WHITE PAPER NO. 2
Reading time: 10 minutes · the full development: 25
The Blind Spot
The human factor in decisions of capital: why it stays outside the field of audit — and how it can be read
De-risking the human factor in your decisions of capital.
Contents
01 · THE ARGUMENT
The thesis, in three minutes
The recognition
Look at a closing room. Every part of the file has its binder, its expert and its signed opinion: the accounts are audited, the legal work is reviewed, tax and employment matters are covered, strategy has its diligence, cybersecurity now has its own. One part alone has no binder, no standard and no signature: the team that will execute everything else. The most decisive variable of the transaction is handled by means that would be judged unacceptable for any other — a few dinners, references taken inside a network with an interest in the deal, a general impression.
Instinct is not contemptible. It is simply alone, in the one place where everything else is instrumented.
The reversal
This blind spot is not an oversight, and that is why no additional attentiveness lifts it: it is structural. Three mechanisms hold it in place — a confidence bias, the free-form interview producing conviction rather than prediction; a conflict of position, since whoever carried the deal cannot put a second opinion on the human bet they themselves wanted; and a gap in supply between the practices calibrated for large caps and HR evaluation, competent on the individual and mute on the investment case.
But the first reason is simpler, and it has just fallen away: reading it cost more than the attention budget of a deal allowed.
The reversal proper to your seat
Our white paper no. 1 establishes that analysis has been repriced downwards, definitively. The consequence, for whoever invests, turns twice. Your own analytical apparatus is commoditised along with the rest — memoranda, models, comparables. And the variable that apparatus never learned to read becomes, by contrast, the first territory of your judgment. The blind spot stops being a tolerable defect: it becomes the exact place where the advantage moves.
What follows
This variable can be read — the research establishes it, and that is the first step of the demonstration that follows. It can be read on one condition: having a framework that thinks it as a system, and not as an addition of profiles. What the team knows how to do, who actually holds the functions that carry performance, what circulates between those people, what binds them — all of it referred to the plan they must serve. What that reading returns is not a score: it is a location — the fulcrum, the place, often far from the symptom, where limited action produces maximum effect.
Thirty minutes on a real situation — book a conversation
02 · THE EVIDENCE
1. What the research establishes
The practitioners' intuition — that the leadership factor decides transactions — has been put to the academic test. The reference study is by Steven Kaplan, Mark Klebanov and Morten Sorensen, published in The Journal of Finance in 2012: several hundred in-depth evaluations of candidate chief executives in privately backed companies, cross-referenced with the subsequent performance of the transactions.
Two results are worth holding on to. The first: structured evaluation captures a real signal — characteristics measured before taking office correlate with later success. The human factor is therefore not the ungraspable “qualitative” that gets left to instinct: it can be investigated. The second is the more uncomfortable one: it is the execution traits — consistency, exactingness, the capacity to hold a course and to get things done through others — that discriminate success, more than interpersonal ease and social brilliance.
And the informal interview — the dinner, the presentation to the committee — measures precisely the opposite: it rewards charm, ease and the quality of the story. The gap between what instinct measures and what counts is not a nuance: it is the whole justification for instrumentation.
Measurement also has its hierarchy, long established. Schmidt and Hunter's meta-analysis (1998) of the predictive validity of selection methods places the structured interview at the top — standardised questions, behaviourally anchored rating scales, trained evaluators — and leaves the free-form interview at the bottom, disqualified less by its weak validity than by the false confidence it inspires: it convinces the evaluator without informing the decision.
The negative side has to be stated with the same clarity: typologies that file a person into a type in four letters or four colours do not meet elementary psychometric standards of reliability and predictive validity; their commercial success is no substitute for evidence, and they have no place in an investment decision. Even valid instruments have their limits: their validities are real but probabilistic — they shift the probabilities, they do not predict a destiny. Hence the only serious method, triangulation: structured interviews, a validated personality inventory, factual data — each source controlling the blind spots of the others, convergences making evidence, contradictions making questions.
An assessment that promises certainty lies; an assessment that ranks probabilities and owns its unknowns illuminates. Instruments, though, measure people. What decides is the system those people form.
03 · THE SYSTEM
2. Two companies under one roof
Inside every company two companies cohabit. The official company: the org chart, the processes, the delegations, the packages — the one the documents describe and the audit photographs. And the real company: who talks to whom, who follows whom, who actually decides, who believes in it — the one every executive knows and no binder contains. The health of a team reads in neither one nor the other: it reads in what they do to each other.
