CX Release Notes

Deployment Guide

CX5.11.0 → CX5.12.0

Add Admin-Managed Do Not Contact (DNC) List with Outbound Enforcement

Lets authorized administrators manage a Do Not Contact list; adding phone numbers or email addresses individually or in bulk, and removing them when no longer needed directly from Unified Admin.

The platform now blocks outbound voice, SMS, WhatsApp, and email attempts to any number or address on the list, whether the attempt comes from an agent's manual dial or an automated outbound campaign flow, and clearly indicates when an attempt is blocked.

Add Delivery Status Tracking for Outbound Campaign Replies (X, LinkedIn, Instagram, Facebook, SMS via SMPP)

Adds delivery status tracking for outbound campaign messages across five channels, so the system now records whether each send was successfully delivered, failed, or bounced with per-channel detail (e.g., Messenger read receipts for Facebook/Instagram, SMPP DLR codes for SMS).

Campaign operators get consistent delivery-outcome visibility across all supported outbound channels in reporting, matching the pattern already available for Email.

Restrict Manual Outbound Channels by Team

Lets administrators optionally link one or more teams to an outbound channel. Agents then only see channels assigned to their team (plus any channel with no team restriction) when choosing a channel for Voice, SMS, WhatsApp, or Email outbound engagements, instead of every configured channel.

Refine Ongoing-Conversation Dashboard to Active/On-Hold Only

Filters ongoing-conversation dashboards to show only Active and On-Hold conversations. As a result, Inactive conversations which previously displayed a confusing "NO QUEUE" label, no longer appear on this dashboard at all.

Restore API Connectivity Across Core CX Components

Closes authorization gaps introduced by the platform's API-gateway rollout across Conversation Manager, Customer Widget, Quality Management Backend, the QM Connector, and the CX Data Platform.

Agents, supervisors, and Quality Management users no longer hit unexpected access errors when using conversation, widget, or AI-evaluation features that depend on these APIs.

Fix Customer SLA Timer Not Triggering on Web, Facebook

Issue: On web, Facebook, and conversations, the customer SLA timer did not start or expire as expected.

Resolution: The SLA timer now correctly starts and expires for these channels, restoring accurate SLA tracking.

Fix Agent SLA Timer Not Displaying on Agent Desk

Issue: The agent SLA countdown timer did not appear on Agent Desk for affected conversations.

Resolution: The agent SLA timer now displays correctly, giving agents accurate visibility into their remaining response time.

Fix Missing Visual Distinction Between Comments and Direct Messages

Issue: Public comments and direct messages appeared identically in the conversation view.

Resolution: Comments and direct messages are now visually distinguished, so agents can immediately tell which type of message they're viewing.

Fix Missing Outbound Email Option on Agent Desk

Issue: Agents were unable to start an outbound email conversation from the AgentDesk.

Resolution: Agents can now initiate outbound email conversations directly from the AgentDesk.

Fix Inability to View Full Email Thread History

Issue: Agents could not view the full email thread history for a conversation.

Resolution: Agents can now view the complete email thread, giving full context of prior correspondence.

Fix Forced-Logout Task Cleanup Exception

Issue: Forcefully logging out an agent with active tasks could throw a Redis-lookup exception, leaving the logout incomplete.

Resolution: Forced logout now completes cleanly even when Redis and in-memory task state are briefly out of sync.

Fix Queue Flush API Access Denied Error

Issue: The queue flush API returned "Access Denied" due to a missing role and an outdated Keycloak integration.

Resolution: Token validation and role assignment corrected; the API now works as expected.

Fix Agent Name Not Displaying on Real-time Dashboard

Issue: The dashboard showed the agent's system username instead of their display name.

Resolution: The dashboard now shows the agent's actual name and surname.

Fix Oldest Chats Not Routed First

Issue: When an agent became available, newer chats could be routed ahead of older ones in the same queue.

Resolution: The oldest chat in the queue is now routed first when an agent becomes available.

Fix Realtime Reports Memory and Startup Issues Under Load

Issue: Realtime Reports could hit Java heap OutOfMemory errors and 504 timeouts under high load or when the historical database was unreachable, due to large nested object deserialization blocking request threads.

Resolution: Reduced object footprint, added data expiry, and moved historical API calls to async handling, eliminating the heap exhaustion and thread pileup.

Fix Agent Details Dashboard Null Values on Disconnect

Issue: Abrupt agent disconnects (e.g., closing the browser tab) could leave stale agent stats showing null values on the dashboard.

Resolution: Agent stats are now correctly cleared on any logout/disconnect path, not just a graceful logout.

Fix AHT/AWT Not Displayed on Agent Desk Summary Dashboard

Issue: A required API mapping annotation was accidentally dropped during a branch merge, breaking the endpoint that supplies these metrics.

Resolution: Endpoint mapping restored; AHT and AWT now display correctly.

Fix Actual SIP Error Codes Not Shown for Outbound Calls

Issue: AgentDesk showed a generic error for failed outbound calls regardless of the underlying cause.

Resolution: AgentDesk now surfaces the actual SIP provider error code and message (e.g., 604) to the agent.

Fix Voice Recording Solution (VRS) Connectivity to Conversation Manager

Issue: VRS, which runs outside the Kubernetes cluster, lost connectivity to Conversation Manager once the platform's authentication gateway was enabled.

Resolution: Conversation Manager is now also reachable through a dedicated NodePort route for VRS, restoring the integration.

Fix Queue Performance Panel Displaying Null Values

Issue: The Queue Performance panel showed null for Total Queued, Answered, Abandoned, Service Level, and wait-time metrics.
Resolution: Resolved by recreating the Team Statistics dashboard ConfigMap and restarting the Grafana pod, so the panel picks up the latest dashboard definition.

Fix Twitter Downtime Backlog Comment Rendering

Issue: After a Twitter/X connector outage, one comment from the replayed downtime backlog could show as a generic "TC" placeholder instead of the customer's actual message on the next session.

Resolution: Fixed in the Twitter connector. The downtime catch-up comment now correctly displays as the customer's message on the next session instead of the "TC" placeholder.

Fix SLA Timer Race Conditions and Expiry Storm in Conversation Monitor

Issue: Under concurrent event processing, the wrong SLA clock could expire (e.g., a late agent message incorrectly restarting the Customer SLA), a threshold-configuration bug could trigger a runaway expiry storm driving high CPU/memory, and Customer SLA could incorrectly start after an agent disconnected on a re-queued email conversation.

Resolution: Conversation Monitor now tracks event timestamps rather than processing order to prevent stale events from flipping the wrong SLA clock, copies SLA threshold configuration per-timer to eliminate the expiry storm, and skips Customer SLA start when a conversation is already queued after disconnect.