances with a large influx of new requests it could slow the whole system down for minutes every 30 seconds. The rules for who is closed by timeout, and when, are unchanged.
- Preparatory technical changes for features in upcoming releases; they do not affect current system behavior.

## Please note

- **Client Deduplication** is enabled on your instance by the ConnectiveOne team on request: it requires a database structure update and an instance restart. Until the module is enabled, the section does not appear; once enabled, only the instance owner has access to it.
- **The operator-silent auto message**, **the scenario transition by closing status**, and **the AI disclosure in the widget** are off by default and are enabled per instance — contact your ConnectiveOne manager. Until they are enabled, system behavior matches release 5.14.x exactly.
- The "Operator is silent" phase works only together with the cascade option **"Stop when an operator connects"**, and the cascade node itself must be placed after the connect-to-operator action.

## Related documentation

- [Configure dialog statuses](/en/settings/how-to/configure-chat-statuses.md)
- [Configure the operator queue cascade](/en/scenariobuilder/how-to/configure-operator-queue-cascade.md)
- [About the operator queue cascade node](/en/scenariobuilder/explanation/operator-queue-cascade-node.md)
- [Operator queue cascade parameters](/en/scenariobuilder/reference/operator-queue-cascade-parameters.md)
- [View scenario history](/en/scenariobuilder/how-to/view-scenario-history.md)
- [Reply to a client](/en/operatorline/how-to/reply-to-client.md)
- [Email view modes](/en/operatorline/explanation/email-view-modes.md)
- [Find and merge similar contacts](/en/broadcastusers/how-to/search-and-merge-similar-contacts.md)
- [View a client card](/en/broadcastusers/how-to/view-client-card.md)
- [View client channels](/en/broadcastusers/how-to/view-client-channels.md)
- [What is the chat widget](/en/channels/explanation/what-is-widget.md)