Artificial intelligence systems do not access organizations directly. They reconstruct them. This reconstruction can be observed, characterized and governed.
An artificial intelligence redefines it in a few seconds.
While organizations have long structured the information they publish, they still have few means of understanding how this information is interpreted and reconstructed by artificial intelligence systems.
A deeptech company. A proprietary infrastructure. A structured human interpretation.
Limits are part of the method.
Saying exactly what is known. And what is not.
An organization knows what it publishes; it has no direct way of observing what a system reconstructs from it.
When a conversational system is queried, it assembles a response from the information environments available to it. What it returns about an organization is not the organization itself: it is a reconstruction, faithful or partial.
An organization may publish precise information, yet observe a partial or ambiguous reconstruction in the systems queried. And when a name is shared, the answer may describe an organization other than yours.
A methodology of observation, not evaluation.
The OMINA methodology defines how an organization can be observed in the information systems used by artificial intelligence: under which conditions, with which limits, and what the results allow, and do not allow, to be concluded.
Three levels of analysis, never conflated.
The real entity.
Its encoding across information environments.
The way systems assemble it when they are queried.
Every methodological statement bears on the last two levels exclusively.
The 2026 · 2027 study
Discover the roadmap and what the study of one hundred companies will establish.
These systems do not draw on a memory alone: they also search at the moment of the question.
What they return combines what they have retained and what they find, and both can be observed.
Observation
Signals detected
The infrastructure queries the systems, records what appears, and what does not, and preserves every piece of evidence.
Interpretation
Coherence and meaning
OMINA reads the evidence and establishes a structural reading: what the reconstruction reveals, within the exact limits of what was observed.
Governance
Reconstructed image
The reading allows the organization to understand how it is represented, what builds that representation, and how this may inform its decisions. These orientations are established by OMINA, never generated by the instrument.
Execution: observation executed by the OMINA analytical instrument, under defined protocol conditions; every execution is preserved under evidence custody before being served. Interpretation and governance established by OMINA.
An observation never produces a simple « present / absent ». It produces distinct states, never conflated.
Under the conditions tested, the content appeared when the systems were queried.
Queried, with no appearance: a finding, not a void. The absence was established under defined conditions; it constitutes a result, never a judgment.
These conditions were not tested. No conclusion is drawn from what was not observed.
The observation is complete; no distinct content follows for this dimension. The corresponding finding is carried by the overall profile: nothing failed, nothing is missing.
Measured absence and absence of measurement are two different pieces of information. Conflating them would be a category error; the methodology forbids it. A qualified absence is never a judgment of the organization: it describes a state of the information environments, under defined and dated conditions.
An observation is a state, not a trend.
No. OMINA does not intervene in AI systems and does not modify their responses.
OMINA observes the way AI systems describe an organization, what is retained, what is absent, what is confused, and identifies what is fragile and where it comes from.
OMINA then establishes orientations: what should be addressed as a priority, not what to publish in order to improve how it appears. Decisions and their implementation remain the organization’s own.
Not only from its website. An AI system assembles its answer from the information environments available to it at the moment of the question: what it has retained, and what it finds.
What the answer returns depends on the coherence of these environments with one another, and an organization has no direct way of seeing what they hold about it. This reconstruction is what OMINA observes and characterizes.
For years, organizations have built their digital presence, presented their activities and their offerings, and explained what sets them apart. But they have no direct view of what AI systems retain from all this when they speak about them.
OMINA provides that reading. Beyond being visible: what does AI actually understand about my organization?
You will be informed of OMINA’s publications and of the progress of the study on one hundred companies. Nothing else.