Reference to Series 1: WP 1 · Organizational Sentience (the six properties with their failure patterns), WP 2 · Perspective Before Data (the codebook, Husserl's half-full glass) and WP 4 · Leadership Climate as the Regulation Layer (target bands instead of maxima). The evidence, with references, is there. This episode turns it into a checklist and a definition.
ABSTRACT
This episode works on what lies before all data: the words in which things are reported. Organizational sentience is made testable through six properties, and the weakest place in almost every operation is language — the same word means different things to two speakers, and nobody notices because both agree. Added to that is the distinction between quantities you maximize and quantities that belong in a band. Gate 7 opens when five terms are defined in writing and three quantities have a band with an assigned indicator.
Suggested citation: Huesch, G. (2026). The Sentient Organization. Series 2 · The Superintelligent Leader, Episode 7. Working Paper Series 2026, Global Leadership Consultancy.
THE GATE
| The lock | The same word means different things to two speakers — and both agree |
| The key | A codebook with five defined terms, three target bands with an indicator |
| It opens when | the five terms stand in writing where they are used, and three quantities have a band with an assigned indicator |
PART 1 · THE FINDING
WHERE IT COMES FROM
Three papers from Series 1 carry this episode, and they interlock.
WP 1 · Organizational Sentience develops the six properties that make organizational sentience testable, each with a failure pattern.
WP 2 · Perspective Before Data starts one level earlier: before every question of data lies a question of perspective, and that decides what the data means at all.
WP 4 · Leadership Climate as the Regulation Layer supplies the regulation logic: target bands instead of maxima.
All three are evidenced there with references. This episode quotes two sentences verbatim and turns the rest into work.
BODY TEMPERATURE IS NOT MAXIMIZED
Body temperature is not maximized. It is regulated into a band, and whoever maximizes it dies of it — not of an error, but of the success of the maximization.
Revenue is maximized. More is better, and the limit lies outside, not in the quantity itself.
The difference between these two kinds of quantity is half the subject of this episode, and the expensive part of it is: most leadership quantities are temperature and get treated like revenue. Safety, satisfaction, speed, utilization, harmony — each has an optimum in the middle and a price at the maximum, and in almost every organization an arrow points up.
WP 4 shows this on regulated needs. On the AGI-26 poster by Mikeda and Goertzel the sentence stands:
"Needs are regulated toward target bands, like body temperature — balance, never maximization."
Maximum safety is incapacity to act. Maximum satisfaction is incapacity for criticism. Maximum speed is an organization that no longer checks anything. Those are not exaggerations but the end points of the arrows that actually appear on the slides.
THE HALF-FULL GLASS, AND WHY NO MEASUREMENT SETTLES IT
WP 2 starts one level earlier, and there it becomes fundamental.
On Hongju Pae's poster stands Husserl's example: the same half-full glass is half full to one person and half empty to another. No additional measurement settles that. You can determine the fill level to the millilitre and have not touched the question, because the question concerns not the content but the way it is given.
Pae calls this the missing structural condition and formulates his claim thus:
"My claim: scaling is not enough. A structural condition is missing: Perspective — how a world is given, not what is inferred."
For an organization this means something practical: before every question of data lies a question of perspective, and in most organizations it has never been worked on. There is a codebook — an unspoken assignment of words to states of affairs and of states of affairs to valuations — and it decides what counts as simple, urgent, unrealistic or self-evident. Nobody wrote it. Everybody uses it.
And here comes the sentence that connects this episode to AI: an AI system does not learn an organization's truth but its mode of disclosure — and returns it with higher confidence. Whoever feeds an unworked codebook into a model gets his own presets back, now in clean sentences and without hesitation.
NINETY MINUTES ABOUT ONE NUMBER
Twelve per cent market share is on the slide.
The sales director says: "held". The head of strategy says: "eight are missing". Both read the same number, both are right, and neither notices that they are not arguing about the number.
The ensuing discussion revolves for ninety minutes around twelve per cent. Further figures are fetched, comparison periods invoked, competitor data supplied. At the end there is a compromise on a measure that both agree to and both understand differently.
The discussion should have revolved around the codebook, and that takes twenty minutes. The question is not how high the market share is. The question is whether, in our language, this market is one to defend or one to conquer — and that is a definition, not a measurement.
This episode is therefore not language care. It is the cheapest productivity measure available to an operation, because it replaces ninety minutes of leadership circle with a definition in a template.
