How to prepare your team for an AI workshop so it is not a wasted day
The difference between a workshop after which people work differently and one that leaves behind a certificate is made before the workshop. Here is what to prepare.
1. The participant survey (10 minutes per person)
I send it a week before the workshop. You can use these questions when organising training with someone else, or on your own.
- Your role and what you do day to day, in three sentences.
- Which programs do you use most (email, Excel, CRM, accounting system, other)?
- Have you used AI tools already? Which, for what, on which account (personal/business)?
- Name 2–3 tasks that take up most of your time or wear you down most. The more specific, the better.
- What worries you about AI at work? (Data, mistakes, being replaced, “I won’t keep up”.)
- What would have to happen for you to say afterwards that the workshop was worth your day?
Question 4 is the important one. The answers become the exercises. Question 5 tells me how much time to spend on safety and how much on defusing fear.
2. Choosing tasks: 2–3 per person, no more
A good workshop task is repetitive (you do it every week, not once a year), has a clear output (a table, an email, a report, a list) and can be started and finished in half an hour. A bad task is “I want to understand AI better” or “automating the whole order process”. The first has no output; the second is a project, not an exercise.
| Instead of | Better |
|---|---|
| “Faster customer service” | “Replies to the 10 most common delivery questions, in our tone” |
| “Reports” | “The monthly sales report by region, currently 3 hours from three files” |
| “Contracts” | “Comparing a supplier’s contract with our template, a list of differences” |
| “Excel” | “A formula that calculates commission from a table of thresholds, currently done by hand” |
3. Anonymised sample files
Every task needs a real file: an email, an invoice, a spreadsheet, a contract. Real in layout and content, but without personal data or trade secrets. The instructions I send:
- Replace names, customer company names, addresses and numbers (tax IDs, national ID numbers, bank accounts, phones) with made-up ones. Keep the format: a tax ID still has the right number of digits, an email still looks like an email.
- Amounts can stay, or be multiplied by a constant factor if they are sensitive. They must still add up.
- One file per task is enough. Two if the task is a comparison.
- One person in the company collects the files and checks them before sending. I do not take them away after the workshop.
If anonymisation seems like too much work, that is the first lesson: AI will do it in half a minute, provided you do it thoughtfully the first time. I show this at the start of the workshop.
4. Accounts and permissions
- Business accounts in the tool you will use (Claude, ChatGPT, Copilot, Gemini), created and tested the day before. Logging in during the workshop eats half an hour if someone has forgotten a password.
- Business plan settings: training on your data switched off, chat history as you decide, an administrator named.
- If the workshop involves company data (reporting, CRM): a separate read-only account, sensitive fields excluded, created a week in advance and tested. Not an administrator account.
- Equipment: everyone has their own laptop and network access. With one computer between two people, one person works.
5. Organising the day
Up to 12 people in a group. The first part together, then work in groups by role: accounting at one table, sales at another. A decision-maker present at least at the start and at the end, when we build the list of ideas to implement. No meetings in between; “I’ll pop out for half an hour” ends with missing your own exercise.
6. The week after the workshop
- Everyone uses AI on the tasks from the workshop, at work, not “when there is a spare moment”. The spare moment never comes.
- One person in the department collects the prompts that work in a single document. An hour a week.
- Questions and mistakes go to the trainer by email. With me, that is 30 days after the training.
- Management reviews the list of ideas from the workshop and decides: what the team does itself, what is worth quoting from a developer.
- After a month, a short meeting: what worked, what to adjust, whether to extend to other departments.
The survey, the task selection and the anonymisation instructions are the first things I send after the call. If you want to start there, get in touch and I will send the survey form before we even fix a date.