Three phases. Nine moves. Five rituals. Six anti-patterns. The minimum viable set of things a leader actually does to shift a culture from rollout to movement. No frameworks worth printing on a mug. No certainty we don't have. Start Monday.
Stop pretending the rollout hasn't started. 75% of your people are already using AI in private. Your first job is to make the shadow visible — without punishment.
“We don't have the answers. We have a direction and a willingness to figure it out with you. I'd rather we be wrong fast together than right slowly apart.”
Movements need a place to try things before they're load-bearing. Without a sandbox, every experiment becomes a referendum on someone's job.
“This experiment is allowed to fail. If it fails, we share what we learned. If it succeeds, we share what we did. Either way, the next person goes faster.”
Tools spread on their own. Cultures don't. By day 30, the conditions that make AI compound — or evaporate — should be visibly embedded in how you operate, hire, promote, and measure.
“Here is what we believe today. Here is what we believed last quarter that we no longer believe. Here is what we don't yet know. The document will keep changing because we will keep learning.”
Movements run on cadence, not memos. These are the only five recurring practices you actually need.
Three demos. No decks. One prompt, one workflow, one lesson. Voluntary. Recorded only if the presenter opts in.
One person publishes a prompt they actually used, with the input, the output, and what they'd change. Shipped in #ai-in-practice, not buried in a wiki.
Three people present an experiment that did not work. Goal is pattern recognition, not post-mortems. The org learns where the walls are.
Leader publishes the current organizational stance on AI. What's changed. What's still uncertain. Signed, dated, public.
Small-group conversation, manager-led: 'What does it mean to be good at your job now?' No agenda. Coaching, not communications.
Mandatory enterprise-wide training
Voluntary peer demos with the people already winning
Mandates breed compliance theater. Demos breed envy — the productive kind.
A center of excellence that owns AI
A network of practitioners that owns their workflow
Centers of excellence become bottlenecks. Distributed ownership becomes the operating system.
Measuring license utilization
Measuring workflow reinvention
Logging in is not learning. You can buy seats. You cannot buy reinvention.
A polished 18-month roadmap
A public 90-day stance you'll happily revise
Certainty you don't have erodes the trust you need. Honest uncertainty compounds it.
Pilot → evaluate → scale
Many small bets → public learning → emergent scale
Pilots filter for safety. Movements filter for energy. You need energy.
Communicating the change
Coaching through the identity work
Communications answer 'what.' People are asking 'who am I now?' Coaching is the unlock.
You will get what you measure. Pick numbers that reward reinvention, not consumption.
Most of activation is interpersonal, not institutional. These are the lines worth rehearsing — for the conversations you'll have this week.
“You're right that we don't know if this will work. That's exactly why we're keeping the bet small and the learning loud. What would you need to see in 60 days to update your view?”
“You're already two steps ahead. I don't want you to slow down — I want you to teach. What would it take to make your workflow legible to the team?”
“Your craft isn't disappearing. It's relocating. The question I want us to sit with is — what part of your judgment is now the most valuable thing you do?”
“We're optimizing for learning velocity this year, not headline ROI. The companies that compound are the ones whose people reinvent their own work. That's the line we're measuring.”
Print this. Mark it up. Cross out what doesn't apply. Add what your organization actually needs. Then put it down and go have the first conversation. Movements don't start on paper.