The practical answer
Layer one: your own leverage
Start with your own work, because credibility comes from use, not from mandates. Leaders spend much of their week preparing, synthesizing, and communicating. AI is a strong partner for all three: preparing for a hard conversation, summarizing a long thread into decisions, drafting a message in your voice, stress-testing a plan by asking what could go wrong. Keep the judgment. Let the tool carry the drafting and the first pass.
Layer two: team capacity
Most teams do not need an AI strategy. They need two or three workflows redesigned so the tedious parts move to the tool and the human parts get more attention.
- Choose workflows with volume and clear quality standards, such as reporting, customer replies, research, and documentation.
- Redesign the work, not just the tool. Decide where AI drafts, where a person checks, and where a person decides.
- Make time for learning. Fluency takes practice inside real work, not a one-off training.
- Measure the capacity gained and where it was reinvested. Time saved that vanishes into more meetings is not a win.
Layer three: the norms only a leader can set
Adoption fails on human systems, not on models. People worry about their jobs, their data, and looking foolish. Leaders answer those worries with clarity.
- Say what AI is for here, and what it is not for.
- Set the data rules plainly: what may go into which tools.
- Set the quality rule: a person owns every output that leaves the team.
- Reward transparency. Nobody should hide that they used a tool, and nobody should hide that they did not.
- Watch who is being left behind, and bring them along deliberately.
Keep judgment where it belongs
AI raises the volume of output; it does not raise the quality of decisions unless leaders protect the moments where judgment matters. Decide in advance which decisions stay human: anything touching people, money at scale, safety, and reputation. Design those steps so a person is not merely rubber-stamping but genuinely deciding.
A ninety-day starting sequence
- Month one: use AI daily on your own work and share what you learn openly.
- Month two: redesign one team workflow with clear checkpoints, and measure it.
- Month three: publish the norms, celebrate honest wins and honest failures, and pick the next workflow.
Where people get stuck
- Announcing an AI initiative before using the tools personally.
- Delegating adoption to IT and treating it as a software rollout rather than a change in how work is done.
- Measuring activity, such as licenses and logins, instead of capacity and quality.
- Leaving the fear unaddressed, so people use the tools quietly or not at all.
- Letting output volume rise while decision quality quietly falls.
I would use an AI tool on my own work every day for a month, in the open, and I would tell my team exactly what helped and what did not. Then I would ask them which part of their week they would most like to hand off, and start there, with a clear checkpoint.
I would write the data and quality rules on one page before the second workflow began.
A recommendation from Maasha Kah, not a guarantee. Your situation decides the order.