An autopoietic agentic system makes a living, aware company possible
Every day, a company makes decisions, solves problems, corrects methods and generates knowledge. Some of what it learns becomes documented procedure. Other knowledge remains in exchanges among people, tools, the reasons for decisions and the experience generated by work itself.
MAIOS — Meta AI Operative System opens another possibility: an agentic system that participates in work with an understanding of the reasons, keeps relevant knowledge connected and develops the competences through which it will continue to operate.
Its direction is autopoietic: the results of work can help transform how the system understands, learns and works the next time. A company kernel is not fully formed in the abstract. It meets the people, responsibilities, sources, processes and tools of a particular organization.
Explore the sources: MAIOS Project Kernel · SSK Working Paper 0.12.
1. AI assists operators and participates in work with an understanding of the reasons
An assistant can produce an answer. A system that participates in work must also recognize what situation it is dealing with, which sources give that situation meaning, which previous decisions are still relevant and what changes as a consequence of the work.
For a designer, the reason for a technical decision may be decisive. For a production lead, it may be a procedure and why it was last revised. For someone working with a customer, the commitments, actual product configuration and the responsibilities of the people involved may matter most.
Everyone need not use the same assistant or have unrestricted access to the same information. A kernel can make different competences pertinent to different roles and environments, according to the actual means and permissions available. Participation means contributing to understanding and, where possible and authorized, to action.
The relationship also works in reverse: a reasoned correction from an operator can teach the system; a new relationship made visible by the system can open a possibility for the person.
Explore: MAIOS Project Kernel · How it works.
Competences meet different roles
Select an illustrative role: the relevant sources change, not the Kernel identity.
Conceptual example: no company data is read or modified.
2. System awareness grows through competences that evolve in work
In MAIOS, a competence is not merely a command or a set of instructions. It is a capacity that can carry domain knowledge, sources, criteria, method, conditions of use and learning.
During work, a reusable difference may emerge: a missing distinction, a correction, a better verification method or a successful approach whose reasons become understandable. If that difference returns to the competence that should behave differently later, work has produced not only an outcome but a potentially improved capacity.
Synthetic situated awareness is broader than the sum of competences. It forms when perceived context, sources, pertinent competences, the system's own situation and consequences become mutually meaningful for the work underway. This is not a claim of human consciousness. It is an operational capacity to understand enough of the present to act and continue coherently.
Consider a technician who identifies a new relationship during maintenance. That knowledge can change a diagnostic competence. At a later task—even a different one—the system may reach a method that was not previously available or applicable. Learning matters when it changes work, not merely when more documents are stored.
Explore: SSK: competences and the generative core · Project Kernel.
3. Rules govern competences and how they participate
A company may hold substantial expertise yet lose coherence when it needs to bring that expertise together. MAIOS introduces a Kernel that makes it possible to determine which competence should participate, why it is relevant and with what situated weight.
These rules are not only prohibitions. They also generate coherent possibilities. KA — Kernel Assiomatico keeps the possibility field open rather than allowing the first answer or current tool to become the system's limit. FDLA corrects substitutions of source, object, meaning or intent as work takes shape. The logical router connects the points and guides their engagement in the actual context.
Governing competence participation is distinct from authorizing material effects. Understanding a change does not automatically authorize editing a factory system, record or account. Each action depends on its actual means, responsibilities and authority.
As new information emerges, the relevant competence composition can change. An issue first understood as production-related may reveal a design origin. Competences keep their own meaning while their participation changes with the understanding of the issue.
Explore: KA — Axiomatic Kernel · MAIOS operating Kernel.
4. Meta-competences provide the logic for self-improvement
Meta-skills act on the competence system itself. They can recognize a relevant capacity, compose existing methods, deepen a competence or form a new one when work, knowledge or an emerging possibility makes it useful.
The decisive difference is that even the way competences are formed and improved can evolve. If a system realizes that a competence lost a relevant reason, it can improve the method that preserves reasons in future work. If several competences need to collaborate in a new situation, their composition may become a useful capacity in its own right.
Self-improvement is not permission for uncontrolled change. KA and FDLA sustain semantic coherence; domain competences retain the knowledge they learn; permissions, tools, verification and responsibilities govern real-world effects. Improving a competence and directly changing a company's operational system are separate resultants.
Explore: What is a meta-skill? · Project Kernel Competence Formation.
The movement connecting all four principles
Work brings together operators, sources and competences. A decision, result, correction or possibility forms. Its consequences can return to the competences that should behave differently, changing how later work becomes understandable.
Work → competent participation → result → consequences → reusable learning → later work.
This is where memory and continuity differ: storing information makes it available; continuity emerges when what happened changes how a later situation is understood and acted upon.
The interactive diagram illustrates this relation through selectable states. It is not telemetry from a live company.
From experience to a new capacity
Select a phase to see how a consequence can change later work.
A person and AI engage with a real work event.
From possibility to company-specific incarnation
MAIOS Project Kernel 5.1 is a publicly available distribution for studying and incarnating the semantic-operational core in different domains and projects. The SSK Paper documents the research and its conceptual distinctions. MAIOS has exercised several competence-formation and continuity methods internally.
A real company is a new field, not a copy of our system. Its incarnation requires authorized sources, people, responsibilities, available systems, appropriate integrations and observable consequences. Organizational transformation becomes an exercised capability only through that work.
A useful first encounter may be to find one area where reasons behind decisions, experience and people's knowledge could become reusable capacity. The right scope emerges from the company's own context.
Explore: MAIOS · Project Kernel · AI setup.
Sources and scope
Descriptions of the Kernel and meta-competences derive from MAIOS documentation and the SSK Paper. Company examples are illustrative. This article does not claim actual third-party deployments or measured company outcomes. Awareness here means synthetic situated awareness, not proof of subjective conscious experience.