THE SIX PROPERTIES WITH THEIR FAILURE PATTERNS
This is the checklist from WP 1, in the form in which it can be applied to one's own operation. Each property stands with its failure pattern, because the pattern is what one recognizes — the property itself is invisible as long as it works.
| PROPERTY | FAILURE PATTERN | |
|---|---|---|
| 1 | Contact with reality | The organization learns of external change only through its own figures |
| 2 | Self-perception | The state becomes visible only in conflict |
| 3 | Governed uptake | A claim attains validity because it comes from above, not because it was tested |
| 4 | Regulation | Things are maximized, not regulated into a band |
| 5 | Legibility | Submissions from other areas are accepted, not examined — and that is called trust |
| 6 | Mission continuity | A commitment holds as long as its author is present |
Failure pattern 6 was Gate 1. Failure pattern 5 was Gate 5. Failure patterns 4 and 3 are the work of this episode. Sentience is testable, not atmospheric — that is the yield of this table, and it consists in replacing "our culture is good" with six answerable questions.
WHAT THE LEADER RECOGNIZES HERE
1 · Sentience is testable, not atmospheric. Six properties, each with a failure pattern. That is a checklist and not a feeling, and it can be evidenced on minutes and templates.
2 · Before every question of data lies a question of perspective. The codebook decides what counts as simple, urgent or unrealistic. Whoever does not work on it delegates meaning to whoever happens to speak first — and later to the model that read along.
3 · Target bands, not maxima. What is regulated is regulated into a band. A direction without a band is not a control but a drive with the accelerator jammed, and the price comes at the end of the drive.
PART 2 · THE WORK
THE OBSERVATION TASK
One week, ten sentences.
Note ten sentences that fell in your meetings this week and contain a judgement — "that's critical", "that's approved", "that's unrealistic", "we'll decide that later", "that's strategic". Verbatim, with half a sentence of context.
Words that almost always pay off: decided, approved, critical, strategic, urgent, unrealistic, costly, done.
And then the part that carries the episode: for three of these sentences, ask what the speaker exactly meant. Not reproachfully, not as a test — curiously, afterwards, one to one. "You said that was approved. What does that mean here exactly — that it is signed, or that nobody objected?"
With at least one of the three you will be surprised. In larger houses, with two.
THE IMPLEMENTATION TASK
Monday, one thing. Five terms, one sentence each.
Define five terms in writing that fall daily in your area and were never defined. One sentence each, no paragraphs.
The most important candidate in almost every organization is "decided". Does it mean: somebody agreed? It is in the minutes? It has been communicated? The objection period has lapsed? Four houses, four answers, and in every one of them everybody holds their own to be the obvious one.
Then the part that decides the gate: hang the definition where those terms are used — in the template, not in a glossary. Nobody reads a glossary. A footer in the decision template stating what "decided" means here gets read along at every use.
THE PROMPTS
Prompt 7.1 · The sentience check
``` Below I describe my organization: how it learns what is happening outside; how it registers its own state; how a new claim attains validity; how it responds to disruptions; how submissions are legible across areas; what is preserved at a change of leadership.
Test six properties. For each, name the failure pattern I can actually find again in my description — with a quotation:
1. Contact with reality — failure: the organization learns of external change only through its own figures
2. Self-perception — failure: the state becomes visible only in conflict
3. Governed uptake — failure: a claim attains validity because it comes from above, not because it was tested
4. Regulation — failure: things are maximized, not regulated into a band
5. Legibility — failure: submissions from other areas are accepted, not examined, and that is called trust
6. Mission continuity — failure: a commitment holds as long as its author is present
Format: per property one finding (present / failure / not assessable), the quotation, and one sentence on what is missing. Where my description yields nothing: "not assessable" — do not guess.
My description: [insert here] ```
Prompt 7.2 · The codebook
``` Below are ten sentences that fall regularly in my organization — from meetings, emails, templates. Verbatim, with half a sentence of context.
Work out the implicit codebook: 1. Which term is used here for which state of affairs, without that being defined anywhere? (e.g. what does "approved" mean here, what does "critical" mean, when is something "decided"?) 2. Where do two speakers mean different things by the same word? Evidence it on the pair of sentences. 3. What valuation sits unspoken in the choice of words — what counts here automatically as costly, as unrealistic, as self-evident?
At the end: five terms whose definition would make the biggest difference, each with a proposal for how the definition might read.
No general criticism of language. Only what is evidenced in my ten sentences.
The ten sentences: [insert here] ```
Prompt 7.3 · The target bands
``` Below I name five quantities steered in my area — utilization, speed, safety, satisfaction, conflict level, whatever applies in my case — each with the direction we are currently driving it in.
For each: 1. What happens if this quantity reaches its maximum? Name the price, not the benefit. 2. Which band (from–to) would make sense instead of a maximum, and how does one recognize that the band has been left? 3. Which indicator shows the departure first — and is it assigned to anyone at all?
Format: three points per quantity. Where I have named no indicator, write "no indicator named".
The five quantities: [insert here] ```
On reading 7.2. What the AI returns as an "implicit codebook" is a reading of your ten sentences and not a determination about your house. Its value lies in arising from your material — and in a machine showing no consideration. The most commonly useful hit is point 2: two sentences in which the same word demonstrably means two things. That is the evidence you take into the definition, and it is stronger than any argument about clarity.
