AI Agent resolution and escalation API metadata

Business problem

The client requests to have more insights on the AI Agent performance via the API or reporting. The closed_by field returns null when a ticket is closed by the AI Agent, and the necessary signals (such as ai_resolved_at) are undocumented and require specific internal relations to be loaded, making it difficult to calculate resolution, escalation, and abandonment rates.

Wished result

Provide a documented and reliable way to identify AI-resolved conversations via the API, such as a dedicated resolution_source field or public documentation of existing AI metadata, to enable accurate KPI reporting on autonomous resolutions vs. human escalations vs. abandoned conversations.