Featured Highlights
Outbound AI Agent Safety Guard: We have added an outbound safety filter for text generated by AI agents. This automatically scans agent responses before they are sent to customers, ensuring that all communications comply with your brand guidelines, safety policies, and tone preferences.
Improvements & Bug Fixes
- Nested List Reply Options: Added support for nesting options inside interactive List Replies.
- Unassign Conversation API: Added a new API endpoint to programmatically unassign conversations from agents, along with updated developer documentation.
- Admin MCP tools: Added a set of administrative Model Context Protocol (MCP) tools that let developers query AI agent configurations, inspect live conversation exchange logs, and track platform events.
- Disconnect MCP Servers: Updated MCP authentication settings to allow administrators to disconnect and remove active servers.
- JSONPath Array Retrieval: Added tags to order details payloads and added support for array items retrieval in JSONPath expressions.
- Trigger Matching Fix: Fixed trigger matching logic to prevent over-scoping rules by connection mode.
- Subscriber Placeholder Fix: Fixed a bug in subscriber placeholder data filling.