aiio
DEEN

Vision

Organizations that understand themselves.

The real news of the AI era is not that machines understand organizations. It is that an organization is, for the first time, in a position to understand itself — continuously, rather than as of the cut-off date of a project. The term for it is older than any language model. What is new is that it can now be delivered on.

Read the theoryThe tools

Knowledge is not understanding.

Data becomes information, information becomes knowledge. Understanding is the step that filing does not produce — it has to be established continuously. That is why an organization can be fully documented and still not know itself: the manual was correct on the day it was signed off, and the organization carried on.

The derivation in the long read

What we hold to be true.

Every one of these statements is open to challenge, and every one has consequences for what gets built — and for what is deliberately not built.

An organization is more than the people currently working in it.Individuals understand parts of it. They leave, move department, retire — and what was written down nowhere leaves with them. The organization itself has no organ with which to observe itself: it has documents, systems and experience, but no continuous overview.
Understanding comes before intelligence.What cannot be read cannot be assessed, and what cannot be assessed leads to no decision. The order cannot be reversed — not even by a larger model.
It is not the machine that understands the organization.It reads what is there and puts it together. What comes out of that is a proposal with a source reference. It becomes insight only once people inside the company examine it, correct it and use it. This boundary is not caution, it is the claim itself.
What an organization can do counts for more than what it knows.Companies no longer compete over information — by now everyone has it. They compete over the ability to turn what is present into repeatable action. Knowledge that becomes no capability is storage.
The theory is not the point.It stands here because it explains why the tools are built the way they are — and why some things are deliberately absent. Anyone who only needs the tools does not need this page.

Where the term comes from.

Organizational Intelligence is not a word from the AI era. It has been in organizational research since the sixties, and the diagnosis behind it has been reformulated several times since. What was missing was never the idea. What was missing was the means to deliver on it.

  1. 1967Wilensky

    The sociologist Harold L. Wilensky publishes a book with Basic Books under exactly this title: “Organizational Intelligence”. His subject is not technology but structure — why governments and corporations fail on information they already hold in-house. The diagnosis is close to sixty years old and has not become wrong.

  2. 1978Argyris and Schön

    Chris Argyris and Donald Schön describe organizations as learning systems and separate what a company says about itself from what it actually acts on. That names the uncomfortable half of the problem: an organization follows rules it does not itself know.

  3. 1991Organizational memory

    James Walsh and Gerardo Ungson set out in the Academy of Management Review where the things an organization “remembers” actually sit: in people, structures, workflows, systems and in its own history. Distributed, not in one place — and therefore never fully readable by any single system.

  4. 1995Knowledge management

    Ikujiro Nonaka and Hirotaka Takeuchi distinguish tacit from explicit knowledge and turn it into a management discipline. What gets built from that over the following two decades is mostly storage: knowledge becomes accessible, understanding does not follow from it.

  5. 2007In our own house

    With Quam, the same work begins in Magdeburg from the practical side: documenting processes so that they become auditable. Two decades of projects in industry and the public sector are the reason this theory was not written at a desk.

  6. TodayWhat is new

    Not the term, but the technology. Language models, vector search and knowledge graphs read holdings that previously only people could read — contracts, minutes, standards, tickets, line-of-business applications grown over years. That turns a description of the problem into a capability for the first time.

The terms.

These are the names as they stand in the long read, each with what it amounts to in plain terms. The order is not a ranking: every stage presupposes the one before it.

Organizational Memory
The memory of the organization.Knowledge, connections and decisions are preserved in a form someone can pick up again later. A filing system alone does not achieve that — it holds files, not connections.
Organizational Self-Understanding
The organization reads itself.Out of that memory, a picture of its own workflows, structures and decisions emerges continuously. Continuously is the decisive part: a picture of last quarter is no basis for a decision taken today.
Organizational Intelligence
The picture turns into decisions.Understanding becomes interpretation, interpretation becomes decisions that can be justified. Only at this point does it become visible whether the picture holds — in the quality of the decisions built on it.
Organizational Capabilities
Decisions turn into repeatable ability.A capability is what a company can still do on the twentieth occasion without the same people standing beside it. Only what reaches this form survives a change of staff.
Organizational Self-Empowerment
The organization develops its own capabilities further.The newest term in the series and the least settled. It replaces the earlier formulation “self-improving organization”, because that described a result and this one describes a capability. Provisional, and carried as such in the long read.
Organizational Resilience
The company withstands change.Not a building block of its own, but the consequence of the earlier ones: an organization that knows itself and keeps developing its capabilities comes through migrations, regulation and outages better. This is where the theory meets commercial questions.

