AI Auto-Reply Config, Smarter Prompts & Live Chat Feedback
New AI auto-reply controls, cleaner classification prompts, and live chat feedback you can act on. Async processing and per-workspace AI settings round it out.
Smarter AI Configuration & Processing
We introduced a new AI auto-reply configuration with enum support and refined AI prompt logic for better tag and priority classification. Async processing for AI agent actions has been enabled, improving performance and reliability. Each workspace now enforces unique AI settings for cleaner management.
Improved Classification Prompts
Support message classification prompts have been simplified with clearer instructions and improved examples. This ensures higher accuracy and more consistent labeling across conversations.
Feedback & Knowledge Source Management
Live chat now includes feedback submission with validation and rate limiting, allowing customers to suggest better Q&A content. AI settings also display detailed counts of knowledge sources by type and parent, helping teams better manage their knowledge base.
Improvements
- Event handling improved for tags and conversations.
- Event-driven processing added for saved reply embeddings.
- Inboxes can now dynamically access chat widgets.
- AI setting responses now include detailed knowledge source counts.
- Config merging improved for AI settings, with chatbot enablement flag added.
- Updated mappings, validation, and source type naming for clarity.
Bug Fixes
- Fixed message filter logic for AI actions.
Deprecations & Cleanup
- Removed old AI chatbot feedback method.
- Deprecated legacy controller and request classes for knowledge source management.
- Removed unused enums, factories, and token models.
- Simplified token usage tests for easier maintenance.
