The jtel cloud TTS service now supports the generation of voice files using both the Elevenlabs and Aristech TTS engines.

Added storage usage visibility for container-stack installations.

• Added datastore usage information to the Main UI.
• Displays the configured jtel storage promise.
• Displays the customer’s used storage and remaining available storage.
• Added storage usage information and graphical visualization to the License Usage view.
• Storage usage is calculated from the relevant ACD file and database volumes.

This gives administrators a clear overview of the customer’s configured storage capacity and current datastore usage directly in the portal.

Live Agent AI TAC suggestions are now integrated with the existing Transaction Code workflow in both Agent Home and the Mini Client.

When a TAC is suggested by Live Agent:

• If the Transaction Code popup is opened automatically after the call, TACs suggested by Live Agent are highlighted in the existing popup.
• If no popup is opened automatically, the agent can click an AI TAC suggestion to open the Transaction Code history popup for the corresponding finished call. The selected suggestion is highlighted.
• Live Agent does not set or persist a TAC automatically. The agent must confirm the TAC using the existing Set/Reset functionality.
• Existing Transaction Code rules, including mandatory TACs and the maximum number of TACs per call, remain unchanged.

The popup integration can be controlled separately for Agent Home and Mini Client using:

portal.Acd.AgentHome.LiveAgent.AI.Assistant.TAC.Popup
portal.Acd.AgentMiniClient.LiveAgent.AI.Assistant.TAC.Popup

This keeps AI TAC suggestions advisory while reusing the existing jtel Transaction Code workflow and validation.

Added support for using Azure Speech Services via a private/custom endpoint for both ASR and TTS. This allows Azure Speech resources with public network access disabled to be used through Azure Private Link/private networking.

The existing Client Master Data parameter:

Azure.SpeechServices.EndPoint

is used to configure the Azure Speech Services endpoint. For a private Azure Speech resource, configure the custom endpoint using the WebSocket scheme, for example:

wss://<speech-resource>.cognitiveservices.azure.com

Private endpoint mode is enabled by setting the existing region parameter to:

Azure.SpeechServices.Region = private

The configuration behavior is therefore:

Azure.SpeechServices.Region = <Azure region>
    -> standard public Azure Speech Services

Azure.SpeechServices.Region = private
    -> use Azure.SpeechServices.EndPoint as private/custom endpoint

Existing customers using public Azure Speech Services do not need to change their configuration. Public Azure continues to use the existing region and API key configuration.

Added a new setting Dial next phone number for same contact in the subcampaign Dialler Parameters. When enabled, preview dialling immediately tries the next valid phone number of the same contact after no answer or an invalid/unreachable number. Busy calls are excluded.

The next number is selected according to the configured phone number/time window order and remains assigned to the same agent.

TTS cache handling has been improved to prevent audio from a previous TTS request from being stored under the cache entry of a following request.

Note: this behaviour was only observed with early versions of Elevenlabs TTS support.

In rare timing situations, especially when a call was interrupted during TTS playback, a stale temporary TTS file could remain associated with the call. A subsequent TTS request could then incorrectly store that previous audio under the new cache key, causing the wrong audio to be played on later calls.

The temporary TTS file reference is now cleared before a new uncached TTS request is started, preventing stale audio from being written to the wrong cache entry.

Existing TTS cache files should be cleared after installing the update to ensure that any previously corrupted cache entries are regenerated.

A new parameter has been added to control whether Live Agent conversation data is written to the Call Log:

LiveAgent.CallLog.DisableConversationLogging

When enabled, Live Agent continues to use ASR, transcription and AI processing normally, but conversation-related data is no longer written to the Call Log.

This includes entries containing:

• Live Agent ASR transcription results
• AI pipeline input containing the conversation
• After-call AI input containing the conversation

Normal technical Call Log entries remain unchanged.

The parameter can be configured on client level and overridden on ACD Group level.

Default behavior remains unchanged when the parameter is missing, empty or set to false.

A new IVR object ASR Options has been added to configure ASR behavior directly within a routing application.

The object provides the following options:

• ASR Provider – optionally overrides the ASR provider configured for the client. Currently supported providers are Azure and Whisper. If no provider is selected, the client/default ASR configuration is used.
• Voicebot Recording – enables or disables recording of the caller's Voicebot ASR utterances.

The configuration is applied to subsequent ASR operations and can be changed multiple times within the same call. This allows, for example, switching between the client default provider and another provider during a routing application.

When Voicebot recording is enabled, the recorded utterances are stored in the client-specific Voicebot recording directory and are available from the Call Log.

Voicebot ASR recording is independent from Live Agent call recording. Live Agent transcription does not create a separate ASR recording; normal Live Agent call recording continues to be controlled by the ACD Group recording configuration.

Live Agent functionality is now available in the Mini Client.

Agents can access Live Agent information directly from the new jtel.ai tab in the Mini Client, including transcription, AI suggestions, sentiment information, call summary, transaction codes and customer satisfaction results.

