Use · team after training
AI adoption after training: turning knowledge into everyday work
Training gives knowledge, but rarely changes how people work on its own. We help companies move from training to uses that happen every week. With an accountable person, time to practise and measurement of who actually uses AI.
Starting point
The team has been trained and licences bought, yet a month later only a small share of people use AI regularly. Nobody owns what happens next.
What we do
We agree a criterion for regular use and measure the baseline. We choose uses on real tasks, run short reviews and describe working rules for the ones that work.
What we measure
The number of people meeting the criterion, uses with rules and an owner, and in the next stage the time and quality of work in those uses.
- In short
- After AI training three things are needed: a person accountable for what comes next, time to practise on real tasks and measurement of who really uses AI. In one of our deployments, after eight weeks 21 of 30 trained people were using AI regularly, against eight at the start.
- Proof from a deployment · AI adoption in a team
- The criterion for regular use was met by 21 of 30 people, that is 70% of participants. This is an observation after a whole set of measures was introduced, without a control group.
| Metric | Before | After |
|---|---|---|
| People using AI regularly | 8 of 30 | 21 of 30 |
| Uses with rules and an owner | — | 2 of 6 |
| Practice time provided by management | — | 1 h a week |
Measurement basis: Four-week task records for all 30 participants before the change and after eight weeks, in the same group. Criterion: at least one completed professional task a week whose result was checked and used.
Read the case study: Eight weeks after the training, 21 of 30 people were using AI regularly
Why does little change after AI training?
Training ends on the day ordinary work begins. Without an accountable person, time to practise and tasks from real duties, new knowledge does not make it into everyday work.
Knowledge is usually not what is missing. What is missing is a decision about who should use it and on which tasks, and a way to check that it is happening.
What do we do after training?
- A usage criterion. For example, at least one completed work task a week whose result was checked and used. Logging into the tool does not count.
- Baseline measurement. We base it on participants’ task logs, not on what people say.
- Ownership and time. The company names an accountable person, and management provides the tools and time to practise.
- Reviews on real tasks. We help define the expected result and how to check it, for materials, summaries or replies to clients.
- Rules or a stop. Uses that work get written working rules and an owner. Those where checking takes too long are stopped.
How to measure the effects of AI training?
More people using AI is the first step, not proof of savings. The next stage is measuring time and quality in the uses that have entered regular work.
Usage, quality and time metrics are covered in the guide How to measure the effects of AI training?
Training or an ongoing engagement?
We run training when the team lacks the basics — as part of an implementation, not as a separate product from a price list. Training alone rarely changes how people work, so we do not stop there. The ongoing engagement starts with choosing uses, ownership and measurement, and ends with solutions the team uses without us.
Working with a team after training is one of the uses we run as part of AI implementation for companies, delivered in Poland, in Polish.
Frequently asked questions
Do you run AI training for companies?
Yes, when the team needs it. We run it as part of an implementation, not as a separate product from a catalogue. The weight sits on what happens after the training: we work with the team on its real tasks, choose uses, run reviews and measure adoption.
How soon can you see results?
In the deployment described, the measurement after eight weeks showed 21 of 30 people using AI regularly, against eight before the change. Every team is different, so we first measure the baseline and agree the criterion.
Does more people using AI mean time saved?
Not automatically. The number of users shows whether the team has started working with AI. Time saved has to be measured separately, in specific uses.
What if a use does not work?
We stop it and record the reason. In the deployment described, one of six uses was set aside because checking the result took too long.