Applied AI training for teams

A workshop so your team uses AI on their actual work, with the house rules written down.

Not a theory course and not a tool demo: I work on the tasks your team does every week. They leave with the prompts that work for their own cases, with a rule for what may and may not be handed to an AI —here that is set by the GDPR and the European AI regulation— and with the hours saved actually counted.

Who it is for

Teams already using AI on their own, with no shared criteria
Practices and advisers handling client documentation
Businesses that want hours back without buying another tool
Small teams where nobody has time to research it

Why it ends up being needed

In almost every team there is already someone using artificial intelligence. Usually two or three people, each in their own way, not telling anyone because they are not sure it is allowed. The rest suspect they are missing something and have neither the time nor the appetite to go and research it themselves.

What does not work is the generic course. Four hours of textbook examples leave everyone entertained and on Monday nobody changes anything, because the speaker's example looks nothing like the report you have to produce every Tuesday. What you learn on your own work sticks; what you learn on somebody else's does not.

So the workshop is prepared around your cases and worked on your documents. People leave with the prompts that work for those specific tasks, with the rule written down about what may be passed to an AI and what may not — set here by the GDPR and the European regulation — and with a count of the hours it saves, so it can be checked next month.

What I deliver

Prior diagnosis — What your team still does by hand, and where AI genuinely saves hours.
Tailored workshop — Four hours on your own cases, not on textbook examples.
House rules — What may and may not be handed to an AI, on one page and in writing.
Prompt library — The ones that work for your tasks, tested in front of you.
Recording and materials — For whoever joins later, without running the workshop again.
30-day follow-up — A short session for whatever came up once they used it for real.

What is included

Diagnosis and a four-hour workshop
House rules and prompt library
Recording and materials
30-day follow-up

What gets decided before a line is written

Which cases get worked on — They are chosen before the workshop, in half an hour with whoever runs the area: the tasks that repeat every week and eat time. A workshop without that half hour ends up as a tour of tools, which is exactly what changes nothing on Monday.
Which documents people bring — Yours, pseudonymised. The first part of the workshop is how to do that, because it is the skill you need afterwards: replacing names, identifiers and account numbers before pasting anything. Working with invented examples teaches the tool and does not teach the judgement.
Where the line sits — It leaves the workshop written on one page: what may be passed to an AI, what gets replaced first and what never leaves. It is the difference between a team using AI with shared judgement and one where each person improvises their own boundary.
What gets measured — Before the workshop, how long each chosen task takes today is written down. Without that count, the only result of the workshop is the feeling that it was interesting, and that cannot be defended to anyone or repeated next year.

What is not included

Tool licences or subscriptions, contracted in your name.
Building automations into your systems: that is the AI and automation service.
Technical training on programming with models, agreed separately.
Ongoing support to the team beyond the follow-up session included.

What you need to have

Half an hour beforehand with whoever runs the area, to choose the cases.
Real documents from the team, pseudonymised during the workshop itself.
Up to eight people per run, because each one works on their own cases.
A room, or an online session, of four uninterrupted hours.

Frequently asked questions

Do people need to know how to code?

No. The workshop is for whoever does the work: administration, sales, accounts, customer support. If there are technical people on the team, that side gets covered separately and in more depth.

How many people fit?

Up to eight, because each one works on their own cases and beyond that it stops being a workshop and becomes a talk. If there are more of you, it runs in two groups.

What about the confidentiality of the documents that get used?

You work with your own documents, pseudonymised, and the first part of the workshop is precisely how to do that. I sign an NDA beforehand if you prefer.

Can people from different areas attend?

Yes, and it usually works well: administration, sales and customer support share more problems than you would think. What does not work is mixing very different levels of use without saying so beforehand.

What if the tools change in three months?

The judgement does not change: which data may leave and what a person reviews holds just as well with another model. Prompts are handed over versioned so they can be adapted, and at thirty days there is a short session to adjust whatever did not stick.

Is it any use if we already use AI daily?

Then the workshop shifts focus: less introduction and more judgement, a review of what you are already doing, and the compliance part, which is what tends to be missing when usage started informally.