Allometry Master Thesis.

Canon, not marketing. Frozen at v2.3 until the first reconciled outcome, now October 1, 2026. Fifty sections, nine appendices and the v1.1 to v2.0 migration map. Business Shape is the organizing idea: horizontal at the kernel, specific through Templates, governed through Tools, accountable through the Ledger, read continuously by the Pulse. The law is Y = a·M^b. Supersedes v1.1 wherever the two conflict. Companions: the manifesto · the six-pager · the open letter · the physics of profit · v1.1 archive.

Frozen (Sept 2 2026): v2.3 is frozen until the first reconciled outcome, now October 1, 2026 (founder ruling, Sept 5). Per §50.7, new ideas enter BACKLOG-2026-08-06.md before canon; nothing enters this file before that date. Status: Working canon, fifty sections. Date: September 2026. v2.3 (Sept 2 2026): third logged revision, see §34.8: the Pulse named as the reading layer (§3.5, §13.5, §19.1, §23.1, §32.3), the counted market and diligence checklist (§26.5, §26.6), the ninety-day calendar and objection map (§25.8, §25.9), a fourth why-now (§25.7). v2.2 (Sept 2 2026): second logged revision, see §34.7: preface, named Doors and the distribution machine (§25.5–25.7), business-model math and milestone gates (§26.4, §29.5–29.6), use of capital (§50.9), and four build appendices (F–I). v2.1 (Sept 2 2026): logged revision of v2.0, see §34.6: the law Y = a·M^b and the physics of profit restored to the core (§3.5), Objective made a first-class object and loop step, agent-to-agent commercialization named (§14.5), the Decision Graph given two evidence inlets (§11.4), the Oracle rule (§21.6), and the pricing hypothesis under test (§26.3). Purpose: Product and company doctrine, not website copy. Supersedes: v1.1 (MASTER-THESIS-2026-08.md) wherever the two conflict. Appendix E maps every v1.1 section to its v2.0 destination.

Allometry builds a living model of a company, shows what should change, explores what the business could become, and helps make the change.

The v2.0 reset promotes Business Shape from a feature to the organizing idea of the company. Margin intelligence, Revenue Recovery, commercial commitments, capital, Work Calls and physical capacity remain important, but each now occupies its proper place inside a larger and industry-agnostic system.

Allometry is horizontal at the kernel, specific through Templates, governed through Tools, accountable through the Ledger and improved through evidence.

Why this company exists

Every company is already obeying a law it has never measured. Revenue is the mass. Everything else, cost, capacity, cash, complexity, margin, scales with it at an exponent nobody wrote down. When the exponents are right, growth compounds. When they are wrong, the company becomes an elephant-sized mouse: bigger every year and weaker per gram, and no dashboard says why.

The software a company runs on was built to record what it does, not to ask what it should become. The CRM knows the deal closed. Accounting knows the invoice went out. Nobody knows whether that work bent the exponents toward the business the owner wants or away from it, because nobody was measuring, and nobody was standing at the door when the commitment was made.

Allometry stands at the door. It measures the exponents from executed work, decides against them, and enforces the decision at the moment the company commits: a quote, an order, a hire, a location, an acquisition. Then it reconciles what happened with what was expected and gets better. Every commitment becomes a small, recorded vote on the Shape of the company. Over enough votes, the company can see itself, choose what it should become, and make the change with evidence instead of decks.

That is the whole thesis. The fifty sections that follow are the mechanism.

How to read this document

Allometry is pre-product-market fit. "Canon" means a stable operating assumption that prevents product drift. It does not mean permanent truth. Every assumption remains revisable when customer and outcome evidence contradicts it.

The v2.0 hierarchy

1. Profile: What is true about the company now.
2. Objective: What the owners want the company to become, in measurable terms: the enterprise value, cash, resilience, control or liquidity they are pricing toward.
3. Diagnose: What creates value, what constrains it and what will break next.
4. Explore Shapes: What credible configurations the company could become.
5. Choose a Move: What measurable change management is prepared to make now.
6. Authorize: Whether the Move fits evidence, policy, risk and authority.
7. Execute: What humans, agents, systems, vendors or machines do.
8. Reconcile: What actually happened compared with what was expected.
9. Learn: How the Profile, Templates, policies and next decision improve.

