Nortakai Method: Stop your AI starting from scratch in every conversation
Your team already uses AI, but each person in their own way. We put what your company knows in order, so every conversation, tool and person starts out knowing their part.
- What it's for
- So your team doesn't have to explain everything over and over. We measure it before and after, with your numbers.
- How it starts
- With the context diagnosis: three weeks, at a fixed price, and you run it yourself.
How much of your team's time goes on explaining everything all over again?
- The same instructions, copied in several places.
- The right document has five versions, and the AI reads the wrong one.
- Every new conversation has to be explained again.
- Nobody knows what was decided, or why.
We have not found a reliable figure from outside for a company like yours. The diagnosis measures it with your own numbers.
What the official statistics say
Among small Spanish companies (10 to 49 people) that have considered using AI but do not use it, 84% cite a lack of relevant expertise (across the EU, 71% of those with 10 to 49 people) and 48% cite difficulties with the availability or quality of the necessary data.
Eurostat, 2025. Weak evidence.
We put the AI your team already uses in order
Not one more tool: a way of working, the same one we use.
- One single source for each thing.So what is current is found first time.
- A profile per role.Each person starts out with their AI knowing their part. It replaces no one.
- Handover, in order.So nothing gets lost between conversations, tools and people.
- Decisions, with their reasons.And a person's sign-off on anything that goes outside, costs money or affects a customer.
Three steps and a measurement
Context diagnosis · 3 weeks.
Where your team's time goes and which three things to do first. It can be hired on its own.
Implementation · about 4 weeks.
Five modules, in your account and by your hands: we design, review and teach.
Measurement · 4 weeks.
Real use, measured as in the two diagnosis weeks when we touch nothing.
And then, every month, for at least 6 months, a health report on context and AI spending, if your console exports it, and a short session.
These are our own estimate: where they come from
Durations for one area in a company of 10 to 50 people, with one person on your team as the context owner. They are our own estimate: we add up what each module takes and how it works for us, our only case so far. The first project measures them.
Your answer, with your numbers. You decide what reaches us.
- You run it yourself, on your own computers, and it only extracts figures: sizes, dates and counts. It never sends the text of your documents or conversations.
- You review what comes out before sending it to us. Lists with file names stay with you.
- It measures your AI spending if your console exports it; if not, we tell you why.
What it opens and what it doesn't
By default it does not open any file. With two options that you switch on, it reads your AI instructions on your computer and looks for exact copies, and what you need to act on stays in a folder of yours. You can check that what you run is what we published: its fingerprint is further down this page.
Without tying you to us
- Your profiles, rules, decision log and reports stay in your own files. If you stop, what is implemented and paid for keeps working; you just stop getting new versions of the method.
In order on the legal side too
- The training for Article 4 of the EU AI Act is included: three sessions and an internal record that helps you keep evidence of your AI literacy measures.
- A one-page AI use policy, approved by your management.
- We do not store your credentials. If during the implementation we see or process personal data of yours, we sign a data processing agreement first.
- We do not build profiles that assess or select people.
Questions we get asked
"We already use ChatGPT or Copilot."
The organisational part works with any tool: single sources, role profiles, handover and the decision log. Our reference is Claude; with another tool, we adapt the profiles within the same project. If you have OpenAI team accounts and want the diagnosis to read their usage, tell us what your console exports and we will tell you whether we can read it. It would be the first time.
"Everyone uses their own personal account."
We do not ask you to ban it overnight. The diagnosis asks, in an anonymous survey, which tools people use, including on their own accounts. With that, the method puts it in order: a profile per role, a one-page policy and a list of what needs a person's sign-off.
"We don't have time."
Before we start, we tell you how much time we will ask of your team. We estimate that the context owner puts in 2 to 4 hours a week during the implementation, and each role lead about 4 to 6 hours in total. It is our own estimate, based on what each module takes, and the pilot measures it. If you can't right now, we believe it is better not to start.
"What about our data?"
You run the diagnosis yourself, and only the figures you review reach us. We work in your account, with your hands on the keyboard.
"What is the return?"
We have no customer figures and we do not make them up. We measure your case before and after, with the same definitions: see how.
"You are a small company."
That's true. The first project is a pilot. We run one implementation at a time, with a fixed schedule in writing, and the method is documented so that you can run it yourself.
Start with the context diagnosis
With the diagnosis you decide whether to go on, which modules and in what order.
Verify the diagnosis
Nortakai's context diagnosis runs on your own computer. To check that what you run is what we published, open a terminal in the diagnosis folder and type python -m nortakai.paquete. The last line must show the same fingerprint as this one.
- Version
- Context diagnosis v1 (diagnostico-v1)
- Published
- 31 October 2026
- Package fingerprint (SHA-256)
d96eabc124791ed6277edf9f4be245f517775233ed65070682c1288927c79556
If it doesn't match, don't run it and write to us at contacto@nortakai.com.