The terms are derived chapter by chapter: to the theory →

Claims and limits.

A theory that explains everything explains nothing. Beside every statement, therefore, stands what expressly does not follow from it.

An organization can understand itself.Not on its own. It takes tools that read the holdings, and people inside the company who examine the result. Without that second half, no understanding emerges — only one more archive.
Machines can read the holdings.They do not understand them. What a model outputs is a proposal with a source reference — not insight into a company it does not know. This distinction has stood explicitly in the text since version 0.15.
The picture emerges continuously.It never becomes complete. Anyone promising completeness has smoothed over the gaps — and the gaps are almost always the places where an undertaking later gets stuck.
The terms are not ours.The composition is. Wilensky, Argyris, Nonaka, Walsh — the prior work is citable, and it is cited. What has been assembled from it is our claim, and open to challenge.
The theory is public.It is not finished. Version, date and reason for the change are stated alongside, and chapters get rewritten when a term turns out to be imprecise.
It explains why the tools are built the way they are.It does not sell them. No sentence on this page is a product promise; what the tools actually do stands on the product sites.

What was changed most recently.

The text is not closed. Every version is recorded with its date and reason, and the changes of recent weeks concerned terms, not wording.

  1. 3 July 2026Version 0.10

    Self-understanding is carried as a capability rather than a state: as something an organization establishes continuously, and therefore as a management task rather than the outcome of a project. Until then, knowledge had been at the centre.

  2. 8 July 2026Version 0.14

    Theory and architecture are separated. Since then the statement about organizations stands independently of the question of which systems deliver on it — the previous version could be read as a product description.

  3. 10 July 2026Version 0.15

    One sentence goes explicitly into the text: artificial intelligence can contribute to Organizational Intelligence, it is not that intelligence itself. Before, it stood between the lines. It is the reason no model takes over the role of understanding in this theory.

  4. 10 July 2026Version 1.0 RC

    First complete version: observation separated from management history, decision quality carried as an observable consequence, chapters taken through to the reference architecture. From here the text can be examined as a whole rather than chapter by chapter.

  5. In progressNew edition

    The coming version places the prior work the terms rest on, and answers the two objections that come up most often: that an old term is being sold again here — and that texts about AI and organizations mostly sound as though a machine wrote them. Both belong in the text, not in a footnote.

Further reading.

The full theory is a text of roughly forty minutes. Anyone who only wanted to know what the tools rest on is finished at this point.

  • Theory of Organizational IntelligenceCanonical in English · version 1.0 RC · as of 10 July 2026 · roughly 40 minutes

    The long version: the observation, the derivation of every term, the demarcation against process and knowledge management, through to the reference architecture. Written by Christian Graup, who at aiio is accountable for what gets built.

    Read the theoryPublic
  • Revision journalKept since version 0.10, newest version first

    Every revision with its date, its core thesis and the chapters affected. Anyone wanting to check whether the subject is being worked on, or whether something was written once and then left alone, reads here — the short version stands above.

    View the journalPublic
  • Organizational Intelligence — the living documentEN/DE · Dr. Christian Graup and Jobst von Heintze · continuously extended

    Why organizations need an Organizational Intelligence System: the problem of the missing layer, the OI ladder as a six-step model, plus the measurement catalogue, case studies, maturity model and reference architecture.

    Open the documentPublic

Who stands behind the theory, and for how long: to the company →

Let us talk.

For objections, a conversation is the better place than a form: twenty minutes about the thesis, not about the price list. Anyone who thinks the order is wrong is the more interesting person to talk to.

Get in touchThe tools
processcollector.com →process-magnet.com →processforge.com →
Vision — organizations that understand themselves