The public shorthand is: See the business. Shape what comes next. Make the change.

The complete operating loop is: Profile → Objective → Diagnose → Explore Shapes → Choose a Move → Authorize → Execute → Reconcile → Learn → Re-profile

01Core thesis

Every company has a Shape.

Its Shape is the configuration through which it creates and captures value: what it sells, to whom, through which channels, using which people, assets, systems and capital, with what cost structure, capacity, cash cycle, constraints and risks.

Most business software records activity inside the current Shape. It records the customer, invoice, employee, project, order or asset. It rarely helps management understand why the company produces its present economics, which elements should change, what adjacent configuration is credible, or whether a proposed transformation actually worked.

Allometry builds a living, source-backed Profile of the company. It diagnoses the current Shape and its next breakpoint. It explores improvements and alternative Shapes. It helps management select a Move, prepares and executes approved work, then reconciles expected and realized outcomes.

The company model becomes more accurate each time the business acts.

Allometry therefore connects three forms of intelligence that are normally separate:

• Descriptive intelligence: How does the company actually work?
• Strategic intelligence: What should it improve or become?
• Execution intelligence: What should happen now, under whose authority, and did it work?

The shortest complete expression is: Allometry helps a company see how it works, decide what it should become and make the change.

Category

• The category identity is: Business Shape intelligence
• The functional description is: Strategic planning that executes and learns.
• The system description is: A knowledge, decision and action layer for the company.
• The long-term infrastructure description is: The economic intelligence and authorization layer through which companies, agents and capital coordinate.

These are levels of explanation, not four products.

02The problem

A company's operating truth is fragmented across accounting systems, CRM, ERP, project tools, emails, contracts, spreadsheets, documents, employees and founder memory.

This fragmentation creates five failures.

2.1 The company cannot see itself

Financial statements describe consequences. Operational systems describe activities. Neither necessarily reveals the company's complete causal structure:

• Which customers create attractive contribution after cost-to-serve
• Which products or services consume scarce capacity
• Where cash becomes trapped
• Which offers depend on the founder
• Which channels create repeatable demand
• Which assets are underused
• Which capabilities could support a new product or business line
• Which apparent growth destroys economic value

2.2 Strategy is detached from operating reality

Strategy documents are periodically produced and rapidly become stale. They are rarely connected to live data, operating constraints, explicit assumptions or daily decisions.

2.3 Business software preserves the current configuration

CRM, ERP, accounting and workflow systems are optimized to record and administer what the company already does. They do not ask whether the company should change its offer, customer mix, channel, revenue model, asset configuration, capital structure or ownership model.

2.4 AI produces advice without accountability

Generic models can generate plausible strategies, but they usually lack company-specific truth, economic calculation, authority boundaries and an outcome record. Advice is cheap when no system remembers whether it worked.

2.5 Transformation is treated as a project instead of a learning loop

Companies make major changes through decks, consultants and intuition. The hypothesis, decision, approval, investment, execution and realized result are not preserved as one continuous record. Institutional learning disappears.

The resulting opening is not another dashboard or agent builder. It is a system that makes the company legible, makes alternative futures comparable, makes action governable and makes learning cumulative.

03Business Shape

A Business Shape is the economic and operating morphology of a company at a point in time.

It is represented by a set of connected dimensions:

A Shape is not a label such as "agency" or "manufacturer." Two companies in the same industry can have very different Shapes. One agency may be founder-led and project-based. Another may have productized retainers, proprietary data and distributed delivery. Their constraints, value and next Moves differ.

3.1 Current Shape

The current Shape is the best source-backed representation of how the company operates now.

3.2 Target Shape

A Target Shape is a selected future configuration that management intends to pursue.

3.3 Adjacent Shape

An Adjacent Shape is a credible future configuration that reuses enough of the company's existing demand, distribution, capability, data, brand, assets or relationships to justify exploration.

