Responsibility as part of the model.
Team is not a decorative roster. In an agentic workbench, responsibility is operational data: who can act, what they are focused on, what they own, whether they are available, and where their recent work is visible. Without that context, even a strong model can stall at the handoff.
The internal response-time analyses show the same pattern: operators often answer quickly once launched, while the real delay can happen before the right person or agent is awakened. A useful Team perspective therefore connects identity, role, focus, availability, and work ownership, rather than treating people and agents as names outside the system.
The team is not an org chart. It is the live responsibility map of the work.
This responsibility map should sit beside Messages, Execution, Defects, and Portae. A user who sees a stalled assignment should be able to understand not only the card, but the operator context around it.