Embassy builds living relational models of organizations so leaders can simulate consequential decisions through the human system before making them in the real world.
The decisions it is built for: a succession. An acquisition about to be announced. A founder handing the company to a chief executive. A $20 billion fundraise. A joint venture entered on expectations the two sides have not yet compared.
A cognition lab for relational intelligence. Founded by John Wolf.
I
What the nodes taught us
Embassy’s product architecture was discovered through real human systems.
A node is one person modeled inside one consequential decision. Each one revealed something the system needed to represent, remember, simulate, correct, or name. Seven of those discoveries follow, with the vocabulary they produced. The cases are described without names until each participant has approved named use.
01
How someone answers is part of the answer.
The useful signal is not limited to the content of a person’s answers. It also lives in what they omit, where they contradict themselves, which subjects they animate around, and which identities appear to be negotiating inside the same sentence. In one engagement, a short body of voice notes and a one-hour conversation produced weeks of useful inquiry, and the model surfaced a tension between the person’s devotion to a family system and their own authorship, a distinction the advisory work on its own had not produced.
BecameNode Intelligence · Signal Fidelity · Truth Leakage · Identity Compression
The signal is not only what a person says. It is the pattern formed by what they emphasize, contradict, protect, omit, and cannot yet name.
02
The org chart is not the human system.
A multi-state retail operator appeared to have a financing decision: sell, recapitalize, raise minority capital, or remain independent. Mapping the actual human system revealed a different problem. The operator still carried growth, real estate, vendor relationships, legal escalations, personnel issues, and routine approvals; the semi-retired founder retained legacy authority; and one longtime employee held knowledge that existed nowhere else. The financial model could describe a transaction. It could not answer whether the transaction would create freedom or simply make the cage nicer.
A company needs a representation of how authority, trust, information, identity, incentives, dependencies, and timing actually move, not merely how roles appear on paper.
BecameLiving Relational Model
The org chart shows where authority is supposed to live. The living relational model shows where it actually moves.
03
Identity can become decision infrastructure.
A node profile can do more than describe someone. A biography that reads as a run of unrelated roles usually carries one pattern beneath them, and once the pattern is named the biography becomes strategically coherent and starts doing work in live decisions. Evaluating a technical partner stops being a résumé comparison. The questions become whether this person wants to share the company or build inside it, whether the relationship carries shared identity and consequential authority, whether the two can compress intelligence at a compatible speed, and who makes the founder more confident when the product is failing and the investor is skeptical. The distinction between a cofounder and a vice president of engineering becomes relational architecture rather than title negotiation.
BecameRecognition Event · Identity Compression · Velocity Mismatch · Role Architecture · Decision Architecture
Identity compression turns a scattered biography into a usable operating thesis: the pattern a person carries from room to room, regardless of title.
04
The room can be simulated.
Generic role-play says: pretend you are a skeptical chief financial officer. Embassy can represent the actual one. A capital-preservation bias. A belief that expansion is premature. A recent loss of influence with the founder. A reluctance to contradict the chief executive publicly. Strong evidence for one claim, a moderate-confidence inference for another, and material unknowns still unresolved. Once the nodes are grounded, the system can ask what changes when the founder communicates the decision, when the incoming chief executive communicates it, when the board is privately aligned first, or when a different sequence is used. It can surface who may resist, who may misread the signal, where trust may break, what information is missing, and what still requires human judgment.
BecameRelational Simulation
Relational simulation does not claim to predict people perfectly. It makes the assumptions, counterreads, dependencies, and failure modes surrounding a consequential move visible before reality makes them expensive.
05
Correction compounds.
A compelling read can still overcompress the human being. In the retail case, an early interpretation risked flattening legitimate gratitude, mentorship, and mutual contribution into an overly complete story about inhibited authorship. The revised reading restored those realities while retaining the unresolved question about the person’s current structure, and it distinguished a mixed desire for liquidity, security, continued involvement, and relief from an executed decision to exit. So the system retains the original inference, the counterread, the human correction, the approved expression, the reception, and the outcome. The difference between the first interpretation, the human correction, and what reality later reveals is the judgment delta. Embassy does not merely learn about people. It learns where its model of people was wrong.
BecameCounterread · Human Adjudication · Judgment Lineage · Judgment Delta
Every correction improves the model. Embassy preserves where an interpretation changed, who changed it, and what reality subsequently said back.
06
The brief can carry the room.
The output does not have to remain inside the original encounter. An Embassy brief can be reread, corrected, annotated, selectively shared, compared with later events, and carried into another trusted relational container. The brief is not a transcript of the room. It is the room made portable. That is what makes the Strategic Coherence Brief more than a polished report: a boundary object, a carefully authored representation that different people can use to examine the same system without pretending they already agree.