3.4 Shape distance

Shape distance measures the difficulty of moving from the current configuration to another. It includes:

• New capabilities required
• Capital required
• Time to evidence
• Customer behavior change
• Operational complexity
• Organizational change
• Regulatory burden
• Data and system change
• Reversibility
• Downside recovery

The highest theoretical value is not always the best Shape. The chosen Shape must also fit management intent, organizational capability, time horizon, risk tolerance and willingness to operate it.

3.5 The law: Y = a·M^b

The company is named after a law, and the law is the core of the model.

In biology, a trait Y scales with body mass M as Y = a·M^b. Metabolism scales at roughly the three-quarter power: an elephant burns far less energy per gram than a mouse, because evolution scaled the transport network with the body. An organism is not a size. It is a set of exponents. Change the size without changing the exponents and the organism breaks. There are no elephant-sized mice.

Businesses obey the same law and almost nobody measures it. Revenue is the mass. Cost, capacity, working capital, management load, quoting complexity, customer concentration and margin each scale with it at their own exponent. A company that doubles revenue while margin scales at 0.8 and complexity scales at 1.3 is an elephant-sized mouse: the structure cannot carry the size. Every business has a Shape. The Shape is its exponents.

This gives the Business Shape dimensions in §03 their arithmetic:

• Coefficient. For each Economic Unit, the measured pair (a, b): how that unit's cost, revenue, capacity consumption, cash cycle and risk scale with volume. The Template supplies the prior exponent for the family. The company's Ledger earns the local one from executed work.
• SKU morphology. The exponent map across the company's units: which products, customers, channels, jobs or locations scale the company at a good exponent, and which quietly bend it. Profitable-looking work that consumes disproportionate capacity. Customers whose concentration is a liability. SKUs whose margin exists only in the price book. Morphology is sizing as a product: it turns the P&L into a Shape.
• Adjacent possible. The set of Shapes reachable from the current one where the exponents are favorable: units the company already knows how to make, sell, deliver or finance whose coefficients, added to the current mix, move the whole toward the Objective. An adjacent Shape is not an idea. It is a reachable configuration with a computed effect on the exponents.
• Shape distance. Becomes computable: the capital, time, capability and organizational change required to move the exponents from where they are to where the Target Shape needs them, weighted by reversibility and downside.

The physics of profit sits underneath the law and gives every Move its constraint and its target:

• Work. W = F × d. In the operating economy a unit of work is a call: requested, priced, committed, done, proven.
• The first law: revenue is conserved, profit is not. Every dollar of revenue becomes either profit or heat. Heat is the stale price book quoting January's inputs in August, scope nobody re-priced, the same order keyed into three systems that disagree. Nobody decides to lose margin; it dissipates. At the first design partner, a Quebec pallet manufacturer, ninety days of completed work surfaced approximately $300K in annualized margin leakage patterns. Not one bad decision. Heat.
• The second law: left alone, every price book drifts toward disorder. Costs move weekly, prices move annually. The gap decays quietly, one quote at a time, and never in the company's favor.
• The floor (defense). Π = Σ (p − c(t)) × q, subject to p ≥ (1 + f) × c(t), enforced at the moment of commitment. The sum is the P&L; the constraint is the future; the (t) is cost now, not cost when someone last updated the sheet. A business that enforces the inequality at commitment cannot quietly leak.
• The target (offense). Choose p to move V → V*, subject to the floor. V* is the enterprise value the owners are pricing toward: the Objective. The price of the unit is derived from the value of the whole organism. That is what allometric pricing means.
• The demon. Enforcing the floor costs information: the system must read the ERP, the spreadsheets and the field system continuously and resolve the number that is actually true. That reading never stops. It is the Pulse: the named layer of the product that reads the company's systems continuously, resolves the number that is actually true, and pays the demon's fee in information so that the floor, the Coefficients and the Profile are never stale.

The three sentences that follow from the law are the company: Allometry measures the exponents, decides against them, and enforces the decisions at the moment the company commits. Every commitment is a small vote on the exponents, and the exponents are the company.

04The universal operating loop

The product is organized around one closed loop.

4.1 Profile

Construct the best current, source-backed representation of the company.

4.2 Objective

State what the owners want the company to become, in measurable terms: the enterprise value V* they are pricing toward, and the cash, resilience, control, liquidity or legacy constraints that bound it. The Objective is chosen, not inferred. It is what makes Shape exploration specific instead of generic, and it is the target every Move is scored against. A company can change its Objective; it cannot run the loop without one.