The functionality is controlled independently from Agent Home using the following new resources:

portal.Acd.AgentMiniClient.LiveAgent
portal.Acd.AgentMiniClient.LiveAgent.AI.Transcription
portal.Acd.AgentMiniClient.LiveAgent.AI.Assistant
portal.Acd.AgentMiniClient.LiveAgent.AI.Assistant.Agent
portal.Acd.AgentMiniClient.LiveAgent.AI.Assistant.Caller
portal.Acd.AgentMiniClient.LiveAgent.AI.Assistant.Suggestion
portal.Acd.AgentMiniClient.LiveAgent.AI.Assistant.Summary
portal.Acd.AgentMiniClient.LiveAgent.AI.Assistant.TAC
portal.Acd.AgentMiniClient.LiveAgent.AI.Assistant.Satisfaction

The new Mini Client Live Agent resources are disabled by default and must be enabled as required.

A new License Usage view is available under Statistics for administrative users.

The view provides detailed usage information for Agent, Live Agent, and Voice Bot licenses, including:

• Usage details for a selected period
• Filtering by Agent, Live Agent, Voice Bot, or all license types
• Configured license limits
• Maximum usage within the selected period
• Bar chart showing license usage over time based on peak concurrent usage per interval
• Configurable chart granularity depending on the selected period
• Visual indication when a configured license limit is exceeded
• Automatic refresh for the current day

For Live Agent and Voice Bot licenses, usage is calculated from license usage events generated by the Blazar telephony servers. Usage from multiple servers is combined to determine the overall concurrent usage.

The existing license usage summary in the main view has also been extended to include Live Agent and Voice Bot peak usage alongside the existing Agent license statistics.

The new License Usage view is controlled by the resource:

portal.Statistics.LicenseUsage

By default, access is provided to System, Reseller, and Client administrators and is not available to normal User accounts

A routing application can now be executed after a Live Agent call has ended.

The routing application is configured using the following client parameter:

LiveAgent.AfterCallPipeline.RoutingApplicationsID

The parameter value must contain the ID of the routing application to be executed.

The execution mode is configured using:

LiveAgent.AfterCallPipeline.ExecutionMode

Supported values are:

AUTOMATIC – the routing application is executed automatically after the Live Agent after-call AI processing has completed.

MANUAL – the agent can trigger the routing application manually using the CRM button once the after-call AI result is available.

ACD group settings can override the corresponding client settings.

This enables customer-specific post-call processing, such as storing AI call results, updating CRM data, triggering follow-up actions, or continuing additional workflow steps.

WhatsApp can now be integrated using ChatWerk in addition to the existing Tyntec provider.

New SysAdmin parameters:

• ACD.Whatsapp.ChatWerk.Base.URL
• ACD.Whatsapp.ChatWerk.Credential

Chat Connector:

• Select the WhatsApp provider (Tyntec or ChatWerk).
• Existing Tyntec configuration remains unchanged.
• For ChatWerk, configure:
  • WhatsApp Channel ID
  • WhatsApp Webhook Secret

Incoming WhatsApp enquiries can now be handled by both agents and bots in jtel. Replies, including text messages, media, and interactive buttons, are delivered directly back to the customer in WhatsApp.

The system now counts Live Agent licenses when a call is actually connected to an agent and Live Agent transcription is active. The license counter is released again when the Live Agent session ends.

This enables monitoring of concurrent Live Agent usage for future reporting and license control.

A new Development Mode is available for Chat Connectors. When enabled, the complete chatbot response is displayed as formatted JSON in the chat window, which helps with testing and troubleshooting custom prompts. The prompt should be configured to return a valid JSON object when Development Mode is used. When Development Mode is disabled, standard text, button, and image responses continue to be rendered normally.

Azure OpenAI can now be used as an additional LLM provider for the Summary Bot.

To configure Azure OpenAI, define the following parameters:

• LLM.Provider = AzureOpenAI
• LLM.APIKey = <Azure OpenAI API key>
• LLM.Endpoint = <Azure OpenAI endpoint>
• LLM.Model = <Azure deployment name>

For Azure OpenAI, LLM.Model must contain the deployment name configured in Azure. The deployment name may be different from the underlying model name.

Example:

LLM.Provider = AzureOpenAI
LLM.Endpoint = https://<customer_domain>.azure.com/
LLM.Model = gpt-5-mini

A new IVR object LLM Bot has been added to simplify sending caller or IVR input to a configured Chat Connector chatbot/RAG endpoint. The object provides a dedicated configuration with the inputs Chat Connector, Sender, Message, Timeout, and Result variable prefix. The selected Chat Connector provides the chatbot configuration, while Sender and Message define the request content. The object returns one of three outputs: OK when the request succeeds, Other when the request completes with a non-success response, and Error when the request cannot be executed or fails technically. This makes chatbot integration in routing applications easier and avoids manual low-level request configuration.

Added support for OpenRouter, enabling access to a wide range of leading AI models through a single integration. Administrators can now choose from models provided by OpenAI, Anthropic, Google, Meta, Mistral, Qwen, DeepSeek, and many others.

Added the ability to enable or disable the Teams button for Teams Rules using the resource portal.Acd.Header.Teams.ToggleRules.

ACD media event handling has been improved to make working with new events faster and more convenient.

