Team-level
Decisions that are made within a team and are specific to that team's work
Driver runs it · Approver signs off · Contributors support execution · Informed updated
Where this team sits on the adoption spectrum
Uses the read to sequence where the central team engages first.
Team training level & needs
Flags domain-specific fluency needs that general training won't cover.
Use-case identification (spotting friction)
Adds the feasibility read. Friction-spotting stays with the people doing the work.
Which workflows get redesigned, and in what order
Brings ROI patterns and flags where a process should be redesigned before it's automated. The priority call sits with the function.
Building the routine, accessible automations
Provides the patterns and reviews against build standards. Coaches the champion rather than doing the build.
Building the hard, cross-functional, production-grade systems
Multi-agent systems, permission-aware retrieval, and anything that spans functions or touches regulated data.
Workflow maintenance once live
Sets the monitoring and evaluation standard. Worth naming the maintenance owner at launch, since AI workflows need ongoing attention.
Context maintenance — shared skills, projects, prompt libraries
Owns the org-wide library the team's work feeds into. Mirrors how the team already owns its documentation.
Specialized tool requests beyond the org platform
Assesses the case and whether it fragments shared systems. Specialized tools enter as additions to the org platform.
Measuring the value of AI for this team
Supplies usage and cost data. What gets done faster, what gets done better, and what gets done that wouldn't have happened otherwise.
Managing people to new AI standards
A management responsibility. Sits with team leads and People.