4.3 Diagnose

Explain how the current Shape produces its economics, where value is created or lost, which constraint dominates and what breakpoint approaches.

4.4 Explore Shapes

Generate and compare credible improvements and adjacent configurations. These can include optimizing the current Shape or reshaping the company.

4.5 Choose a Move

Convert strategic possibility into one bounded, measurable change with an explicit hypothesis, expected outcome, cost, owner, time horizon and kill condition.

4.6 Authorize

Apply evidence thresholds, policy, permissions, capital constraints and human approval. Uncertainty is allowed. Unstated uncertainty is not.

4.7 Execute

Prepare and perform the work through people, agents, software, vendors or machines. Use the smallest reliable tool and preserve rollback wherever possible.

4.8 Reconcile

Compare what happened with the expected range. Identify whether the variance came from facts, assumptions, execution, timing or external conditions.

4.9 Learn

Update the Profile, local coefficients, Template priors, policies and future recommendations. The outcome becomes input to the next loop.

4.10 The two speeds

Allometry operates at two connected speeds:

• Strategic loop: Explore and select a future Shape over months or years.
• Operating loop: Detect and execute the next Move over hours, days or weeks.

The Target Shape directs the operating loop. Operating outcomes determine whether the Target Shape remains credible.

05Canonical product objects

The product vocabulary must remain compact. Every interface, Template and tool maps to the same core objects.

No vertical may redefine these objects. A Template may extend them with domain-specific fields and relationships.

06Knowledge Graph

The Knowledge Graph is the source-backed representation of what the company is, how it operates and what has happened.

It contains:

• Entities and typed relationships
• Current and historical values
• Source and provenance
• Confidence and freshness
• Ownership and permission
• Calculated economic measures
• Contradictions and unresolved claims
• Domain-specific extensions supplied by Templates

The Knowledge Graph answers:

• What does the company sell?
• Who buys it?
• How is it delivered?
• Which resources and capabilities are required?
• What are the real unit economics?
• Where do demand, work, cash and risk concentrate?
• Which facts are known, inferred, disputed or missing?

6.1 Claims before certainty

Allometry does not silently turn inference into fact. Every material claim carries:

• Value
• Source
• Source date
• Extraction or calculation method
• Confidence
• Last verification
• Permission scope
• Historical versions
• Contradicting evidence, if present

6.2 Knowledge is broader than structured data

Useful company knowledge exists in:

• Accounting and banking records
• CRM, ERP, FSM and ecommerce systems
• Contracts, proposals and invoices
• Documents, email and calendars
• Websites and public sources
• Interviews and management judgment
• Operational events and sensor data
• Previous decisions and outcomes

The graph normalizes economic meaning without requiring every source system to use the same schema.

6.3 The Profile is not the graph

The Knowledge Graph is the underlying evidence network. The Profile is a governed, human-legible view of that graph for a specific purpose and moment.

A financing Profile, an operating Profile and an acquisition Profile may expose different views of the same governed company truth.

07Profile

The Profile is the source-backed reality of the business today.

It should contain only what is useful for understanding the current Shape, diagnosing it and selecting the next Move.

7.1 Core Profile

• Company identity and ownership intent
• Business archetype and current Shape
• Economic units
• Offers, products, services and SKUs
• Customers, segments and concentration
• Channels and demand system
• Revenue architecture
• Delivery and operating model
• Cost structure and unit economics
• Capacity and utilization
• Assets, capabilities and latent advantages
• Cash cycle and capital requirements
• Organization and decision authority
• Systems and data quality
• Current constraints and breakpoint
• Risk and resilience
• Plausible enterprise-value range and method
• Evidence confidence and missing information

7.2 Profile quality

A Profile is valuable when management says:

• This accurately reflects how our company works.
• It exposes something important we could not previously see.
• It identifies what we should investigate or change next.

Profile quality is measured by evidence coverage, freshness, contradiction rate, management attestation and predictive usefulness, not by document length.