When the new setting AcdEvents.OpenNextNotOpenedEvent is enabled, clicking the media event notification icon can open the next new event directly instead of only showing the event list and also open the related CRM record automatically when available. The next event is selected based on priority and waiting time, so higher-priority and older new events are handled first.

When the new setting AcdEvents.OpenCrmUrlOnOpenEvent is enabled, opening an event can also open the related CRM record automatically when a CRM link is available. This helps agents access the relevant customer or case information more quickly.

For the supervisor event list, CRM auto-opening is controlled separately by the new setting AcdEvents.OpenCrmUrlOnOpenEventSupervisor. This allows CRM records to be opened automatically for supervisors only where this is desired, without affecting the agent behavior.

A new action is also available in the transaction-code popup, allowing users to complete the transaction-code step and close the event directly in one flow.

In some actions, the incorrect type of data change log could be produced. For example, updating a trunk group would produce an “inserted” log not an updated log. This problem has been fixed.

Fixed an issue in preview dialling where an invalid phone number could be marked valid again during call finalization.

This could cause the same invalid number to be selected again on the next attempt and prevent the contact from being closed with the configured "no telephone number available" result when no valid phone numbers remained.

Fixed an issue where SMS text was JSON-escaped for all SMS providers when sent via the Mini Client.

JSON escaping is now applied only when the configured SMS provider is Lox24. Other SMS providers continue to use the previous message handling to avoid encoding issues, for example with umlauts and other special characters.

An issue was resolved where, after an Agent-to-Group transfer followed by a Number/Synonym/PBX transfer, the cleanup could use the previous agent ID instead of the current agent ID. This could leave the second agent incorrectly marked as busy until the final transferred call ended.

Fixed an issue where the StatusChangeOnStartClient parameter did not work correctly. The problem occurred because the web session was not properly closed when exiting the MiniClient. The session is now closed correctly, ensuring that the configured status change is applied when the MiniClient is started again.

Fixed an issue where some Chat History Report entries displayed an error instead of the expected chat data.

Also improved the handling of different chat history entry formats so that available information is displayed correctly.

Corrected an issue where email addresses containing UTF8mb4 characters in the name part were not processed correctly.

Email addresses with UTF8mb4 characters in the display name are now handled correctly.

The IMAP connector now supports email messages whose top-level MIME type is application/pdf.

When replying to these messages, the original PDF can be included as an attachment in the response email, ensuring that the original content is preserved.

• Improved Live Agent stability. AI processing errors no longer interrupt the live transcription or assistant.
• Replaced the legacy suggestions model with a new LLM-based Entity Extractor for improved detection of customer numbers, telephone numbers and ticket numbers.
• Improved AI response handling by automatically removing unwanted Markdown formatting from LLM-generated JSON responses.

A typo “DSGVO” instead of “DSVGO” was fixed in the call logs.

Added support for configurable chat priorities.

Incoming chat and WhatsApp conversations now use the priority configured in the corresponding Chat Connector, allowing more flexible and consistent routing behavior.

Fixed an issue where the Create Callback IVR object would fail if the subject or body text contained a single quote character. Single quotes are now correctly handled, ensuring callbacks are created as expected regardless of special characters in the input fields.

An issue has been fixed where FTP recording filenames were not parsed correctly, causing agent information to be missing. Recording filenames now include the available agent information as expected.

Voice recording files uploaded via FTP could become corrupted when the system was configured to use the Blazar telephony server with the default FTP upload settings. This issue has been resolved — voice recordings are now reliably uploaded and remain intact on the destination server.

The sorting functionality in the Media History table view is now working as expected.

Introduced new SWYX connector configuration settings for the PBX connector address and communication port.

• pbxConnectorAddress (default: acd-tel1) – Defines the SWYX PBX Connector host.
• pbxConnectorPort (default: 40406) – Defines the SWYX PBX Connector communication port.

This feature is related to the new “call answer” feature implemented for the SWYX PBX.

Fixed a bug where anonymous (CLIR) calls were not correctly recognized by certain trunk translators. When the group rule "Callback 3" was configured, callers with a suppressed number should be prompted to enter their phone number via keypad before a callback is scheduled. Due to this bug, the anonymous call was not detected, causing the system to skip the number entry prompt and immediately schedule a callback without a valid caller number.

The email processing workflow has been improved as follows:

• If no corresponding AcdEvent can be found for an email, the email is automatically moved to the JTEL_Error folder for further investigation.
• If the corresponding AcdEvent reaches the Finished state, the email is automatically moved to the JTEL_Processed folder.
• A rollback mechanism has been implemented to ensure consistency. If an error occurs while creating an AcdEvent, all partially completed processing steps are reverted to prevent inconsistent data.

Fixed a race condition where an agent could still see or interact with a media event shortly after logging out, before the event was redistributed. The agent view is now updated promptly, and a warning popup notifies the agent when the event is no longer assigned to them because it has been redistributed.

Resolved an issue where ASR WAV recordings could not be downloaded from the call log. The root cause was that certain ASR IVR objects did not specify the ASR call recording name, causing the download to fail. The recording name is now correctly resolved in all cases, ensuring reliable access to ASR recordings from the call log.