Q2 2026 (Apr–Jun 2026)
Enhancements
-
Dynamic Masking of Sensitive Data On-Screen
Building on the new masking settings, personal data shown in SR Recognitions and Data Flow Reports is now automatically hidden on-screen wherever Dynamic Masking is enabled, with related audio playback disabled to prevent exposure — while raw data remains untouched in the database. -
Automatic Language Tagging for Text Outputs
All recognized text and event data across SR and Data Flow now carries the detected language. This groundwork enables language-aware features such as masking and improves accuracy for downstream analytics and translation. -
Refreshed Version Settings Experience
The project version dialog has been redesigned with clearer status options (Draft, Release, Default), descriptive helper text, and a more modern layout, making it easier to understand the effect of each choice before saving. -
Updated SESTEK Branding
The product logo and in-app naming have been refreshed to align with SESTEK's current corporate identity across all screens.
Improvements
-
Enhanced Voice Interaction with AI Agents
Data Flow's AI Agent integration has been substantially upgraded: it can now handle multipart audio responses from agents, share audio artifact URLs and synthesized TTS output directly with connected AI Agents, and expose a queryable Audio Logs API with optional OGG output — all built on a more robust, natively re-engineered voice-handling core for improved reliability. -
Consistent Page Header Layout
Page headers (titles, breadcrumbs, and action buttons) have been aligned to a single consistent layout across all Core Web pages, improving visual consistency and ease of navigation. -
Richer Context Shared with Kore.ai
Conversation identifiers from telephony (UCID) are now passed through to Kore.ai integrations, giving customers better traceability of calls across systems. -
Faster Database Performance
Database connection pooling has been introduced at the driver level, meaningfully reducing the time backend requests spend waiting on database transactions. -
Zero Audio Retention with Compliant Transcript Retention
Tenants with strict data-retention requirements can now configure audio retention down to zero, so call recordings are never persisted, while transcripts and metadata are still retained for a minimum of one month to satisfy legal and compliance obligations.
Bug Fixes
-
Fixed Missing Validation Warnings on Required Fields
Resolved an issue where no warning appeared when a required field was left empty. -
Fixed Validation Errors When Creating a Data Source
Resolved validation issues in the data source creation form. -
Fixed Unwanted Scrolling on the Data Flow Projects Page
Resolved an issue causing unwanted scroll behavior on the Data Flow Projects page. -
Fixed Timeout Setting Not Applying to the SR HTTP Node
Resolved an issue where the configured timeout on the SR HTTP node was not being applied. -
Fixed Condition Parameter Changes Being Lost on Tab Close
Resolved an issue where changes to condition parameters were lost if the browser tab was closed before saving. -
Fixed Speed-Type Notifications Not Reaching RTG
Resolved an issue where speed-type notifications were not being delivered to RTG. -
Fixed a Confidence Score Error with Early Silence Detection Enabled
Resolved a confidence-score error that occurred when the VAD Early Silence Trigger was enabled. -
Fixed a Crash Affecting Certain Data Flow Projects
Resolved an issue that could cause a crash in certain Data Flow projects. -
Fixed a Checklist Error in the Agent Assist Node
Resolved a checklist display error in the Agent Assist node. -
Fixed the "How Many Turns to Send to Generative AI" Setting Not Applying
Resolved an issue where the Agent Assist node did not correctly apply this configuration value. -
Fixed Agent-Only Scenarios Incorrectly Triggering on the Customer Channel
Resolved an issue in the Agent Assist node where an agent-channel-only scenario was also incorrectly triggered on the customer channel. -
Fixed Incomplete Entity Support in the Inverse Text Normalizer Node
Resolved an issue where the Inverse Text Normalizer node did not support all expected entity types. -
Fixed Custom Node Edits Not Saving
Resolved an issue where edits made to a custom node were not being saved. -
Fixed a Typo in the Node Config Address Field
Corrected a typo in a node configuration address field. -
Fixed a Connection-Drop Issue in the Virtual Agent Integration
Resolved an issue causing dropped connections during Virtual Agent integration. -
Fixed Masking Settings That Could Not Be Disabled
Resolved an issue where masking settings could not be turned off once enabled. -
Strengthened Credential Validation in Core Web
Fixed a credential validation issue in Core Web to strengthen authentication security.