They reinforce each other. A decision taken on Monday is applied on Tuesday. Processes help instead of hindering; the best people stay — and one knows why. Clear structure releases the energy of the living organisation; a healthy living organisation legitimises the structure and corrects it. The investor's reading: the team will carry the thesis — support it, do not over-steer.
They ignore each other. The org chart says one thing, reality does another. The processes exist and everyone routes around them; everything goes through unofficial channels; official meetings record decisions taken elsewhere. Neither destruction nor reinforcement: a disconnection — and value leaks out in silence. The reading: reconnect the living organisation and the structure before reorganising — reorganising a disconnection only moves its lines.
They destroy each other. Every reorganisation feeds the distrust; every departure calls for more control — which provokes the next departure. The spiral feeds itself: the very mechanism that made the system strong now sets the speed of its fall. The reading: the risk self-amplifies — act fast, at the fulcrum.
These three situations are the three regimes of a single loop — virtuous, broken, inverted. The lineage is owned: Burns and Stalker distinguished mechanistic and organic regimes as early as 1961. The model's own move holds in one sentence: these two regimes do not work as a slider — more or less structure, more or less living organisation — but as a feedback loop, which reinforces itself in whichever direction it turns. Bureaucracy is not too much structure: it is structure that the living organisation has stopped feeding. The relevant diagnosis is never “where to place the slider” — it is: is the loop turning, and in which direction?
04 · THE APPARATUS
3. Three levels, six dimensions
The loop is the movement; one still needs the apparatus that describes it. The fuseki “human capital assessment” — the matrix in our Goban repertoire that reads a leadership team — articulates three logical levels and six dimensions of examination. The precision matters: these six dimensions are not six boxes on a checklist, and they do not belong to the same plane.
At the centre, Vision — the reference standard. The strategic plan, the investment thesis: its clarity, how far it is shared, how far it is embodied, how it is communicated. It occupies the centre for a logical and not a decorative reason: it is what everything else is assessed against. Nothing is judged in the abstract — the same executive profile is an asset in a conquest phase and a risk in a structuring phase; the same tight-knit team is a strength on paper and a fragility if its cohesion rests on two people the plan will separate. Vision is therefore not scored — one does not measure a standard against itself — it is qualified: formulated? shared? embodied? communicated? A failing Vision does not only distort its own box: it renders every other reading indeterminate.
In the body, four quadrants — the operation, scored. Key Resources, the question of the stock: what human assets does the organisation hold, in alignment with its operating model — and where is the gap? Critical Functions, the question of allocation: which functions actually carry performance, are they identified, equipped, delimited — and held by the best people? Synergies, the question of the system: are the dynamics between people and between teams fluid, cooperative, aligned with the roadmap? Commitment, the question of fuel: what real buy-in is there to the plan — and, in an LBO, the actual alignment of the package beyond its legal mechanics?
In the crown, Governance — the control, scored on its own plane. The demarcation from Critical Functions is one of the most operative in the model: Critical Functions belong to the operational — who runs the company; Governance belongs to the administrative — who governs the company: the bodies, the rules of the game, the relationship between shareholders and management, the mechanisms of control and of transmission. The same fact — an omnipresent founder — then reads twice: as operational concentration (a key-person dependency, which calls for a succession plan and a strengthened second circle) and as an administrative failure (an absence of counterweight, which calls for effective governance). To treat one and leave the other is to have treated nothing.
05 · THE TRANSMISSION
4. The transmission
What remains is the link between the crown and the centre — and it is the model's second loop, of a different nature from the first. Governance does not sit above the system for decoration: it holds the transmission between the operation and the intention. It is the body that measures the gap between what the system produces and what the thesis aimed at — and that corrects: the system, or the thesis itself. Where the loop in the body amplifies in whichever direction it turns, the transmission balances — it is the thermostat of the thesis. It has two states: held — steering is exercised, the gap is read and reduced — or broken: the organisation drifts, operating perhaps, but no longer in the service of its plan.
The full reading of a situation therefore crosses the regime of the engine with the state of the transmission — and it is this combinatorics, not a collection of scores, that qualifies an organisation. Two cases are enough to show what it sees and nothing else does.
Success that drifts: a virtuous engine under a broken transmission — the company performs, the teams strengthen, every score is good; but nobody is measuring the gap to the thesis any more, and the organisation is moving away, efficiently, from what it was bought for. No audit classifies this case as a risk; the combinatorics designates it.