BecameStrategic Coherence Brief · Boundary Object · Relational Receipt · Portable Power Map · Context Rights · Relational Value Interoperability
A Strategic Coherence Brief makes the room portable: inspectable enough to challenge, precise enough to act on, and alive enough to change when the human system changes.
07
Language changes what the system can see.
Each case kept producing phenomena the inherited language could not hold. Once a distinction had a name, it could be noticed, discussed, tested across cases, corrected, and used in a decision. The lexicon is not brand decoration. It is part of the representational technology. A case exposes something the model cannot yet see; a distinction is created; the distinction enters the ontology; the next case tests whether it survives. The case teaches the system how to see.
Naming changes what a system can perceive. What becomes perceptible can become discussable, testable, and governable.
The sequence the discoveries add up to.
Recognition→Representation→Simulation→Human adjudication→Authorized expression→Reception→Outcome→Learning
The core vocabulary.
- Node Intelligence
- An evidence-grounded representation of one person as they matter to a specific decision: what they know, want, fear, control, cannot say, and may be misread about.
- Identity Compression
- A concise articulation of the pattern through which a person repeatedly creates value across different roles, industries, and contexts.
- Living Relational Model
- A revisable map of the decision-relevant human system: people, relationships, trust, formal and informal authority, incentives, information, identity stakes, commitments, and timing.
- Relational Simulation
- The practice of pressure-testing a consequential move through the actual human system expected to receive and carry it.
- Strategic Coherence Brief
- The decision artifact produced from the living relational model and relational simulation: what is happening, what remains uncertain, how the options may propagate, and what must become visible before action.
- Counterread
- A plausible alternative interpretation retained beside the primary reading so that coherence does not become false certainty.
- Human Adjudication
- The moment a responsible human accepts, rejects, narrows, or corrects a machine-supported interpretation.
- Judgment Lineage
- The inspectable chain connecting source evidence, interpretation, counterread, human correction, authorized expression, reception, outcome, and model update.
- Judgment Delta
- The difference between what the system first inferred, what a human corrected, and what reality ultimately revealed.
Working language, born in the cases.
- Signal FidelityHow faithfully consequential human reality reaches the model.
- Truth LeakageWhat escapes indirectly when direct expression is blocked.
- Recognition EventThe moment a person or system recognizes itself in a representation.
- Velocity MismatchIncompatible speeds of perception, decision, translation, or execution.
- Designed PermeabilityPreserving valuable direct access without dissolving role clarity or accountability.
- Context RightsThe information someone needs in order to carry responsibility coherently.
- Relational Value InteroperabilityWhether trust, judgment, influence, and context can travel across roles without being blocked, duplicated, or stripped of attribution.
- Executive Timing IntelligenceKnowing when a signal can move, who must receive it first, and what readiness is required.
- Belonging VigilanceAttention organized around the possibility of exclusion.
- Embodied Authorization LagWhen a role or permission has changed formally but the person has not yet learned that it is safe to inhabit it.
- Warmth–Legibility GapWhen someone is relationally valued but the value they create remains difficult for the system to name or reward.
- Cultural UnderwritingReading whether a person, institution, or field can carry the future it claims to want.
- Conviction at Operating SpeedThe capacity to develop and transmit conviction quickly enough to support real execution.
Several of these remain working, case-generated language rather than settled product categories. They show the ontology growing in contact with actual human systems.
The model proposes. The human adjudicates. Reality teaches.
Embassy is not a machine for predicting people. It makes decision-relevant assumptions about a human system visible, challengeable, and correctable before they become consequences.
II
The engagement
What an engagement is.
Financial systems capture the transaction. The CRM captures the contact. Nothing captures the judgment, so every organization rebuilds it from scratch, every time. The reasoning behind a consequential decision lives in a few people’s memory and leaves when they do. Embassy’s work begins there.
Today, Embassy is advisory work, led by its founder and scoped to a single decision.
The work has three parts. First, Embassy builds a model of the people inside the decision: not a list of names and titles, but what is true about the relational system around them. Who holds formal authority and who holds the informal kind. What each person has declared they want and what they appear to want. Where trust is sufficient for the proposed sequence and where it is not. What information has not propagated. Which commitments, ruptures and identity stakes are in play.
Second, Embassy asks the question conventional strategy leaves out. Not only what should we do, but what happens to this human system if we do it. What happens if we announce the acquisition on Monday. What happens if the founder communicates it, and what happens if the incoming chief executive does instead. What happens if the board is privately aligned first. Who resists. Who misreads the signal. Where trust breaks. Which sequence gives the decision the best chance of landing.
