One operating model, four teams. Every agent is governed from R1 to R4, and the system learns from every cycle.