Background
Where InFinIT Control comes from
InFinIT Control is a trade name of InFinIT Partners B.V. and is under construction. Below: who is behind it, where the idea comes from, and the principles we build on.
Iwein Borm
Iwein Borm taught artificial intelligence at Delft University of Technology at the age of seventeen, and graduated cum laude twenty years ago with a thesis on multi-agent systems: software in which independent parts work together. He then spent more than ten years responsible for the systematic trading strategy of a large hedge fund: billions in daily orders across hundreds of markets worldwide, in a hyper-automated environment where everything had to add up to the cent and comply with every rule. Applied common sense ran through everything there.
In 2019 he founded InFinIT Partners: first a software company with its own development team of thirty-five people, later a venture builder and boutique advisory firm. As an adviser he has worked for governments and large organisations.
He knows governing from the inside. He served four years as an elected councillor, chaired the finance network of a national political party for nine years, sat on the core group that wrote its new statement of principles, and now chairs its network for economic affairs and innovation, where digitalisation and AI are standing topics.
What those worlds have in common: if you are not in control of what you do, and do not consciously choose which way you want to go, you will not get there. All of that experience now comes together in InFinIT Control.
InFinIT Control is a trade name of InFinIT Partners B.V. The common thread at InFinIT Partners is applied common sense, built on two starting points. The rules determine the outcome. And expect people to behave human. InFinIT Control is that story, turned into software. That is how we practise what we preach.
The name says what it is about: a grip on your organisation. Infinite control over your processes.
In today’s buzzwords: an agent harness and control plane with humans in the loop, a digital twin of the organisation, decision intelligence and rules in everyday language. All in one, and most of it without AI.
Our philosophy
InFinIT Control is designed around eleven principles. They come from our own practice and they determine how the software works.
Applied common sense.
We know what happens when an organisation has to succeed in the real world. What matters then is not whether you have worked out the standard cases properly, but how robust you are in a regime change*When something that worked for a long time stops working because the world has changed: a new technology, a new law, a new insight., and how well you handle exceptions and emergencies. That takes common sense, and people who can take responsibility. That is why an exception belongs exactly where it is handled: as a row in a register*A data table in InFinIT Control: rows with fixed columns and rights per register, row or column. Registers hold your organisation’s data and agreements. or as a rule in the process logic of the process*A process has two parts: the declaration, which says what it may read and write, what it asks a person and which services it calls, and the process logic, the rules it carries out.. So the people who own the process can adjust the exception themselves.
As simple as possible, but not simpler.
Every nuance that is needed gets a place. Complexity that adds nothing does not.
Maximum homogeneity.
The same form everywhere: processes, data and rules are recorded in one way, in one control framework in which everything is laid down and in which people and machines both exist. The same roles, the same agreements, the same form, whether a person or a machine performs the action.
Everything revolves around decision logic, and a person signs it off.
Decision logic is the set of rules that decide what happens in each situation. Most tasks do not need AI; they need such rules. AI can help find them in your existing processes and put them into a framework; then people sign them off, per run mode*The mode a process runs in, for instance development or production. Per run mode a process can have its own declaration: what it may then read, and where it writes.. The processes themselves are, in principle, not AI processes: AI helps write the logic; the system carries out the rules. Where AI does sit inside a process, the same agreed limits apply.
The system advises, people decide.
Including what the system may handle by itself. It blocks only at limits that people have set: roles and rights, laws and regulations, contracts. Everything else it flags, visibly.
The system earns its autonomy.
Only when the consequences of a particular kind of decision are well understood may the system handle it by itself. The decisions that really matter stay with people. Moving that line is always a decision for people, never a side effect.
Rights belong to a role, not to a person or a machine.
Who may see, read and write what? Which decisions does an employee take alone, which does a supervisor review, and which are put to the board? Those are agreements per role. An agent*A process that carries out tasks on its own, with or without AI, under the same roles and rights as an employee. works under the same agreements as a person and is identifiable in everything it does.
One organisational brain that everyone contributes to, all the time.
All digital processes come together in one organisational brain*All digital processes together in one whole that everyone contributes to, all the time; what one records, another can use. There is deliberately only one: two chefs in the kitchen doesn’t work.. People and agents work in the same smart harness, and what one records, another can use. So the brain grows with every decision, every exception and every insight that someone adds.
The safeguard is ‘you can see it’.
Not ‘it is not allowed’, but governance*The agreements on who may see, read and write what, and who decides what. The system keeps to them. you can see: every change is recorded in a register, the previous text is kept*Three things under every change: the justification register (why, and whether a person or a machine wrote it), the trail (who did what, when, with the previous text kept verbatim) and the oversight layer that watches and can roll back in time., every justification says whether a person or a machine wrote it, and there is time to roll back. What is visible, a person can judge.
A lexicon with clear definitions.
The lexicon records what a term means and what it does not, and makes the distinction visible where the same term means something different in different places. Your organisation can rename anything without changing how anything works.
The same logic in simulation and in production.
Whatever is automated*Mostly without AI, sometimes with a little AI, sometimes with AI throughout. must remain testable. In a digital twin, a simulation environment, you replay the past: had we always worked this way, what would the outcome have been? Only when it adds up does the same logic go to production, as far as possible unchanged.
That is how we build InFinIT Control: people in control of their organisation’s processes.