AI Change Management When a Firm Adopts It

AI Change Management When a Firm Adopts It

Last Updated: November 2026

AI change management is the work of moving a whole business from one way of working to another, once software starts doing part of someone’s job. It covers the order people learn in, the staff who never asked for this, the old habits that have to close, and the reports that say all is well. Tools are the easy half; the firm is the other half, and that half decides whether any of it holds.

AI Smart Ventures has guided growing businesses through AI adoption across whole teams, not single desks. The pattern is steady: leaders describe a change their staff cannot see yet, and the distance between those two accounts is where the work sits.

Get the order wrong and you spend a year with a tool nobody trusts and a team that has learned to look busy with it. Get it right and the software turns dull, which means people stopped noticing it and started relying on it.

Key Takeaways

  1. Sequence by checkers, not by keenness: in Wharton and GBK Collective’s October 2026 report, 32% of IT teams and 26% of HR had changed how work gets done, against 8% in product and engineering.
  2. Usage rose while the mood cooled: the same study put daily use among managers at 62%, up 16 points in a year, while the excited share fell 8 points to 51%.
  3. Training looks better from above: 78% of managers said staff get enough AI training, while only 56% of the staff agreed.
  4. The people who hung back are late, not hostile: 7% called themselves resistant, yet 79% of them said AI saves them time.
  5. A manager’s job changes first: a Nielsen survey for monday.com found 78% of 1,700 directors and senior managers had changed how they judge their people’s work.

Read together, those five stop being a software question. Each is a management call: who learns when, who may be slow, and what a manager reads on a Tuesday. The last one matters most, because when two levels of the firm disagree about how the change is going, you still have to decide which account you believe.

How should you order the waves of a rollout?

Start with the group that already checks its own output, then widen in waves; keenness is the wrong thing to select for. You want work that repeats, someone who reads it before it leaves, and a manager with time to sit in. The Wharton Human-AI Research and GBK Collective study of October 2026 found 32% of IT teams and 26% of HR had changed how work gets done, against 11% of legal and 8% of product and engineering teams.

Departments that had actually changed how work gets done by October 2026, from Wharton and GBK Collective: IT 32%, HR 26%, legal 11%, product and engineering 8%, shown beside the share of managers describing an aligned organization at 39% against 26% of junior and mid-level employees

Three waves is usually enough at any headcount. Wave one proves the work, writes down what it learned, and keeps its existing check. Wave two is the awkward one, because it holds people who did not ask for this, so pair each of them with a wave one colleague in the same job. Wave three is everyone else, by which point the instructions exist and the stories are local. Leave about three weeks between waves, or wave one has nothing to hand on. That staging is what makes AI upskilling add up.

What do you owe the staff who did not volunteer?

A written reason, a named way in, and their current sign-off left alone. Most people who hang back are not fighting you. In the Wharton and GBK data, 7% described themselves as resistant to AI, and they used it daily far less often than colleagues, 48% against 63%, while 56% worried about their own role against 36% of the rest. Yet 79% of that group said AI saves them time.

So treat them as late, not lost. Give them the task they complain about most, never the showcase one, and let them keep the check they already run. Do not stand them in front of the room as a convert, because that costs you the trust you just bought. One quiet rule helps more than any session: nobody has to defend being slower, and everybody has to say what they tried. AI literacy grows faster when the question stays that practical.

What happens to a manager’s job here?

It moves from checking what people did to checking what they did not do by hand. A Nielsen survey for monday.com, reported on 21 October 2026, asked 1,700 directors and senior managers in seven countries. More than half said judging their team’s work had got harder, nearly half now spend more time reviewing output, 78% have changed how they rate people, and 39% use AI to draft reviews.

Three habits have to change, and none is a tool choice. Ask how an answer was produced before you accept it, and read samples of real work rather than usage totals, because a total says nothing about quality. Then guard practice hours in your own diary. Salesforce research from June 2026 asked 538 US managers about this, and 78% feel personally responsible for their team taking up new technology, while 48% feel pressure from above to show it. Those two numbers are how performed progress gets made.

AI Smart Ventures has trained over 20,000 professionals in Applied AI, and its AI Training work puts managers in the first wave rather than the last.

When does the new way become the only way?

When the old route is shut and the new one sits in the written process people follow. Until then you run two systems and pay for both in attention. The Wharton and GBK report found 54% of managers saying over half of last year’s AI pilots reached production, while only 28% call AI fully embedded in core processes. Most firms sit in that gap, and it does not close itself.

Shutting the old route is small, dull housekeeping. Update the template, the checklist and the handover notes on one named date, and say which step is gone. Then watch where people go when they are busy, since that is the real test of AI implementation. In Morgan McKinley’s 2026 global AI report, 48% of employees said they use personal AI accounts for work, and 34% had no policy or did not know of one. If your route is slower than a private one, the private one wins.

Why do managers and staff describe different firms?

Because the people who provide the training rate it, the people who receive it rate it, and the two readings differ. The Wharton and GBK study surveyed about 1,400 US managers and employees in July 2026. It found 78% of managers calling training adequate against 56% of junior and mid-level staff, and 79% seeing a clear long-term vision against 64%. In no group did both levels describe the same workplace.

