Run it
A framework only works if a team actually runs it. This is the meeting-in-a-box: the recurring rituals that turn the model into a rhythm, plus a starter set of metrics to show value early. Adapt every agenda to your campus, they're starting points, not scripts.
The rituals
Build shared understanding and surface priorities before any tool or policy decision. This single conversation is the lightest possible entry point, and often the highest-leverage.
- 10 minFrame: why now, and why mission-first. Set the ground rule, this is a conversation, not a vote.
- 20 minWhere are we really? Each person names one place AI is already helping and one place it worries them.
- 25 minAgainst the mission: which current or proposed AI uses clearly serve it, which don't, which are unclear?
- 20 minSurface the gaps: what's unowned, unclear, or invisible (shadow AI, no policy, no front door)?
- 15 minName 2–3 first moves and who owns each. Schedule the next session.
What to measure
You can show progress long before long-term outcomes land. These are leading indicators, pick a handful that fit your goals and your data, not all of them. Each pairs with an artifact you're already maintaining.
Adoption of the decision process itself, the leading sign governance is real, not theoretical.
Proves governance enables rather than blocks. Target: most decisions inside two weeks.
Visibility is the substrate of governance, rising registry coverage means fewer blind spots.
Closes the accountability gap, every consequential system should answer to a named human.
Shows risk is tracked and acted on, not just logged.
A healthy number isn't zero, it's evidence the institution catches and closes issues.
A concrete, early cost win from coordination, easy to show a CFO.
Proves the portfolio moves, the antidote to the pilot graveyard.
Faster, clearer answers to 'am I allowed to?' A felt improvement for the community.
Capability spreading across groups, not concentrated in a few champions.
The long-run trend line, is the institution actually getting more capable year over year?
Put these four sessions on the calendar for the year, pick three metrics to report each quarter, and the framework runs itself.