Think about how every major technology of the last decades entered your company. The ERP, the CRM, the cloud, even email in its day: leadership decided, the contract got signed, and then came the hard part. Convincing people to use it.
An entire “change management” industry exists because of that hard part: trainings, internal campaigns, champions, adoption metrics. Because resistance was the enemy, and it came from below.
With AI, the exact opposite happened.
The wave that didn’t ask permission
Generative AI didn’t enter companies through the technology committee. It entered through your people’s personal phones. They tried it at home, for their own things: drafting a delicate email, answering a question, preparing a class, translating something. And when it proved useful, they brought it to work. No permission, no training, no project.
It is the first technology wave that reaches the company from the masses, not toward them. Leadership isn’t pushing adoption: it is running behind it.
What this changes (everything)
The consequence runs deeper than it looks: the classic change management problem disappeared, and a new one took its place.
The classic problem was resistance: “I bought the tool and nobody uses it.” With AI, that problem barely exists. Your people already want to use it. They already do.
The new problem is ungoverned adoption: everyone on their personal account, pasting client information into tools nobody evaluated, with uneven quality and zero traceability. It isn’t bad faith; it’s people solving their work with what’s at hand.
Punishing it is the wrong reaction. Banning AI at the company doesn’t stop the use: it hides it. And a hidden practice can’t be improved, measured or protected.
Change management, inverted edition
If the wave comes from below, leadership’s job is no longer to push. It is to channel:
- Inventory the real usage. Ask without punishment: who is using what, for what? That’s your use-case map, for free. Your people already did the discovery for you.
- Give what works an official channel. Approved tools, company accounts, clear rules about which data can leave and which can’t.
- Make the sanctioned path the easiest one. If the official tool is worse than the personal account, people will go back to the personal account. Policy doesn’t beat convenience; it has to outcompete it.
- Invest where there’s already traction. The use case your team adopted on its own is the best candidate to deepen with real integration, your own data, and measurement.
The new role of leadership
In previous waves, leading adoption meant beating inertia. In this one, it means giving direction to a movement that already started. Less evangelizing, more judgment: deciding what gets formalized, what gets integrated into the business systems, and what stays as personal assistance.
That is exactly the job of our AI implementation practice: taking what is already happening inside your company and turning it into systems that perform. And if you need someone on your side of the table to organize the map, that’s where the fractional CTO comes in.
The revolution is already inside. The question isn’t how to get your people to adopt AI. It’s whether the company will catch up to its people.