Crisis seen, therefore steerable: an inverted engine under a held transmission — the spiral has started, but governance sees it, names it and arbitrates; the prognosis is entirely different from that of the same spiral under a blind governance. Same photograph, two films — and two opposite decisions.
Thirty minutes on a real situation — book a conversation
06 · THE MOVEMENT
5. The film, not the photograph
An audit photographs the official company — org charts, processes, indicators: a sharp image, accurate and necessary. But the real company cannot be photographed: it is observed in motion. To read a team is to watch the film — both companies, and what they do to each other, over time.
Systemic — the symptom is almost never the cause. In a system, the problem rarely appears where it is born. Commercial underperformance can be born of a blocked governance; turnover, of a package that does not bind; a project that stalls, of the best people allocated to the wrong functions. The model works back from the symptom to the mechanism — and treats the cause, not the noise.
Dynamic — a state is not a trajectory. The same photograph hides two opposite films: the company in the process of repairing itself, and the company in the process of coming apart. The diagnosis gives the direction and the speed of the movement — and it is the movement, not the instantaneous state, that decides the action: support, reconnect, intervene.
This is the integral approach: the official company, the real company, and the dynamic that binds them. Where the audit stops at what can be counted, the Goban reads what is at play.
07 · THE CLOSE
The skilful move
What makes the eleven p.m. questions investigable is not a new method: it is a price that has changed. The human condition of an investment decision — this team, for this plan, in this phase — was never illegible; it was too expensive to read for the attention budget of a deal. It was the tolerable blind spot of a world in which analysis was expensive; it becomes the differentiating variable of a world in which analysis costs nothing.
In the game of go, the skilful move is called a tesuji: the precise placement that accomplishes more than brute force. That is the exact ambition of this practice — neither the heaviness of an established apparatus nor instinct alone: a reading that is structured, honest and actionable, in the service of one thing only — the right decision.
Everything gets audited. Except what decides everything. It is time that stopped being true.
Thirty minutes on a real situation: a transaction under way, a portfolio company that worries you, a succession to prepare. Nothing to prepare, no commitment — book a conversation
The apparatus
What makes this text defensible rather than persuasive: what the blind spot costs, the questions that open it, the mechanisms that hold it, its lineages, the protocol, its occasions and what it refuses.
What the blind spot costsfive situations every investor will recognise
The cost of this blind spot does not show up as a line of fees: it shows up in portfolios. Five situations say it, and every investor will recognise them.
The key person discovered after closing. The org chart named the chief executive; the real company obeyed someone else — a technical director, a discreet number two, sometimes a manager with no title at all. Their resignation, six months after the transaction, reveals that the value rested on a person the file never mentioned.
The conquest leader out of their depth in the structuring phase. The founder who took the company from zero to thirty million is rarely the one who will take it from thirty to a hundred: the qualities that created the value — intuition, speed, omnipresence — become the obstacles of the next phase. The file evaluated a track record; what needed evaluating was a fit with the phase now opening.
The silent fracture in the executive committee. Five brilliant profiles, five reassuring interviews — and two camps that have not cooperated for two years, held together by growth that forgave everything. When growth slows, the fracture starts paying its interest. No CV mentioned it: it exists only between the people, never in them.
The succession that was never prepared. The departing chief executive promised a transition; nobody evaluated the successor — and, above all, nobody evaluated them against anything: against what the company will need in three years, not against the portrait of the current incumbent.
The misaligned LBO. The package was negotiated, signed and made legally secure. But the legal mechanics of a management package say nothing about its actual effect: does it bind the people it is meant to keep, over the life of the plan, at the moments when the plan will hurt? Alignment of interests is a psychological and collective fact before it is a clause.
These situations have one thing in common: not one of them was unreadable. Each left signals — but signals that had no place in any binder, because no category of expertise had a mandate to collect them.
The eleven p.m. questionsthe opening, in full
The status of this text, stated at the outset because it governs the reading: it is a theoretical demonstration. It invokes no engagement result, no client case, no proprietary data; the situations it describes are situations types.
A decision of capital is a moment of maximum load. Tens of millions are committed to a thesis — and the whole execution of that thesis rests on four to six people who have been met three times, in settings where everyone plays their best part. The committee expects a degree of certainty that the human file does not supply; the calendar waits for no one; and the person who carried the deal cannot coldly put a second opinion on it — they bought this team, and can no longer judge it without bias.