Third, Embassy works those readings through with the people whose decision it is. Every reading is a claim about people, and claims about people can be wrong, so each one is put in front of the person best placed to correct it. Human judgment determines the move.
What you leave with: the Strategic Coherence Brief.
- What is actually happening
- What each critical person appears to believe
- Where the contradictions are
- What information is not propagating
- The plausible options
- How each option is likely to move through the human system
- What must become visible before action
- Who needs to speak to whom, and in what order
- What should be said, and what should not yet be said
- Where human judgment overrides machine inference
Strategy asks what we should do. Coherence asks what must become mutually visible so that the right action becomes possible.
III
The room
What happens in the room.
The work happens with the people whose decision it is, not about them.
At some point in the conversation, what a person has known for a long time but never said arrives as a sentence. It is usually short. Everyone in the room recognizes it.
That is the part of the work no model performs. Embassy’s role is to have prepared the ground for it, the questions, the record, the reading of the room, and then to hold what was said so it is not lost on the way out of the door.
The tacit becoming explicit.
IV
The software
The software Embassy is building.
The advisory work produces a method. The method is AI-native and becoming software: a workbench in which the living relational model persists between decisions, the simulation can be run again as the situation moves, and the record of every decision stays whole.
What a node holds.
BeliefsDeclared objectivesLatent objectivesIncentivesFormal authorityInformal authorityTrustCommitmentsIdentity stakesInformation asymmetriesDependenciesTensions and historical rupturesTimingDecision rightsUnspoken knowledge
What the record keeps: judgment lineage.
- Source evidence
- What was known, by whom, when the question was live.
- Model hypothesis
- What the model first read in the room.
- Alternative reading
- The counterread put against it.
- Human judgment
- What a person accepted, rejected or narrowed.
- Authorized expression
- What was said, by whom, to whom, in what order.
- Reception
- How it landed.
- Outcome
- What happened, and what the read failed to anticipate.
The distance between the first reading, the human correction and what actually happened is recorded as a judgment delta. Over enough decisions, the deltas show which signals matter, which interpretations routinely fail, which interventions work in which contexts, and where human judgment reliably improves the machine’s. An organization does not merely make better decisions. It develops a better model of how it makes them.
Three rules hold throughout. Evidence stays distinguishable from interpretation. Interpretation stays open to challenge. Human correction stays visible.
V
The company
How the practice becomes a company.
Today
Embassy Advisory
Founder-led engagements inside live, consequential decisions.
Retainers · Fixed scope · Selective participation
Next
Embassy Platform
Persistent relational intelligence deployed across an organization.
Enterprise subscriptions · Licenses · Implementation
Long term
Embassy Intelligence Network
Seasoned advisors licensed onto Embassy, running engagements on the platform; intelligence products built on patterns across decisions and observed outcomes.
Advisor licenses · Engagement fees · Intelligence products and APIs
Advisory is the entry point. The platform is the recurring business. The network is the compounding advantage. Each stage is funded by the one before it, and the third exists only if the first two earn it.
In the September–October 2026 Harvard Business Review, Felipe Csaszar argues that strategy was built around the limits of human attention, and that AI removes three of them: a company can now search thousands of options instead of a handful, replace static frameworks with living models, and pressure-test a decision through structured challenge. Embassy’s extension is that a fourth limit remains. A company can hold an exquisite model of its market while the people who have to act on it stay poorly represented. The chief executive knows something the board does not. The finance chief sees something she will not say in the room. The founder wants autonomy for his successor and keeps revoking it. Call it bounded relationality.
AI can expand collective human cognition only to the extent that the human system becomes legible. Embassy works upstream of the model, on the fidelity of the human signal, before asking AI to reason across it.
Palantir holds an organization’s operational ontology. Glean holds its enterprise memory. Hebbia holds source-traceable reasoning. Read AI holds what can be observed in a meeting. Embassy holds how judgment formed, where a person corrected it, how the decision was received, and what happened: the layer between evidence and consequence that none of them occupy.
Bloomberg organized financial intelligence. GitHub organized software intelligence. Palantir organized operational intelligence. Embassy is built to organize relational intelligence.
VI
Founder
John Wolf.
Twenty years operating where billion-dollar outcomes depended on human judgment, across private capital, family offices and the fine art market. Built and applied relational intelligence across capital formation, leadership, transactions and organizational change. Embassy is the software expression of that work.
VII
Write
Two doors.
John WolfFounder, Embassyjohn@dancer.biz