How to train your team on AI: where to start, step by step

To train your team on AI, start by taking stock of how people already use it and by setting a few simple rules, then run short workshops on the real tasks of each role. There is no need to wait for a big training plan: a six-step approach over a few weeks is enough to build solid foundations.

Why train rather than let it happen

In many teams, AI is already there, often without any framework: one colleague uses an AI assistant on their phone, another pastes meeting notes into a free tool. The risk is not the use itself, but use without rules or method.

A lack of skills is also the main barrier to adoption. According to Eurostat’s page on AI use in enterprises, accessed on 22 October 2026, 70.89% of European businesses that considered AI without adopting it cite a lack of relevant expertise as a barrier (2025 data).

Step 1: take stock

Before training anyone, look at where the team stands. Three questions are enough, asked without judgment:

• Who already uses an AI tool, which one, and for what?
• Which tasks take the most time and feel repetitive?
• What worries people: confidentiality, mistakes, the future of their job?

The site’s AI level test makes a good starting point: everyone takes it in a few minutes and you get a sense of the skill gaps. Worries about jobs deserve a real answer; our article on what AI agents change for jobs sums up what the data says.

Step 2: set simple rules

One page is enough. It should state clearly:

1. Approved tools, and with which kind of account (personal or business).
2. Forbidden data: personal data about customers or employees, confidential information, passwords. Data protection law such as the GDPR still applies when you paste data into an AI tool.
3. The checking rule: any text produced by AI is reviewed by the person who uses it, and that person remains responsible. A model can produce a hallucination, meaning a false statement delivered with confidence.
4. Transparency: when to flag that a document was prepared with AI.

Our guide to which AI to use while keeping business data private covers the settings of the main tools.

Step 3: choose a shared tool

Training everyone on the same tool makes learning and sharing good practices easier. The main assistants offer team plans. According to the vendors’ websites, as recorded on 22 October 2026: ChatGPT Business at 25 USD per month, Claude Team (Standard seat) at 25 USD per month, Le Chat Team at 24.99 USD per month.

Assistant	Free	Paid from	Web search	Agent mode	Voice	Images
ChatGPT	Limited	$8/mo	Yes	Yes	Yes	Yes
Claude	Limited	$20/mo	Yes	Yes	—	—
Le Chat (Mistral Vibe)	Limited	$5.99/mo	Yes	Yes	Yes	Yes
Gemini	Limited	€4.99/mo	Yes	—	Yes	Yes

Data rules differ. According to the site’s profile of Claude, conversations on the Team and Enterprise plans are not used for training by default. For Le Chat, data is hosted in the European Union by default. Check these points on the official page before you sign up.

Step 4: train by use case, not by feature

This is the step that makes the difference. Instead of a general talk about AI, run short workshops where everyone works on a real task from their job: meeting notes, a reply to a customer, a summary of documents, a job posting.

A simple workshop format:

1. Demonstration on a concrete case from the team.
2. Practice: everyone repeats it on their own task.
3. Discussion: what worked, what disappointed, what needs checking.

Take time to explain what makes a good prompt, the instruction you give the AI: the role, the task, the context and the expected format. The prompt generator helps beginners write their first instructions.

Check yourself A colleague pastes the full customer list into a free assistant to prepare a follow-up email. What do you tell them?

That the file contains personal data that must not leave the tool approved by the company. They can ask the AI for a follow-up template with fields to fill in, then do the mail merge themselves in the usual software. This is exactly the kind of situation your one-page policy should cover.

Step 5: appoint champions and share

Find one or two motivated people per team. Their role: answer simple questions, report difficulties and maintain a shared library of prompts that work. A common page, organised by task, beats tips scattered across chat threads.

Assistants also let you save custom instructions: company background, tone of voice, usual formats. That saves explaining everything again in each conversation.

Step 6: measure and adjust

After a few weeks, run a simple review: which tasks really saved time, which were dropped, what mistakes appeared. Adjust the policy and workshops accordingly. Once the team is comfortable, you can consider more advanced automation, or even agents; our article on AI agents for business helps you spot the use cases that work.

Step	Indicative timing	Expected result
1. Take stock	Week one	Uses, needs and concerns known
2. Rules	Week one	One-page policy approved
3. Shared tool	Week two	Business accounts set up
4. Job-based workshops	Weeks two to four	Everyone masters one real case
5. Champions	From week three	Shared prompt library
6. Review	After one month, then regularly	Policy and workshops adjusted

These timings are only a guide and depend on team size.

Mistakes to avoid

A few pitfalls come up again and again when a team discovers AI.

• Imposing without explaining. A flat ban pushes usage underground; an obligation without support breeds distrust. Explain the reasons behind the rules.
• Training everyone the same way. A salesperson, an accountant and a technician do not have the same tasks. Tailor examples to each role, and allow a slower pace for people less comfortable with digital tools.
• Promising miracles. AI saves time on some tasks and costs time on others, when everything needs checking. Set modest expectations and beat them.
• Forgetting managers. If they do not use the tools themselves, they will struggle to judge their team’s usage and settle practical questions.
• Skipping learning time. Block dedicated slots in the calendar, or training will always come after the day’s emergencies.

For an overview of possible uses, role by role, see our hands-on guide to AI at work.