AI adoption for teams, the complete guide

AI adoption for a team sticks when it starts from how people already work, rather than from a tool. The teams that get real, lasting use out of AI follow the same four-part method: understand each person's starting point, agree the rules together, work the change in a weekly rhythm, and give leaders evidence of what is happening across the team, not a watch list of who is using what.

This guide pulls together everything Working With Me has published on AI adoption into one place: why most rollouts fail, the four steps that make adoption stick, how the method changes for a neurodivergent team, and how to set it up. Read it straight through, or use the links to go deeper on any one part.

Why most AI rollouts fail

Most AI rollouts start with the tools and hope people catch up: a licence rollout, a training day, an email telling everyone to get on with it. MIT's 2025 report The GenAI Divide found that 95 percent of enterprise generative AI pilots deliver no measurable return, and named a learning gap inside organisations as the real barrier, not the technology itself. We cover the research in full in why AI rollouts fail, and how to make adoption stick. The short version: a tool given to a whole team at once, with no plan for how different people actually learn and adopt change, gets used by the people who were going to try it anyway and ignored by everyone else.

The four steps that make AI adoption stick

• Understand your people. Find out where each person actually stands before deciding what they need.
• Agree the rules together. Write down what AI is for, what stays human, and what gets checked, as a team.
• Work in a weekly rhythm. Small experiments, tried and shared, turn a training day into a habit.
• Show the evidence, not the individual. Leaders see the pattern across the team, never a log of one person's activity.

None of these four steps works well on its own. Understanding people without agreeing rules leaves everyone guessing what is allowed. Rules without a rhythm sit in a document nobody reopens. A rhythm without evidence leaves leaders unable to tell whether any of it is working. The method holds together because each step feeds the next.

Step one: understand where each person actually stands

In any team, someone is already using AI for most of their work, someone tried it once and stopped, and someone has not touched it at all. Guessing which is which, or assuming everyone is in the same place, is where most rollouts go wrong first. A Working Profile, each person's own working with me manual, captures their AI confidence and the support they need alongside how they like to communicate and learn. It is written by the person themselves, not assessed or scored by anyone else, so it tells you who needs a worked example, who wants to experiment alone, and who is ready to show the others.

This step matters even more for a neurodivergent team, where the range of starting points and preferred routes in is often wider. AI adoption in a neurodivergent team, getting it right for every brain goes into that in full, including the different routes in that suit different ways of learning and the case for treating every team as a neurodiverse one.

Step two: agree the rules of the game together

Rules written by the people who will use them get followed. A policy written by nobody the team knows gets ignored while the actual use of AI carries on unofficially around it. How to agree the rules of AI at work with your team sets out how to write them in one working session and keep them to a single page: what AI may be used for, what never goes into an AI tool, how AI-assisted work gets checked before it goes out, and who owns the result.

The agreement should live somewhere the whole team can see it, not in a folder nobody opens. That is what turns it from a document into something people actually check when they are unsure.

Step three: work the change in a weekly rhythm

A single training day fades by Friday. What makes adoption stick is a weekly rhythm of small experiments: try one thing, share what worked and what did not in the team check-in, and keep whatever earns its place. An experiment that gets dropped after a week is not a failure, it is the rhythm doing its job.

This is the same rhythm that makes any change stick in a small team, not a special process invented for AI. If a weekly check-in is already part of how the team works, AI experiments simply become one more thing raised there alongside everything else.

Step four: show leaders the evidence, not the individual

Leaders need to know whether AI adoption is actually working, and the honest way to answer that is to look at patterns across the team, not at what any one person has done. Evidence should show whether experiments are happening, what is being kept, and where support is still needed, at team level, without a log of individual activity.

The goal throughout is better work, not more AI use for its own sake. An experiment that gets dropped because it did not help is exactly as much progress as one that gets kept, and the evidence should be able to show both.

Setting it up

The fastest way to put this method to work is the Working With Me workshop, from £2,000 with a four-week platform trial, where a team builds its working manuals and agrees its rules in a single facilitated day. The AI adoption module applies the same Working Profile, agreed rules, and weekly rhythm specifically to AI, and Working With Me keeps the rhythm going afterwards, so adoption stays a habit rather than fading back into a one-off training day.

Common questions