Then the file closes, and what remains are the questions that appear in no binder. Will the founder hold through the phase now opening — or only the one now closing? Will the number two, without whom nothing runs, still be there after closing? Do these five brilliant profiles make a team — or five careers? Does the package that has been signed genuinely bind the people it is meant to keep?
These are the eleven p.m. questions. Every investor knows them; almost nothing in the apparatus investigates them.
That load is not a weakness: it is the nature of the decision. But it explains a fact that ought to surprise us more — the most decisive variable of a transaction is also the least instrumented. One does not willingly look at what one has no means of reading. Everything gets audited. Except what decides everything.
That sentence is the very subject of our first white paper, where it holds for all the decisions of an organisation. Here it bears on one alone: the decision that commits capital to a team.
This white paper proceeds in three movements: the observation — why this blind spot exists and why it persists; the foundations — what sixty years of research establishes about organisations, leadership teams and the measurement of the human; the method — how the fuseki “human capital assessment” reads the company as a system, and how that reading comes down as far as the decision. The reader who wants the general thesis first — why the world changes shape when intelligence stops being scarce — will find it in The World That Is Coming; the reader who wants to see the production mechanics, in Intelligence, by design.
Why the blind spot persiststhe three mechanisms, in full
If the cost is so visible and the evidence so long established, why does the blind spot remain? Three mechanisms hold it in place — and it is precisely because they are structural that no additional attentiveness dislodges them.
The confidence bias. Unstructured expert judgment is, on this ground, a mediocre instrument — occupational psychology has established this for decades, and it is hard to accept for professionals whose trade is precisely to judge. The longer the experience, the higher the confidence in instinct; validity does not follow. The free-form interview produces conviction, not prediction.
The conflict of position. Whoever defended the file to their committee cannot put a second opinion on the human bet they themselves wanted. This is not a moral weakness, it is a mechanical commitment bias: one does not coldly judge what one has bought. Every other diligence is entrusted to a third party for exactly that reason — the human factor is the only part of the file where the deal sponsor is also asked to be the expert.
The gap in supply. The top of the market is served: executive evaluation practices exist, calibrated for large caps, at their prices and in their formats. The bottom of the spectrum belongs to HR evaluation — competent on the individual, mute on the investment case. Between the two, small and mid-cap — where most transactions are done — has nobody who speaks their language, at their scale and on their calendar. Demand without supply eventually stops being expressed: the silence of the market maintains the blind spot that created it.
The blind spot is not an oversight — it is structural. And what is structural is not corrected by additional attentiveness: it is treated by structure. By a method.
The lineagesChandler, Burns and Stalker, Lawrence and Lorsch, Hambrick and Mason, Wageman and Hackman, Edmondson, Schmidt and Hunter, Barrick and Mount
The human factor in decisions of capital is not virgin territory: it is among the most studied objects in management science and occupational psychology. Three converging bodies of research underpin everything that follows.
3.1 — Strategy fails inside the organisation
Why do impeccable plans die in execution?
The question is as old as the discipline. Alfred Chandler set out the first term in 1962: structure follows strategy — a new strategy demands a new organisation, and most strategic failures are organisational lags. Contingency theory established the second: Burns and Stalker (The Management of Innovation, 1961), then Lawrence and Lorsch, showed that there is no good organisation in the abstract — only organisations that are adequate, or not, to their environment and their plan. The congruence models — Nadler and Tushman, the 7S, Galbraith's star — drew the grid from it: performance is born of alignment between the components of the organisation, and is lost in their misalignments.
Contemporary research has sharpened the line. Richard Rumelt: bad strategy is recognisable by its refusal to come down into the organisation — it proclaims objectives without building the path. Roger Martin and A.G. Lafley: the cascade of strategic choices is only real once deployed down to the capabilities and systems that carry it. David Teece, then Rita McGrath: in unstable environments, sustainable competitive advantage gives way to the capacity to reconfigure — and that capacity is a property of teams, not of documents.
To evaluate a thesis without coming down as far as the organisation that must carry it is to evaluate a plan, not a company.
3.2 — The leadership team as explanatory variable
Why do five brilliant CVs tell you nothing about the team?
The second body of research moves the gaze from the organisation to those who run it. Upper echelons theory — Hambrick and Mason, 1984 — established that a company is the reflection of its leadership team: strategic choices, risk-taking and performance carry the measurable imprint of the executives, of their trajectories and of how they see the world. The relevant unit of analysis is therefore not the isolated individual: it is the team.
