Bug Fixes
- Fixed custom phrase matching for HTML-encoded characters (e.g.,
Org®
now properly matchesOrg®
) - Improved HTML entity decoding logic to handle multiple encoding levels from Salesforce
- Enhanced string processing to properly unescape content before custom phrase matching
Bug Fixes
- Fixed HTML language and direction attribute handling
- Resolved issue with excessive HTML wrapper tags being added to translated content
- Enhanced test coverage for HTML translation scenarios
Bug Fixes
- Fixed double-encoding issue in insight results processing
- Improved data handling for analytics operations to ensure reliable data storage
- Resolved JSON encoding errors that affected bulk data operations
Features
- Added advanced language detection capabilities with internally configurable provider selection
- Improved language variant handling for regional and script-specific language codes (e.g., Greek variants like el-CY, el-Latn, el-Cyrl now properly resolve to their parent language)
- Enhanced message query performance through optimized database indexing
- Added support for compressed payloads in Freshdesk integration to handle larger ticket threads
- Optimized Freshdesk initialization to use plain text instead of HTML, reducing payload size
Bug Fixes
- Fixed language script stripping logic to properly handle Unicode CLDR language identifiers
- Resolved substitution validation issues that could cause translations to be returned without proper replacements
- Improved validation flow to prevent early exits that bypassed necessary checks
Bug Fixes
- Improved escape/unescape processing for custom tags to ensure proper handling by translation providers
- Fixed issues with custom phrase recognition and translation
Features
- Enhanced language detection with more comprehensive HTML handling
- Simplified HTML translation processing for improved performance
- Increased input text limits for larger content translation
- Improved HTML text node translation with concurrent processing
- Enhanced word boundary detection for Unicode characters to better support international languages
Bug Fixes
- Fixed support for custom phrases in Chinese, Korean, Japanese, and other Unicode languages
- Fixed regex validation to properly handle Unicode word boundaries
- Improved validation logic for custom phrase matching across multiple languages
Bug Fixes
- Fixes custom phrase nested matching, custom phrases now match on longest phrase first and matches within that longest match are no longer matched/replaced
Features
- Live-Score is now enabled for Genesys clients
- Better telemetry data around languages detected and detection providers
Bug Fixes
- Internal prompt schema improvements
- Fixes issues related to long FreshChat message lookups
- Fixes issues where duplicate messages from Genesys were being returned
- Salesforce Live-Score triggers scores appropriately after customer responds
- Fixes concurrent language model lookup race conditions
Features
- Adds internal prompt management system
- Adds telemetry to prompt management system
- Adds internal centralized intelligence domain
- Adds ability to return masked PII in custom format (e.g. 1111-2222-3333-4444 -> XXXX-XXXX-XXXX-4444)
- Adds conversation simulation for enabled clients
- Adds an optional
UseSource
flag to the translator request payload, allowing propagation of known locale to translation providers
Bug Fixes
- Fixes French translation issues with certain providers
- Fixes one internal detection provider to utilize smaller subset of top languages to assist in more accurate locale detection
- Fixes an issue where needless errors are returned when translation contains something like
notranslate
, typically seen in Custom Phrase matching
Features
- Adds internal support center flags
- Enhances internal logging for faster issue triage
- Adds additional administrative functionality
Bug Fixes
- Fixes issues with Arabic translations not conforming to proper left-to-right formatting
- Salesforce insights no longer error when no conversation exists
- Fixes issues with large text mis-detection
Bug Fixes
- Fixes issue where translation provider determination was not deferring to smallest sample percentage
Features
- Adds internal
Support
api - Genesys application should now handle media messages
- Adds flags query to Studio allowing for simplified flag retrieval process
- Adds language pair combinations to internal translation provider determination
Bug Fixes
- Updates the summarize packages to produce more reliable summarizations
- Fixes issues with Supported language functionality across providers
- Fixes bug when attempting to unmarshal floats returned from AI providers. Changes schema to TypeInteger
- Better internal role checking for administrative users
- Fixes a nil pointer error in the Freshchat service for
system
users - Fixes an issue where multiple defaultRules being applied would result in an incorrect, incomplete unredaction
- Fixes issues that arise when a custom phrase is contained within a word
- Fixes issues with text between anchor tags not being translated
Features
- Improves service selection, specifically related to supported language logic
- Updates service selection to not only fall back to the target language’s parent, but also fall back to the source language’s parent before exhausting the supported language pair check
- Adds the ability to store
System
as the author for certain Freshchat messages - Enables filtering by only active organizations
- Adds additional administrative endpoints for Genesys
- Provides Salesforce specific endpoints for each Salesforce application
- Adds more granular roles for accessing usage data
Bug Fixes
- Fixes a nil pointer error, which is causing a bug with Freshchat
- Fixes issue where an unsuccessful or prolonged call reporting call holds and fails requests
- Disables French as a supported target language for some providers
- Fixes locale definition around
no
not being the parent language fornb
- Fixes translation fallback logic regarding
null
responses - Fix disruptive responses for third-party translation engine performance degradation
- Fixes issues with HTML translation of image tags
- Fixes issues with large HTML payloads not being fully translated
- Fixes issues with how HTML is handled between providers
- Fixes issues with incorrect MIME types being detected
- Fixes issues with Custom Phrases being incorrectly applied to URLs
Features
- Optimized large conversation translations
Bug Fixes
- Fixes occasional raw malformed HTML from being returned after translation
- Fixes HTML image persistence in right-to-left translations
- Fixes HTML numbered list formatting primarily in right-to-left language translations
- Fixes text and font formatting in language translations
- Fixes Salesforce-based merge fields from being translated