The failed approach
I have watched owners stand up and beg their team to adopt AI. It never works.
The all-staff please-use-AI email. The town hall where the owner makes adoption a personal mission. The lunch-and-learn where the consultant demos tools and the team politely nods.
None of it sticks. Six weeks later, two people are using the tools. The rest are back to doing the job they were measured on.
The owner concludes the team is resistant. They are not. They are responding exactly as designed.
People act on incentives
A team paid and measured to do their job will keep doing their job, not adopt your new tool on request.
If the bonus is on client delivery, they will deliver. If the promotion case is on billable hours, they will bill. If the work is reviewed for accuracy and speed, they will optimise for accuracy and speed.
None of those incentives point to adopting AI. They all point to doing the existing work well.
Look at it from the desk rather than the corner office. You are being asked to take on the risk and awkwardness of a new way of working, in exchange for a benefit that lands on the firm’s P and L rather than on your Tuesday. Declining is not resistance. It is arithmetic.
You do not persuade adoption. You arrange conditions.
The reliable lever
The reliable lever is making the desired action the easiest and most rewarding one. Not the loudest, not the most repeated. The easiest and most rewarding.
Map every step in the workflow where AI could remove friction for the team member who actually does the work. Then remove that friction. Then make the new way the path of least resistance.
The team member does not adopt AI because they were told to. They adopt AI because it is the easiest way to get their work done faster, with less rework, in less time.
We learned this the hard way inside our own firm. The workflows that stuck were the ones where the technician doing the AI-assisted task got their own job easier the same day. The ones we mandated from the top stalled, every time, no matter how good the tool was. Same firm, same people, same software. The only variable was whether the person doing the work got something out of it that week.
Not all friction is bad
Some friction protects quality and security and should stay. The skill is removing the friction that blocks the team while keeping the friction that protects the business.
Vicious friction, to remove. Clunky tools. No templates. No training. No reason for staff to change. Approval chains that exist because someone is anxious, not because the work needs them.
Virtuous friction, to keep. Security controls. Approval on client data. Governance gates. The checkpoints that mean the work is safe to send.
The distinction matters because the usual reaction to a stalled rollout is to strip out every obstacle in sight, including the ones holding the firm’s risk position together. That is how a rollout that was merely slow becomes a rollout that is genuinely dangerous.
Stage the adoption
A small group adopts early. Most follow once they see the wins. The last group moves once it is simply normal.
The first group are curious and will try anything if you point them at it. The middle group need to see the first group winning, in their own work, with their own clients. The last group need the new way to be the default.
You do not convert the last group first, which is exactly what most owners attempt, because the loudest sceptic feels like the problem to solve. Let the first group carry the proof. Then the middle moves. Then the tail follows because the new way is just how the firm works now.
The question to ask before any rollout
Before you roll out an AI workflow, ask two things.
Who has to change their behaviour for this to work, and what is in it for that person in the first week? Not in the annual accounts. The first week.
If the honest answer is “nothing yet, but it helps the business”, you have designed vicious friction and it will stall. Fix the incentive before you fix the tool.
Ties to the governed-adoption thesis
Friction is not the enemy. Ungoverned friction is. The whole point of the certification-grade system behind our AI work is to keep the virtuous friction and remove the vicious friction. That is what makes adoption durable rather than enthusiastic for a month.
The standard for this series
Every week is something we have run inside our own firm or inside a real New Zealand client firm. No theory I have not run myself. No invented case studies.
What to do next
If you want to design an adoption approach your team will actually follow, book a 15 minute call with me. If it is not a fit, I will say so. The AI Readiness Audit is the lighter first step.
Week 7 of The AI Divide lands next week. Price to value, and stop selling from your own wallet.
