The short answer.
People use AI at work when it does a named piece of their own work the way their best colleague does it, and when their manager uses it too. Training sessions and encouragement do not change what people do on Monday morning; a workflow built from the team’s own method does. Involve the people who do the work in writing it, start with the piece of work they like least, make checking the output part of the job and not extra, and never tie it to watching or rating anyone.
Training does not change the work. A workflow does.
Training teaches people what an AI assistant can do in general, and general use fades because nothing in the job asks for it. People leave a session with a few good prompts and go back to the way the work was done before.
A workflow changes the work. A workflow is your best way of doing one piece of work, written down precisely enough that ChatGPT, Claude, Copilot or Gemini prepares it the same way every time. When the month-end reconciliation, the client brief or the new-hire packet arrives already prepared in the team’s own format, nobody needs encouragement to use it.
- Named.One piece of work with a name the team already uses, such as “the weekly pipeline note”, not “use AI for sales”.
- Their own.Built from how the team’s best person does it, so the output looks like their work and not a stranger’s.
- Where they work.In the company’s own assistant, on its own account, run by its own people. Nothing new to log in to.
How to roll out ChatGPT, Claude, Copilot or Gemini across a company covers the seats and settings; this page covers the people.
Build it with the people who do the work.
People use a workflow they helped write, because it holds their judgment and they know where it came from. Four moves make the people who do the work its authors.
- Ask how the work is done.Sit with the person who does the work best and ask what they start from, the order they work in, what they check and what they leave out.
- Start with the work they like least.Every team has a piece of work everyone puts off: the same report rebuilt each week, the chase emails, the form copied from one system to another. A workflow that removes that piece wins the team in its first week.
- Make checking part of the job.A workflow prepares; a person decides. Write the check into the workflow as a step, with the reviewer named, so reviewing the output is the work and not an extra task on top of it.
- Let the team change it.Every correction a reviewer makes is logged and goes back to the workflow’s owner. People see their fixes in the next version and trust the workflow more each month.
Build your AI plan with the people who do the work shows how to bring each team into the plan itself.
Managers use it first.
A team uses AI when its manager visibly uses it on real work. A manager who asks for AI use and still does every task by hand tells the team the request is optional. The manager runs the first workflow before anyone else, reviews its output in front of the team, and says what was corrected and why. Who should own AI in a company your size? sets out who leads above the team.
What about the people who are wary of it?
People who are wary of AI at work are usually wary of three things: being judged by it, being watched through it, and having their own way of working thrown out. Answer each one plainly and in writing.
- Nobody is rated or ranked.A workflow is measured by whether the work is used and useful, never by who used it most. Nothing about AI use goes into a performance review.
- Nobody is watched.You find out how work is done by asking the people who do it. Nothing records keystrokes or screens, and nothing monitors anyone. Asked or watched: two ways to find out how your team works explains the difference and why asking wins.
- Their method is the method.The workflow is written from how the team already works. The experienced person’s judgment becomes the standard the assistant follows.
- A person still decides.Anything that sends, pays or changes a record waits for someone to approve it.
Give the most careful skeptic the job of checking the first outputs. The skeptic finds the faults early, and the team trusts a workflow the skeptic has passed.
Six questions to ask your team before the first workflow.
A team lead asks these six questions in one team meeting, before anything is built. Copy them as they are.
- Which piece of work do you most wish you never had to do again?
- Who here does that work best, and what do they do differently?
- What does that work start from, and where does it come from?
- What has to be checked before it goes anywhere, and who checks it today?
- What would you never want an AI assistant to do with this work?
- How will we know, in a month, that it is helping?
The answers choose the workflow, write its steps and check, set its limits and say how you will judge it. Installed, used or useful? Three questions that show AI adoption gives the sheet for that judgment.
One team, from first question to daily use.
Illustrative example.
Take a 55-person property management company. The lead of its accounts team asked the six questions. The answer to the first was unanimous: the monthly owner statement letter, rewritten by hand for every property. The person who wrote the clearest letters walked the team through how she did it: which figures she pulled from the accounting system, which items she always explained, and what she never put in writing to an owner. That became the workflow. The team lead ran it first on her own properties and showed the team two letters she had corrected. The careful colleague who had doubted the whole idea became the reviewer for the first month and added three checks the workflow now runs every time. The team lead told everyone in the meeting that nobody’s use would be tracked or reported. Within the month, the team asked for the next workflow, for tenant renewal notices.
What is different in a company of under 200 people.
Most advice on AI adoption is written for large enterprises, with change programs, champions networks and training catalogs. A company of under 200 people gets people to use AI more directly, and faster.
- The manager is in the room.The person who leads the team also does its work, so going first is natural.
- One person’s method is the team’s method.A small team often has one person whose way of doing the work everyone already copies. Write that person’s method down and the workflow is half built.
- Trust is personal.People know who decided and can ask them. Say in a team meeting, in plain words, that nobody is watched or rated, and the promise holds because everyone knows who made it.
- One good workflow spreads.When one team uses a workflow every week, the next team hears about it at lunch. Word of mouth does the work a rollout program does elsewhere.
This is how a small company gets 10X the ROI of your existing AI: the same seats, used every week on named work that people chose.
Where to go next.
Getting people to use AI is the deploy step of a larger plan. How organizations plan their use of AI sets out all three steps: plan it, develop the workflows, deploy them. How to deploy AI workflows to a whole team, and keep them working takes the next step after the first team.
FullyPowered Planner is where the plan lives: you set out your teams, their roles and the work between them, and choose what to build first. Free workflow templates give each team a starting point, and we develop workflows with your team in a 3-Day Workflow Sprint or a 2-Week Workflow Sprint.