And the team is precisely what the addition of profiles does not show. Ruth Wageman and Richard Hackman, at the end of one of the largest studies ever conducted on executive committees, established a counter-intuitive result: leadership teams are among the least effective teams in their organisations — not for want of talent, but for want of conditions: blurred boundaries, an ambiguous course, meetings with no real object, incentives that reward everything except collective work. Amy Edmondson added the variable that governs learning: psychological safety — the possibility of naming an error or a disagreement without cost — without which a brilliant team turns into a cautious aggregate.
What predicts the performance of a leadership team does not reside in the people but between them — in the system of interactions formed by profiles, functions and interests. It is that system one has to know how to read.
3.3 — Measuring the human: possible, within limits that must be owned
What can be known — and what will never be known?
The third body of research is that of measurement. A century of occupational psychology has established a hierarchy of instruments, synthesised in Schmidt and Hunter's meta-analysis (1998) of the predictive validity of selection methods. At the top: the structured interview — standardised questions, behaviourally anchored rating scales, trained evaluators — whose predictive validity is among the highest known. At the bottom: the free-form interview, disqualified less by its weak validity than by the false confidence it inspires — it convinces the evaluator without informing the decision. On personality, the scientific consensus is the five-factor model: Barrick and Mount established that conscientiousness — organisation, reliability, perseverance — predicts job performance across occupations; a result that echoes precisely the execution traits in the Kaplan, Klebanov and Sorensen study.
The negative side has to be stated with the same clarity: the popular typologies sold on the corporate away-day circuit do not meet elementary psychometric standards of reliability and predictive validity; their commercial success is no substitute for evidence, and they have no place in an investment decision. And even valid instruments have limits: their validities are real but probabilistic — they shift the probabilities, they do not predict a destiny. Which is why the only serious method is triangulation: structured interviews, a validated personality inventory, factual data — each source controlling the blind spots of the others, convergences making evidence, contradictions making questions.
Honesty about the limits of measurement is what separates the rigorous practitioner from the seller of tests — a useful paradox: an assessment that promises certainty lies; an assessment that ranks probabilities and owns its unknowns illuminates.
References. The work cited in this document is that which grounds the hierarchy of validity of assessment methods (Schmidt and Hunter on the predictive validity of selection methods; Barrick and Mount on the five-factor model), the reading of executive teams as systems (Hambrick and Mason on upper echelons; Wageman and Hackman on the enabling conditions of top management teams; Edmondson on psychological safety) and the analysis of organisations as feedback loops (Burns and Stalker). Beyond these, this text cites no secondary source: it sets out a matrix and its use.
The protocol in five movementsframing, gathering, triangulation, delivery, follow-through — and the two disciplines
A model is worth only the move it makes possible. Here is how the reading comes down in practice — in two weeks from the first interview, within an invariant protocol.
5.1 — The protocol in five movements
The framing. The engagement opens on Vision: a reformulation of the investment thesis or the strategic plan, with the people who carry it. This movement is not a formality — the presented subject is rarely the real subject, and without an explicit reference standard no reading is possible: one cannot evaluate a fit without knowing a fit to what.
The gathering. Structured interviews with the key people, on a standardised grid — the most validated predictor in the research — cross-referenced with a five-factor family personality inventory of documented validity, and with documentary analysis. Every signal collected attaches to one of the model's six dimensions: that is the guarantee of exhaustiveness — nothing that counts is left without a box, nothing that is collected is lost.
The triangulation. The sources control one another. Convergences make evidence; contradictions make questions — and they are flagged as such, never smoothed over: a contradiction between what the interviews say and what the instrument shows is information, often the most valuable of the engagement.
The delivery. The deliverable is a committee report, and it reads a system: a matrix of collective functioning, a scored evaluation of the four quadrants and of Governance, a qualification of Vision, ranked recommendations. Individual readings are not poured into it: they go to each person assessed, in an individual debrief, and reach the report only through what they do to the system — by name only where the decision requires it and the person has consented. That is the ethical and practical condition of the exercise: what has been measured is told first to the person who gave it.
The follow-through. The reading closes as a roadmap: decisions to be investigated, with owners and deadlines — not observations. Because the diagnosis does not return scores: it locates the fulcrum — the place, often far from the symptom, where limited action produces maximum effect.
5.2 — The two disciplines
This protocol holds by two disciplines, both of which are part of the product.
