Hero image to be produced. Suggestion: a table of MethodKit cards mid-workshop, shot in the Berghs yellow-on-white style
Your AI does not know how your organisation works. For most it has read the internet but none of the meeting notes. That is why the answers come back confident, generic and slightly wrong. This one-day intensive is about the missing half: where your organisation's knowledge actually sits, how to get one piece of it into a form a machine can use, and how to keep it current.
Many AI trainings teach you to write better prompts. This one starts a step earlier, with the material the model is working from.
Every organisation runs on knowledge that was never written down. It sits in the head of the person who has been there twelve years, in a chat thread nobody can find, in a folder nobody has access to, in a document that was true two years ago. An AI model cannot reach any of it. So it answers from the average of the internet instead, and you get something that sounds right and is not.
Most of the conversation about AI context treats it as a quantity: how much fits in the window. We treat it as something with a shape. There are different kinds of missing context and each kind has a different fix. Knowledge nobody has said out loud is not the same problem as knowledge that was written down and lost, and neither is the same as knowledge sitting right there but out of date.
The day is built on the MethodKit method: physical cards on the table, structured conversation in small groups, and a named output at the end of every exercise. The cards put the whole system in front of you at once, so a room of people can see the same picture and argue about the right thing. The frame is given, what matters inside it is yours to decide.
You work on your own organisation the whole day, not a case study, and you work on it out loud with other people. Sorting cards on a table is a methodical way to see every place context could live in your organisation, judge which of those places are worth the effort, and put them in an order you can act on. You bring a real problem. You leave with a context map: your own organisation laid out, the dark corners marked, and an order you can work through.
For people who have already used AI properly, hit the ceiling, and suspect the problem is not the tool. Communication leads, managers, strategists, consultants, operations people and founders. It suits anyone who is expected to answer for how AI gets used in their organisation.
It is not an introduction to AI. If you have not used ChatGPT, Claude or Gemini seriously for real work, take a foundation course first.
Map the context your organisation holds, and see where it is missing.
You leave with a context map of your own organisation. An artifact, not notes.
Co-Founder, Dark Context & MethodKit
Ola Möller is the founder of MethodKit and co-founder of Dark Context. He works where systems thinking, design and facilitation meet, building card-based tools that give teams a shared language for complicated work. With DW Akademie he built tools that help newsrooms around the world make sense of AI, and with UN Habitat he built a programme where young women shape the planning of their own neighbourhoods. He is based in Stockholm.
Assisted on the day by Andriy Zhukov, co-founder of Dark Context.
Contact us if you have any questions.
info@berghs.se