PART 3 · ABOUT GERALD HUESCH
Anyone claiming that a shared language lowers the cost of checking must say what that language consists of.
Systems shape behaviour. Through the selection of the right instruments, mechanics and leadership codes — derived from values and leadership guidelines — the behaviour of everyone involved is synchronized and aligned. Not through motivation, but through a system that produces the right behaviour when applied. Sales has Salesforce. Finance has SAP. Leadership has LEAD21.
About the author. Gerald Huesch is the creator of the leadership operating system category – "one of the most professional and creative minds in the domain of leadership and difficult change projects" (Boston Business School). With LEAD21 he built the first operating system for leadership – 21 evidence-based methods, the comprehensive solution and the missing link for complete transformation, extreme growth and high-volume post-merger integration: a unique leadership system for human coordination at scale, in which alignment and culture emerge as a by-product. He has trained more than 30,000 executives in over 30 countries (NASA, Rolls-Royce, Airbus, Omnicom and others) and delivers executive leadership training, leadership development and executive mentoring for executives in Germany, the Americas, Asia Pacific and the UAE. Participants from 300+ companies report +87% clarity, +58% efficiency, +50% performance, +46% alignment and −50% negative stress after LEAD21 (participant-reported); a one-year study with 567 leaders (Prof. Beate Hanson & Prof. Mohit Bhandari, McMaster University) showed improvement in 8 of 9 climate dimensions. Dr. Charles Pellerin (NASA) on LEAD21: "Unique in the world, and a capability that nobody else has." gerald-huesch.com/about-en
PART 4 · THE GATE AND THE PATH
THE CHECKPOINT — GATE 7 OPENS WHEN
Five terms are defined and in writing — and stand where they are used, not in a glossary.
And: three quantities have a band with an assigned indicator instead of a direction. Assigned means: a name, not "controlling".
The test for the first part takes a minute: take the next decision template and look whether the definition is on it. If it is only in your Monday email, the gate is closed.
And if the definition triggers resistance? Then you have found something more valuable than a definition. Whoever objects when "decided" gets defined has a reason — usually a room for manoeuvre hangs on the vagueness that the definition removes. That reason belongs on your list for Gate 8, because it is a candidate for the de facto constitution.
STATUS OF THE GATES
| GATE | TITLE | YOUR KEY | STATUS |
|---|---|---|---|
| 1 | The Sole Decider | Dependency list, 3 decision frames | passed |
| 2 | The Toolbox | Inventory of the last three years | passed |
| 3 | Why Good Methods Evaporate | Fallback path, one mandatory question | passed |
| 4 | The Bottleneck Is Not Information | Checking budget, extension, abstention | passed |
| 5 | What an Operating System Must Do | Assessment of the four functions | passed |
| 6 | LEAD21 — and What Is Still Missing | Gap map, requirements specification | passed |
| ▶ 7 | The Sentient Organization | Codebook, three target bands | you are here |
| 8 | The Superintelligent Leader | Constitution list, succession test | to follow |
WHAT COMES NEXT
Gate 8 · The Superintelligent Leader. The last gate. You have defined terms and set bands — Gate 8 asks what of everything you have done in these eight weeks survives a restructuring. And it answers the question that stands over the whole path: how an outsider recognizes that this organization has stayed capable of judgement.
The lock there: what was never decided governs most strongly.
OPEN QUESTIONS FROM THIS EPISODE
The six properties are taken from WP 1, and their completeness is marked as open there. Six are a proposal, not a closed list.
Whether defining terms improves understanding or merely formalizes it is not tested in this episode. The checkpoint measures that the definition exists and stands in the right place — not that it works.
Target bands require limits, and limits are stipulations. Neither WP 4 nor this episode supplies values. Whoever sets a band sets it from experience, and that belongs stated.
Capturing an organization's codebook completely is not possible. Five terms are a selection by effect, not a lexicon.
REFERENCES
Series 1, Working Paper Series 2026 WP 1 · Organizational Sentience. The six properties and their failure patterns. WP 2 · Perspective Before Data. Perspective as a missing structural condition; the codebook; why a model returns the mode of disclosure. WP 4 · Leadership Climate as the Regulation Layer. Target bands instead of maxima.
Quoted verbatim in this episode Mikeda, A. & Goertzel, B.: A Motivational Architecture for Conversational AGI. AGI-26, LNCS 16855, pp. 62–77, and conference poster. The quoted sentence on target bands appears on the poster. Pae, H.: Perspective as a Missing Structural Condition for Subjectivity in Active Inference Systems. AGI-26, LNCS 16855, pp. 137–143, and conference poster. The quoted claim and Husserl's example of the half-full glass appear on the poster.
All further primary sources used in WP 1, WP 2 and WP 4 from the 19th Conference on Artificial General Intelligence (AGI-26), San Francisco, 27–30 July 2026, are identified there with volume and page numbers.