The legal discipline. Evaluating people is a legally framed act — and that is a protection for everyone. Explicit purpose, informed consent from every person evaluated, minimisation of the data collected, rights of access, individual debrief, limited retention: this framework is not a constraint endured, it is what makes the conclusions usable in a governance decision. Material gathered fairly is material one can rely on. Material gathered otherwise is neither fair nor reliable.
The technological discipline. Artificial intelligence takes part in the production — and the fuseki is precisely what makes it possible to employ it while conceding nothing to it. An explicit conceptual framework draws the division of labour: the machine investigates what can be investigated — structuring interview verbatims, bringing signals into coherence across the six dimensions of the grid, completeness and contradiction checks, formatting — under constant human supervision; and judgment keeps what belongs to it alone — contextual weighting, recognition of the regime, the recommendation, the signature. The rule is absolute, ethical, legal and methodological at once: the machine never assesses a person, and no decision is automated. Delegating the production without delegating the thinking: that is exactly what an explicit model makes possible — and what no intuition, however well equipped, guarantees.
Five situations where assessment changes the decisionthe first hundred days, signing to closing, underperformance, succession, the build-up
Assessment of a leadership team is not a periodic exercise: it indexes on the moments when the decision is in play. Five occasions concentrate the essentials.
The first hundred days. Closing is signed; the value creation plan is being calibrated now. Assessment brings the map: the real dependencies, the successions to prepare, the quick human wins — and the fulcrum to start from. A hundred days commit three years; better to commit them on a reading than on an impression.
Between signing and closing. The confirmatory phase is the first legitimate window: management becomes accessible, consent becomes possible. Terms, package, integration — everything still open to negotiation gains from being negotiated with knowledge of the team, before it all sets.
The underperforming portfolio company. A gap to plan always has several possible fathers: the market, the model, the management. Assessment isolates the managerial share of the gap — before concluding it was the market, or replacing the wrong executive: the visible one, who is not always the one who weighs.
Succession and key-person risk. Evaluate successors against the company's future need, not against the portrait of the incumbent. A successful succession replaces a function in a phase — almost never a person like for like.
The build-up. The value of a consolidation is destroyed first in its human integration. Reading both teams before merging them — their respective critical functions, their synergies, their cultures of decision — turns a bet on integration into a plan for integration.
What this practice refusesthe five commitments that are not open to negotiation
One has to end where the misunderstandings begin: with what this practice refuses.
The evaluation sector knows its temptations — overselling prediction, producing falsely reassuring scores, turning probabilities into verdicts, dressing up as science typologies that are nothing of the kind. These temptations are economically rational: certainty sells better than nuance. They are also exactly what discredited evaluation in the eyes of executives who, once “profiled” in four letters, never came back to it.
The opposite course is taken out of conviction — and out of enlightened self-interest: an assessment is only defensible before a committee, and before a judge should it come to that, if it is honest about its limits. Hence five commitments that are not open to negotiation. The people evaluated know what is being measured, why, and for whom — and each receives their debrief. Instruments without demonstrated validity are excluded, whatever their popularity. The conclusions distinguish the established, the declared, the qualified and the not established — a judgment that does not know its own degrees of certainty is not a judgment, it is an opinion. No score substitutes itself for the judgment of those who decide. And measurement stays in the service of the decision — never the reverse: the assessment illuminates the decision; it does not take it.
This is not the most commercially comfortable stance. It is the only defensible one — and, over time, the only one that builds what this practice needs in order to exist at all: the trust of those who open their executive committee to an outside eye.
Where to go nextthe two other panels of the triptych
The general thesis of which this document is an application is set out in The World That Is Coming (white paper no. 1). The production mechanics — how an artificial intelligence chain is designed to investigate everything and assess no one — are set out in Intelligence, by design (white paper no. 3).
FRONTIÈRE — The frontier of intelligence, in the service of your leadership.
“Everything gets audited. Except what decides everything.”
De-risking the human factor in your decisions of capital.
Frontière — Frontier intelligence for human capital assessment & leadership advisory. Founded in Marseille by Alice Arnaud, the practice serves investors and company leadership across France.
Tesuji — the intelligence engine · Goban — the repertoire of matrices · Kifu — the decision interface. “The Goban reads, the Tesuji investigates, the Kifu records.”
Alice Arnaud, founder — alice@frontiere.ai · frontiere.ai
© Frontière 2026 — The Goban, its fuseki and the Tesuji engine are the intellectual property of Frontière.
This text is the argument. Its development — the elaborations, the lineages in full, the objections handled one by one — is available to download.
The full development — 25 minutes (PDF)