The most common AI training mistake is giving everyone the same session. A finance assistant, a sales manager and an operations director need very different things. A better approach is to train in layers: safe-use basics for everyone, practical task-specific training for daily users, deeper skills for champions, and change leadership for managers.
Layer 1: Everyone (safe and confident basics)
What they need:
- What AI tools can and cannot do, in plain English.
- Which tools are approved.
- What information must never go into AI tools.
- How to check AI output for errors.
Format: a short, practical session plus a one-page guide. This is about safety and confidence, not mastery.
Layer 2: Daily users (task-specific practice)
What they need:
- How to brief AI well: context, role, examples and format.
- Hands-on practice on their actual tasks, such as their reports, emails or analysis.
- Refining weak answers rather than starting again.
- A personal or team prompt library.
Format: two or three short, hands-on sessions over a few weeks, using real work. This layer delivers most of the productivity gains.
Layer 3: Champions (depth and support)
What they need:
- Advanced prompting and comparing tools.
- Designing AI-supported workflows.
- Helping and coaching colleagues.
- Spotting new opportunities and testing them.
Format: deeper training plus time to experiment and share.
Layer 4: Managers (leading change)
What they need:
- Confident personal use, so they lead credibly.
- Identifying where AI can improve their team's work.
- Running small experiments and measuring results.
- Handling resistance and talking honestly about roles.
- Setting team rules and scaling what works.
Format: a structured programme combining hands-on use with change leadership. This is the layer most often missed, and the one that determines whether the other layers stick.
Training mistakes to avoid
- Theory-heavy sessions about how AI works, with little practice.
- One-off events with no follow-up.
- Tool-only training that ignores safe use and quality checking.
- Training without time to practise, expecting learning on top of full workloads.
- Skipping managers, who then cannot support their teams.
How to know training is working
- Weekly use of approved tools rises and stays up.
- People bring their own new use cases.
- Quality of AI-assisted work is consistent.
- Fewer questions about basic rules.
For a practical session plan, see how to train your team on AI tools, and for prompting skills, learning prompt engineering for business. Our Leading AI Innovation & Organisational Transformation course covers Layers 2 and 4: hands-on AI skills for your own work and the skills to train and lead your team.
Frequently asked questions
How long does basic AI training take?
Safe-use basics can be covered in an hour or two. Practical, task-specific skills usually need a few sessions spread over several weeks.
Should AI training be compulsory?
Safe-use basics should be, because of data risks. Deeper training works better when people can see how it helps their own work.
Who should deliver AI training?
A mix works well: external training for managers and champions, then internal champions delivering task-specific sessions to their teams.