Insights
How to actually get your team using AI
Getting your team using AI means turning tools they already have into daily habits. You pick a few high-value tasks, train people on their own work, set light guardrails, and track who is actually using it. Most teams have the tools. The gap is confidence and a clear reason to start.
Why is my team not using the AI tools we pay for?
Most teams under-use AI because nobody showed them how it fits their job, not because they refuse to. People are busy, the tools feel generic, and a single lunch-and-learn does not change how anyone works on Monday. The result is a paid licence that sits idle while a few keen people quietly use their own AI in the gaps.
- No clear, role-specific reason to open the tool
- A one-off demo with nothing to practise on afterwards
- Worry about getting it wrong, sharing the wrong data, or looking slow
- No visible example from someone in the same role
What does an AI adoption playbook look like?
A practical AI adoption playbook is a short, repeatable sequence: find the work worth automating, pick the people to start with, train on their real tasks, set guardrails, then measure use and keep going. It is run in weeks, not quarters, and it starts narrow so people get a win fast and tell their colleagues.
- Step 1: Map the tasks that eat the most time in each team
- Step 2: Pick two or three starting use cases with an obvious payoff
- Step 3: Choose a small first group, including one respected sceptic
- Step 4: Train on their own work, not on generic examples
- Step 5: Set light guardrails so people know what is safe
- Step 6: Measure weekly active use and share the wins
How do you pick the first AI use cases?
Pick use cases that are frequent, time-consuming, and low-risk to get wrong. Drafting first replies, summarising long threads, cleaning up notes, and prepping research all qualify. Avoid anything that touches sensitive data or a regulated decision in week one. A good first use case saves a real hour this week and is easy to copy across the team.
How should you train your team to use AI?
Train people on the tasks they do every day, with their own files and their own workflows open in front of them. Skip the theory about how the model works. Give each role a handful of prompts that map to real jobs, then have people practise live until they get a result they trust. Comfort comes from reps, and reps need more than a single session.
- Run hands-on sessions by role, not one generic all-hands
- Give people a short prompt library tied to their actual tasks
- Practise on live work, so the value is obvious that day
- Aim for more than five hours of practice over the first month
What guardrails do you need before rolling out AI?
You need a one-page set of rules people can actually remember: which tools are approved, what data must never be pasted in, when a human has to check the output, and who to ask when unsure. Keep it short and friendly so it removes fear rather than adding bureaucracy. Most hesitation comes from not knowing what is allowed, so saying it plainly speeds adoption up.
How do you measure whether AI adoption is working?
Measure weekly active users per team, the tasks people now run with AI, and the hours those tasks used to take. Adoption is a usage number, not a licence count. Review it every week at first, celebrate the people leading the way, and feed what is working back into training. When usage stalls, it usually means the next use case is unclear, so go find it with the team.
- Weekly active users by team and role
- Number of real tasks now run with AI
- Estimated hours saved on those tasks
- Where usage stalled, and the next use case to unblock it
What the research shows
Most employees already bring their own AI to work, usually without guidance, so the demand is there before any rollout starts.
Comfort using AI nearly doubles after structured training, which is why training, not access, is the real lever for adoption.
Employees rank training as the single most important thing they need to adopt AI, ahead of any new tool.