AI adoption: getting the whole company on board

Ask ten people in your company how they use AI and you will get ten answers. One drafts every email with it. One refuses to touch it. One has built a private library of prompts that nobody else knows about. The company, as a whole, has no idea what any of it adds up to.

Usage is not the problem. In Wharton and GBK's "Accountable Acceleration" report (October 2025, 800+ senior leaders), 82% said they used generative AI at least weekly and 46% daily. The problem is that individual habits don't compound. Reforge made the same point in "Your Team's AI Is Siloed. Here's How to Start Fixing It." (Sachin Rekhi, 21 August 2026): gains stay with the individual until a team shares its skills and makes its context usable by the tools.

What teams get wrong

The usual response is a licence for everyone and a lunchtime training session. It raises usage and changes little else, because it leaves the hard decisions with each person: which tasks to use AI for, what a good result looks like, and who checks it.

When those decisions stay private, quality varies from person to person, and colleagues inherit the cleanup. In a September 2025 survey by HBR, Stanford and BetterUp, 41% of workers said they had received AI-generated "workslop", work that looks finished but isn't, and reported about two hours of rework per instance. It is a self-reported survey, so read the numbers as a signal. Most teams will recognise the pattern.

McKinsey's "The state of AI in 2025" (November 2025) points the other way. The small group of high performers, about 6% of respondents, were far more likely to have redesigned their workflows. They changed how the work is done. Giving people access was only the start.

Five moves from private habits to shared practice

1. Choose the workflows before the tools

Pick three pieces of work that happen every week and cost real time, such as proposals, weekly reporting or first replies to customers. Name them. Everything else stays optional for now.

2. Agree what good looks like

For each workflow, write one page: what the output must contain, what it must never do, and who checks it before it leaves the team. This is the judgment AI can't supply for you.

3. Share what works

Collect the prompts, templates and instructions that already work into one place the whole team can use. The person with the private prompt library becomes its editor.

4. Teach on real work

Run workshops on the team's own tasks, with their own documents, and leave with a finished piece of work and a written standard for it. People learn a tool faster when it solves their own Tuesday.

5. Measure the work

Track time to a first draft, rework and errors in the chosen workflows. Don't count logins or prompts. Usage tells you people opened the tool. The work tells you whether it helped.

An example

Illustrative example, not a Flygen client. A 40-person engineering consultancy notices that five people write proposals with AI, each differently, and the partners can tell which ones by reading them. The team picks proposals as its first shared workflow. In one workshop they agree on a structure, a list of claims that always need a source, and a rule that a partner reviews the pricing section by hand. The best prompts go into a shared folder. A month later, the proposals read like one company wrote them, and the partners review the substance instead of the style.

What to do Monday

• Ask five people from different teams to show you, on screen, how they used AI last week. Write down the tasks, not the tools.
• Pick the three tasks that come up most often and cost the most time.
• For one of them, draft a one-page standard and name the person who checks the output.
• Book a half-day with that team to rebuild the workflow on their own material.
• Decide what you will measure in a month's time, and write it down now.

Getting a whole company to work with AI is mostly a question of judgment: what to use it for, and what counts as good. Our Digital Transformation service starts exactly there.

Where this leadsDigital Transformation