Use that gap as your measure, not a usage dashboard. Ask four plain questions at both levels, a quarter apart: do you get enough practice, do you know where this is going, can you choose when to use it, and does anything still get in the way. Compare the two sets of answers rather than this quarter against last. Where departments had cut staff, the same study found only 20% described an aligned firm, against 53% elsewhere. Tool-first AI agencies rarely ask the second group at all.

Frequently Asked Questions

Does an AI change management certificate exist?

No single recognised one exists. Training providers and change bodies sell AI modules and badges, and some teach useful material, but none is a standard and no client asks to see yours. Judge a programme by what it makes you produce: a wave plan, a written rule, one measure you can read in a month. A slide and a badge means theory, not practical AI.

Which AI change management courses are worth it?

The ones built around your own work. A course earns its place if you leave with your process rewritten, not a general model of change. Ask three things before booking: how many hours go on your real tasks, whether managers attend with their teams, and what gets measured after. In the Wharton and GBK study, 78% of managers rated training adequate against 56% of staff. Ask the people who went.

Do you need slides to explain the change?

You need one page, and slides only if your firm reads them. Most people hunting for a change management deck want a shape they can trust. Hold it to five parts: what changes, who goes when, what stays the same, who to ask, and how you will know it worked. The part about what stays the same gets read twice. Leave the model diagrams out.

What does a workable AI change plan include?

Five things, and it fits on a page: the job AI is doing, the order teams get it, the check that stays human, the date the old route shuts, and the two numbers you will read in a month. Anything longer describes the plan instead of the change. Name the people in wave one, not the roles. A role cannot book an hour or answer a question.

What is pushing firms to adopt AI right now?

Three drivers, in rough order: AI arriving inside software you already pay for, plain-language tools that need no technical skill, and leaders who now use it themselves. Daily use among managers hit 62% in the Wharton and GBK report, up 16 points on the year. Agents that chain several steps are the newest of the three. None of them is a reason on its own; a task that repeats is the reason.

How does AI reshape a process once it lands?

It takes the first pass and moves the person to the check, which changes the shape of a day more than it changes the output. A step that took forty minutes of drafting becomes ten minutes of reading and deciding. Two effects follow: the slow part shifts to review, and skill at reviewing becomes scarce. Draw the process before and after on one sheet, or nobody believes the hours came back.

Which ethics questions come up during a rollout?

Four turn up in most firms: what data people may type into a tool, whether customers are told, how a biased output is caught, and who answers for a mistake. Write the first one down before training starts, because people otherwise guess at it. In Morgan McKinley’s 2026 report, 34% of employees had no AI policy or did not know of one. Guessing is the real risk.

How big should wave one be?

Enough to show a pattern and few enough to fit in a room, which usually means six to twelve in a growing business. One team, one repeating task, one manager at every session. Include someone who thinks the whole thing is overdone, since their questions are the ones the rest of the firm asks later. Below six you learn about individuals, not the work.

Can you run two ways of working side by side?

Yes, for a few weeks, and not much longer. Running both is how you compare quality fairly, so keep the old route live while wave one builds confidence. The cost is attention: two processes mean two sets of habits, and nobody is sure which counts. Set the closing date at the start, say what it is, and move it once at most.

How do you hear the objections nobody says?

Ask in private, in writing, and about the task instead of the tool. People rarely argue with a rollout in a meeting; they just get busy. Three questions do most of the work: what did you stop using, what did you go back to doing by hand, and what would you need to try again. Fix one answer visibly within a fortnight, or you will not be told twice.

Where does the effort go in firm-wide change?

Mostly into calendars, not software. Expect a week to plan the waves, an hour a week per person inside one, and half a day from each manager who attends. Effort tracks how many waves you run and how much work gets rewritten, far more than headcount. Ask any AI advisory partner for named stages, a shut-off date and an exit point. Schedule a consultation to set the order of your first two waves.

Executive Summary

AI change management is a sequencing problem in software costume. Order the waves by who already checks their own output, not by who is keenest, and leave three weeks between them. Owe the people who did not volunteer a written reason and the task they complain about, since hanging back is not refusal. Expect a manager’s job to change first, from reviewing people to reviewing output. Shut the old route on a named date, or both survive. Then ask staff and managers the same four questions and read the difference. That gap is the honest measure of AI adoption.

What Should You Do Next?

This week, write down your three waves and the names in wave one, then pick the date the old route shuts. Ask one manager in that wave to sit through the same session as their team, not a separate briefing. Put the four alignment questions to both levels and keep the answers side by side.

AI Smart Ventures offers AI Training for growing businesses taking a new way of working across a whole team. Schedule a consultation to set your wave order and the measure that proves the change landed.

People Also Read

About the Author

Nicole A. Donnelly is the Founder of AI Smart Ventures and an AI Adoption Specialist with 20 years of experience as a founder and CEO and over a decade leading AI adoption initiatives. She helps businesses integrate artificial intelligence with clarity and confidence, driving innovation and sustainable growth. Nicole has trained over 20,217 professionals in Applied AI, delivered 624 workshops, and worked with close to 1,000 organizations across diverse industries.

Expertise: AI Transformation, AI Strategy, AI Implementation, AI Adoption, Applied AI, Marketing, Business Operations

Connect: LinkedIn | Website

Disclaimer: This content is for informational purposes only and does not constitute professional business or technology advice. Results vary based on industry, existing systems and implementation commitment. Contact AI Smart Ventures for a consultation regarding your specific situation.