Professional-services AI Coordination Playbook
Coordination practices for consulting, accounting, legal, and advisory firms running AI agent teams. Covers engagement management, utilization tracking, knowledge reuse, compliance walls, and client relationship orchestration.
Engagement Management
Engagement Phase Gates
The engagement agent enforces phase transitions: Discovery to Planning, Planning to Execution, Execution to Delivery, Delivery to Close. Each gate requires sign-off data from specific agents (scope confirmed, resources allocated, deliverables reviewed). No phase advances without the gate conditions being met.
What goes wrong without this
A consulting team starts execution before the scope document is signed. Three weeks in, the client disputes the deliverables. The firm has to absorb $40K in rework because the planning gate was skipped.
Scope Creep Detection
The engagement agent monitors hours logged against the original scope estimate by work category. When any category exceeds 120% of plan, it flags the engagement partner with a specific recommendation: raise a change order, absorb the overage, or de-scope something else.
What goes wrong without this
A 200-hour engagement quietly grows to 340 hours. Nobody flagged it because each individual task seemed reasonable. The firm writes off $23K in unbillable time. The pattern repeats on the next engagement.
Single Engagement Owner Agent
Every active engagement has exactly one agent responsible for its health: timeline, budget burn, deliverable status, and client sentiment. Other agents (billing, scheduling, knowledge) feed data to the engagement owner. Nobody else sends client-facing updates about that engagement.
What goes wrong without this
The billing agent sends a budget overage warning to the client at the same time the delivery agent sends a "project on track" update. The client sees contradictory signals and loses confidence in the firm.
Stay in the loop
Get weekly coordination intelligence updates. No account required.