7.3 The Profile is living

The Profile changes when:

• New source data arrives
• Management verifies or rejects a claim
• A Move is executed
• An outcome differs from expectation
• The company crosses a breakpoint
• A new capability, product, location, channel or entity is added
• A source becomes stale or unreliable

Every material change is versioned.

08Diagnose

Diagnosis converts company knowledge into an explanation of the present.

It must distinguish symptoms from causes.

Low margin may be caused by pricing, mix, utilization, scope control, procurement, rework, customer concentration, working capital or an offer that does not fit the delivery system. Growth may be constrained by demand, sales conversion, capacity, management, cash, regulation or founder attention.

8.1 Diagnostic outputs

Every diagnosis should identify:

• Current Shape
• Primary economic unit
• Value-creation engine
• Primary constraint
• Next breakpoint
• Material risks
• Underused assets or capabilities
• Highest-confidence improvements
• Missing evidence that could change the conclusion
• Three credible Shape directions
• One recommended reversible Move

8.2 Breakpoints

A breakpoint is an operating threshold, not merely a revenue band.

Common breakpoints include:

• Founder can no longer personally review every commitment
• Delivery becomes repeatable enough to standardize
• First management layer is required
• First additional product, channel, crew or location is added
• Working capital becomes the binding growth constraint
• Capacity must be financed before demand can be served
• Data fragmentation prevents coordinated decisions
• Acquisition becomes more efficient than organic expansion
• A single company becomes a portfolio or network
• Routine decisions can safely be delegated to software or agents

Allometry should say: You are here. This is what normally breaks next. This is the evidence. This is the smallest system or Move required before you cross it.

8.3 Diagnosis remains contestable

The system must expose:

• Supporting evidence
• Assumptions
• Alternative explanations
• Confidence
• Disconfirming data
• Questions that would reduce uncertainty

The goal is not to sound certain. The goal is to make the next decision better.

09Explore Shapes

Allometry explores both improvement and transformation.

9.1 Improve the current Shape

• Pricing and packaging
• Customer and account mix
• Product, service or SKU mix
• Gross margin and contribution
• Capacity and utilization
• Procurement and cost-to-serve
• Working capital and contract-to-cash
• Channel allocation
• Location economics
• Team and decision design
• Founder dependence
• Retention and recurrence

9.2 Reshape the business

• Add a SKU or product family
• Productize a service
• Add recurring revenue
• Add a software, data or licensing layer
• Enter a new channel or customer segment
• Launch a new location or territory
• Add private label or co-manufacturing
• Franchise or license the operating model
• Reconfigure an asset
• Acquire distribution, capability or capacity
• Sell or close a business line
• Form a platform or holding company
• Partner instead of build
• Spawn a new company when separation creates more value

9.3 Shape generation

Adjacent Shapes should not be generic ideas. They are generated from:

• Existing customer demand
• Unmet requests and lost opportunities
• Distribution advantage
• Underused capabilities and assets
• Proprietary data or methods
• Existing brand permission
• Supplier and partner relationships
• Operational adjacency
• Market structure and competition
• Founder and management fit
• Capital and time constraints
• Previous outcomes from similar Moves

9.4 Scenario comparison

Every serious Shape is compared on:

• Revenue and gross-profit range
• EBITDA and cash generation
• Capital and working-capital requirements
• Time to first evidence
• Time to meaningful scale
• Organizational and operational complexity
• Founder or management involvement
• Risk and downside recovery
• Reversibility
• Strategic option value
• Plausible enterprise-value range
• Confidence and evidence still required

Allometry uses ranges and sensitivities. It does not manufacture false precision.

10Choose a Move

A Shape is a direction. A Move is the smallest meaningful change that creates evidence or advances the company toward that Shape.

