AI training: start with one everyday process, not a tool

I run AI training for companies, and one thing comes up in almost every conversation. A company wants its team to "learn to work with AI", and the training plan starts with getting to know a tool: what it looks like, what features it has, how to write prompts. After a few weeks, most people go back to their old ways of working.

The problem: training starts with the tool

When the tool is at the centre of training, people learn a general skill that they have nowhere to apply. They see that the tool can write a text, summarise a document or answer a question. That is interesting, but the next morning they are back to specific tasks, deadlines and clients. There is a gap between "I know what the tool can do" and "I know what to do with it in my own work", and that is exactly where the benefit of the training disappears.

The second problem is that tools change quickly. If a team has learned the layout of a particular tool, the next update makes part of that knowledge obsolete. Understanding how to organise your process so that AI can help with it, on the other hand, stays valid.

What this looks like in a company

A typical example: requests come into the company by email. Someone reads them, copies the data into a spreadsheet, prepares a reply, and then enters the same information again in another system.

Before:

in the training, everyone learns the tool's capabilities together using generic examples;
everyone tries to do something in their own way;
there is no shared result, and it is unclear what will change at work;
after the training, two or three enthusiasts use the tool and the rest do not.

After:

the training starts with a question: which process takes the most of your time or causes the most errors?
the team maps it out together, step by step;
you choose the steps where AI can help and the steps where a person must stay in charge;
you get to know the tool only as much as this process requires;
by the end of the training, the team has a working solution they can use the very next day.

The difference is that people are not learning "about AI" but solving their own work problem with the help of AI.

The solution: one process, one result

The practical approach is simple, and you can use it either with me or on your own.

Choose one process. Not the whole company, not a whole department, but one repeatable activity: handling requests, preparing a quote, summarising a meeting, answering the most frequent customer questions.
Map it out. On paper or a whiteboard, write down every step, who does it and how long it takes. Even at this point it often becomes clear where there are unnecessary actions.
Find the steps where AI helps. These are usually drafting text, gathering information, classifying, and transferring data from a document to a spreadsheet.
Define the human role. Every result needs a person who checks it. AI prepares, a person approves.
Test it on real but safe examples. Avoid entering sensitive data into tools whose data security the company has not assessed.
Write down the instructions. If the process works, document it so that another colleague can repeat it too.

Only once one process works does it make sense to move on to the next. This way the team builds up experience step by step, and knowledge of the tools appears as a by-product.

What this brings

When training starts with a specific process, the benefits are clear and verifiable:

A clear result. The team sees what has changed at work, not just that "we learned something new".
Greater engagement. People are more willing to learn when it makes their own daily work easier.
Less resistance. Concerns about AI fade when a person sees that the tool takes away the boring part, not their job.
Better organised processes. Even if AI ends up not being used, mapping the process helps to find unnecessary steps and unclear responsibilities.
A basis for next steps. One well-developed process becomes a model that can be adapted for other teams.

It is also important to be honest about the limitations: not every process suits AI, and sometimes the process itself has to be put in order first, and only then technology added.

Where to start

Before any training, run a small check. Ask team members to write down three tasks they do every week that seem repetitive or tiring. Collect the answers and look for what recurs across several people. This is the best candidate for the first training.

Then choose a process using three criteria:

it recurs regularly;
a mistake in it is not catastrophic, and the result can be checked;
there is a person who knows the process well and can talk about it.

On this basis you can build either a short workshop or a longer training programme. If you need help setting this up for your company, get in touch and we will choose the first process together.

Which AI skill should your team learn first?