AI training
Practical training in the tools you already hold licences for, with exercises taken from the participants' own work. Afterwards you get a learning platform built around your business, so what you covered can be practised further. No lecture tour about the future of work.
Short answer
AI training for companies is practical training in the tools you already have, usually ChatGPT and Microsoft Copilot, with exercises taken from the participants' own work. The content is prompting technique, data hygiene and what must never be pasted into a public AI tool. The formats are a talk, a half day, a full day or a programme over time, and every one of them includes a learning platform built around your business.
We do not run a lecture tour. We build AI solutions for companies during the day and teach from the same experience: what works in production, what only looks good in a demo and where the line between the two runs.
It shows most in the questions. When someone in the room wonders whether AI can read their supplier invoices, the answer is not that it is an exciting possibility, but how it would be built, roughly what it would cost and why it still might not be worth it at your company.
You are also already paying for the tools. The licences exist in most organisations, but they are used by a few people and at a fraction of what they can do. The difference between a licence and an effect is that someone has shown how the tool is used in your particular work.
Four parts, in that order.
ChatGPT and Copilot in your own working day. Not a feature walkthrough, but the tasks that recur among the participants: summarise a long document, write a first draft, compare two documents, clean up a list, prepare a customer reply.
Prompting technique that holds. How to describe the task, give the model the right material, ask for a format you can use directly and recognise when the answer is invented. It is a craft with fairly few rules, and it can be taught in a morning.
Data hygiene: what may and may not be pasted in. Personal data, contracts, price lists, code and customer data all have different answers, and participants should leave with a rule they actually remember instead of a feeling that one ought to be careful.
How AI agents work and when they are needed. Where the line runs between a human using a tool and something running automatically in your systems, and which kind of task belongs on which side of that line. It makes it easier for you to see what the next step is worth.
Four setups. All of them can be held on site with you or over video, and all of them include a learning platform built around your business.
01
A walkthrough for the whole company or for the management team: what the tools can do today, what they cannot do and what applies around security. Up to around 30 participants. Works well as a shared starting point before you decide on something bigger.
02
Hands-on in the participants' own work. Everyone brings something they actually do every week and leaves with it solved. Up to around 12 participants, for the simple reason that everyone should have time to get help with their own.
03
A workshop in the morning and work on your AI policy in the afternoon: which tools are approved, what may be sent into them and who decides. You leave with a draft in hand instead of an item on a to-do list.
04
Three sessions with your own work in between and a follow-up a month after the last one. The format exists because that is where the habit sets in. A one-off day gives you enthusiasm, a programme gives you use that is still there after the quarter turns.
After every training you get a learning platform built around your own business. It holds the material we covered, your processes and your examples, the explanations behind why something works the way it does and exercises to practise on. It is always there to go back to, both for those who attended and for those who join you later.
The reason is simple: a talk that is not followed up gets forgotten. What is still there a month later is what you have done yourself, several times, in your own work. The platform is the difference between having heard something and knowing it, and it lets you as a company hand the training over to your staff instead of hoping it stuck.
It is included in every format and adapted to your business. The depth grows with the setup: a talk gives you the material and the basic exercises, a programme gives you a platform that is built on between sessions and follows your own workflows.
Four steps. The preparation is what decides whether the training feels relevant or generic.
01
A short conversation about what you do, who will attend and which tools you already hold licences for. We also ask what takes the most time in the everyday, since that is where the exercises should sit. The call is free and takes half an hour.
02
We swap the exercises for your tasks and your examples. A staffing agency and an installation firm do not get the same morning, even though the fundamentals are the same. If you have an existing AI policy we start from it instead of presenting a generic one.
03
On site with you or over video. Participants work in their own tools on their own tasks, and we walk around and solve what gets stuck. Anything we do not get through on the day is collected and answered in writing afterwards.
04
You get the material, the example prompts and the draft guidelines in writing. The programme format also includes a follow-up session a month later, and by then the questions are almost always completely different from the ones on day one.
Riktpriser
Indicative prices per session, regardless of the number of participants within the limit of the format. Travel and accommodation outside Stockholm are added and agreed in advance.
Indicative prices excl. VAT. Your setup is always quoted at a fixed price before we start.
Every format contains a block on data hygiene, and that is the part we do not shorten. It is also the part with the fastest effect: an organisation where everyone knows what not to paste in has reduced its risk more in a morning than most policy projects do in a quarter.
We go through what never belongs in a public AI tool: personal data, customer data, contracts, price lists and anything covered by confidentiality. We also go through the difference between an account you pay for and a free one, since the terms for what may be used for model training are not the same.
Afterwards you get a draft set of internal guidelines: which tools are approved at your company, what they may be used for and who to ask when you are unsure. A couple of pages you can decide on, not a framework to be implemented.
Four groups we train most often, and what each of them needs to get out of the session.
01
Needs to tell the difference between what is possible today and what was said at a conference, above all in order to say no to the right things. The talk format is usually enough, ideally with time for questions about your own plans at the end.
02
Sits closest to the workflows that can be automated and notices the difference fastest. The half day gives the most here, with the participants' own documents as exercises. It also tends to be where the next automation project is found.
03
Writes a lot of text under time pressure and has the most to gain from drafting, summaries and research ahead of meetings. Data hygiene matters extra in this group, since customer data is the thing closest to hand to paste in.
04
When the goal is a shared foundation and shared rules rather than specialist knowledge. A common combination is a talk for everyone plus a workshop for the departments most affected, a few weeks apart.
We run the training ourselves, no hired speaker. Caesar Katende works on agent architecture and orchestration, Ted Wachtmeister on backend, data and ML as well as operations. Both studied at Lund University, and both build AI solutions that run in production at companies.
The same people all the way: the person who draws the strategy writes the code and runs the training. Nothing gets lost in a handover.
It means the examples come from real builds, including the ones that turned out harder than expected. Participants can also ask about their own case and get a straight answer on whether it can be automated, roughly what it would cost and whether it is worth doing.
Common questions
No. We start from zero and assume that several people in the room have tried ChatGPT a few times and then stopped. Those who are already used to it get the most out of the workshop part, where they work on their own tasks at their own pace and can go deeper.
The talk works for up to around 30 participants. We keep the workshop formats to around 12, since everyone should have time to get help with their own work. If there are more of you we split the group across several sessions, and the price is then per session.
Both work. Talks work well over video, while workshops are usually better on site since we can walk around and look at the screens. We are based in Stockholm and are happy to travel; travel and accommodation outside Stockholm are added.
Yes, always. We align beforehand and swap the exercises for your own tasks. Participants bring something they actually do every week, and the goal is that they have solved it before they leave.
You get a learning platform built around your business, with the material from the training, your own processes and examples and exercises to keep practising on. It stays there and works just as well for those who did not attend, new hires for example. The platform is included in every format, and the depth grows with the setup you choose.
Yes. Every delivery ends with a walkthrough for the people who will use the solution, and it is included in the price. Standalone training is for when you want to lift the whole organisation's use of the tools and not just one workflow.