Samture Academy
Most people stop halfway: an assistant that writes faster. This programme takes you through the second half — an ontology, a semantic layer, a knowledge graph over your own material. That is where a system stops guessing and starts answering from facts you can point at, and where a working model of your organisation becomes possible at all.
Eight weeks, and then the deeper eight. Online, with one live session a week. Madelein Leegwater leads the programme herself. Enrolment is not open yet.
Where this sits
The Samture Excellence Center is not an institute in the making. It is the combined building capacity of Samture and its delivery partners: some 250 engineers, architects and data specialists who are working on foundations for organisations across the Gulf today. What makes them one whole is not a logo but a standard.
The demand in the region is larger than that capacity, and it is growing faster. So we bring people up to that standard quickly: builders from the partner organisations, and through the Academy people who want to learn to build it themselves. Both routes end at the same bar. Whoever clears it, builds.
Who this is for
The craft is called knowledge crafting, and it is not the property of one profession. The people who turn out to be good at it come from marketing, policy, operations, finance, research and engineering. What they have in common is not a background but a daily practice: they work with information where it matters whether it is right, and they want to put something up themselves that works.
The standard measures what you have built, not where you came from. That does not make it gentler but stricter: a background you have, a working foundation you have to show.
Part one asks for no programming experience. You set up your own working environment, step by step, with guidance alongside. Part two goes deeper and becomes more technical — that is where you learn to model and to query. Anyone who finishes part one knows for themselves whether they want to take that step.
What it leaves you with
AI inside your real work, not beside it. Your context, your naming, your rules, your data landscape, and permissions set correctly from the first connection.
It stops being a chat window and starts doing work.
The same system, given meaning: an ontology of your own domain, semantics that hold across departments, and a knowledge graph built over your own material.
It stops guessing and starts answering from facts you can point at.
Once meaning and provenance are in place, a living model of how the organisation actually works becomes buildable: the ground a digital twin stands on.
Not part of the programme. It is what the programme makes possible.
What you actually do
Each theme is four lessons: one that sets the question of the week, two that carry the material, and one worked session where we build it together. Around forty minutes of video, plus your own building — reckon on roughly four hours a week, most of it on your own system.
You build from the first week, and you begin at the bottom: identity and permissions first, then connections, then behaviour. What you put up in that first week is not a demonstration but the bottom layer.
The working environment itself, in the order a foundation is laid: identity and permissions first, convenience second.
You end with a running environment, the same shape for everyone.
The questions your system has to be able to answer, collected from the people who do the work rather than invented at your desk.
You end with fifteen to twenty-five questions that become your measuring stick.
Data landscape, classification, residency, landing zones. The theme nobody else teaches, and the one your security officer cares about.
You end with a data plan you can hand to your own CISO.
Things and relationships. What is an object and what is merely a property of one, and why “relates to” is never good enough.
You end with an entity and relationship map of your own domain.
Context, conventions and memory that survive between sessions, so you stop re-explaining your own organisation every morning.
You end with a system that knows how your work is put together.
Your mail, your files, your systems — with least privilege from the first connection, and anything that sends, deletes or pays asking first.
You end with a connected environment with rights you can defend.
The first real task that leaves your hands. Not all of it: the part the machine does well, with you where the judgement sits.
You end with one working automation on your own process.
Provenance and auditability. Where it came from, when, on whose authority — and where you draw the line on what an agent may decide.
You end with an auditable system, and the wall where part two begins.
What “meaning” actually means
The moment you can say that two mentions are the same thing, you stop having text about things and start having things you can ask about. One ordinary sentence, climbing that ladder:
As prose
“The services agreement with our Jebel Ali logistics supplier was signed on 7 July and runs for two years.”
A person can read it. A machine can find the characters.
As a value
2026-07-07
True, and useless. Nothing owns it.
As information
signedOn: 2026-07-07
Now it means something — inside one document, and nowhere else.
As knowledge
Agreement → Organisation → Person
The date belongs to an agreement, which belongs to a supplier, which has an owner. Nobody wrote that down. It follows.
The question you could not ask beforeWhich supplier agreements expire before December, and who owns them?
| Agreement | Expires | Owner |
|---|---|---|
| Logistics — Jebel Ali | 2026-11-14 | Procurement |
| Data platform — renewal | 2026-11-30 | Architecture |
| Managed hosting | 2026-12-01 | not recorded |
Look at the third row. When the fact is missing you get an empty field, not a confident guess. That is the difference this is built on: an answer you can point at, and a gap you can see.
It fits what you already have
The lessons are recorded in Claude Code, because that is what we build on every day. The method transfers: Codex, Gemini CLI, Copilot and whatever replaces them next year work the same way, because the foundation sits underneath the assistant rather than inside it.
Where a step genuinely differs by platform, both routes are shown instead of one being assumed — including the parts where Windows and Mac disagree about where things live.
Microsoft 365, Google Workspace, or the mix most organisations really have. You connect what you use, and you decide per connection what it may and may not do.
That is deliberate. Samture has no stake in which product you end up buying, and a foundation that only holds on one vendor’s stack is not a foundation.
The bar
Most certificates in this field are issued for finishing the material: watch enough of it and one arrives. That measures consumption and prints competence.
The Samture competence standard is the same standard every builder working on a Samture foundation is measured against. There is one, and it holds for both routes.
How it runs
Madelein leads the programme and is there every week. On the parts where it becomes genuinely technical — modelling, the semantic layer, connecting real source systems — specialists join her: people who do that work daily and bring in what a course otherwise misses. What it looks like when the data does not agree with itself, when the permissions turn out to be wrong, when the client changes their mind halfway through.
The cohort itself is the third source. Everyone builds in the same shape, so you can read each other’s work and say something useful about it. That is why the shape is the same for everyone, and why the group stays small.
Two minutes
Seven statements. Answer them as things stand today, not as you intend them to be. The result tells you honestly where you would start. Nothing is sent anywhere: the score is worked out in your own browser.
Your result
Nothing is sent anywhere. The score is worked out in your own browser.
These seven are about where you would start. For the same question about your organisation as a whole, there are twenty-four on the Capability Index.
Enrolment is not open yet
Look inside before it opens.
The first cohort is being formed
The first cohort is small. Dates, format and pricing are not yet fixed.
Leave your details and you will be the first to see the walkthrough of the working environment — the same environment the lessons are recorded in — before it goes public. And you will hear first when enrolment opens. One email, and no list you have to escape from.
If you would rather talk it through first, say so in the role field and Madelein will reply personally.