Examples:

• Test a new price with one customer segment
• Interview ten customers about a recurring offer
• Launch one productized service
• Pre-sell a new SKU before production
• Reallocate one asset or crew to a higher-value segment
• Remove one unprofitable service
• Pilot one partner channel
• Convert five customers to a recurring contract
• Open a temporary location before signing a permanent lease
• Screen ten acquisition targets before retaining an intermediary
• Automate one bounded delivery step
• Prepare one financing package for a proven capacity constraint

10.1 Move contract

Every Move contains:

• Target Shape
• Current-state baseline
• Hypothesis
• Supporting and contradicting evidence
• Expected outcome and range
• Economic unit affected
• Required actions and Tools
• Owner and accountable decision maker
• Required authority
• Time and capital budget
• Dependencies
• Measurement window
• Success threshold
• Kill, pause and escalation conditions
• Reversibility and rollback path
• Actual outcome, once observed

10.2 Portfolio of Moves

A Plan is not a static roadmap. It is an ordered portfolio of Moves with dependencies and evidence gates.

The next Move is selected by expected value, information gain, urgency, strategic fit, resource requirement, reversibility and confidence.

The system should prefer the Move that most efficiently changes what management knows or improves what the business earns.

11Decision Graph

The Decision Graph represents what the company could do, why it might do it and how a choice becomes an authorized action.

The Knowledge Graph describes reality. The Decision Graph describes possibility, causality, choice and authority.

The Decision Graph connects:

• Objective
• Current Profile
• Constraint or opportunity
• Alternative Shapes
• Scenarios and assumptions
• Hypotheses
• Proposed Moves
• Expected outcomes
• Dependencies
• Policies and authority
• Decisions and approvals
• Commitments
• Work Calls
• Outcomes and variance
• Resulting learning

11.1 Decisions must preserve rejected alternatives

Allometry records not only what management chose, but what it rejected and why. This prevents hindsight from rewriting the original decision.

11.2 Decision quality is separate from outcome quality

A good decision can produce a poor result because uncertainty resolves badly. A poor decision can produce a good result through luck. The Ledger must preserve both the quality of the original reasoning and the realized result.

11.3 The graph is domain-agnostic

The same structure can represent:

• A price change
• A new SKU
• A hiring decision
• A marketing allocation
• A location opening
• A capacity investment
• A hotel repositioning
• A service productization
• An acquisition
• A divestiture
• A financing decision

Templates supply the domain logic. The graph preserves the universal decision structure.

11.4 Two inlets

The Decision Graph is fed from two directions, and they are ranked differently.

• Sovereign inlet (internal). The company's own Knowledge Graph: executed work, invoices, payments, contracts, capacity, management attestations. This is the truth about the company. It is private to the Node and it outranks every external claim about the same fact.
• Market inlet (external). The Index and everything outside the company: price, cost and capacity ranges, demand and payment signals, competitor and supplier Profiles, public filings, licensed datasets, scraped sources, the TAM Builder's targets, partner and buyer agents. This is the truth about the market the company operates in. It supplies priors, comparables, opportunities and counterparties, never the company's own facts.

Both inlets carry provenance and confidence (§21.1). A Move is scored on both: what the sovereign inlet says the company can do, and what the market inlet says the market will pay, supply or demand.

11.5 The graph prices Moves with the law

Every proposed Move is expressed as a change in coefficients: which Economic Units it adds, removes or reprices, and how the company's exponents move as a result. The expected outcome of a Move is the predicted change in V toward V*, under the floor, with a range. The realized outcome is the re-measured exponent after the measurement window. Variance between the two is what the Ledger learns from. This is what makes the Decision Graph allometric rather than a generic strategy tree.

12Governance and policy

Allometry applies deterministic governance to uncertain intelligence.

12.1 Policy types

• Economic thresholds
• Capital and spending limits
• Margin floors
• Risk limits
• Data-access permissions
• Tool permissions
• Approval authority
• Customer or supplier restrictions
• Legal and compliance requirements
• Model and compute budgets
• Autonomy limits
• Escalation and rollback requirements

12.2 Authority belongs to the company

Models may recommend. Tools may prepare. Agents may execute only within delegated authority.

The system must always know:

• Who requested the action
• Which evidence was used
• Which policy applied
• Who or what approved it
• What side effect was authorized
• What can reverse it
• Where the outcome will be recorded

12.3 Govern the Tool, not the personality

Autonomy is granted to a bounded Tool and action class, not to a vaguely defined agent.

For example, the company may allow automatic follow-up emails for approved prospects but require human approval for price changes, contracts, bank transfers, hiring, capital deployment and external commitments.

13Tools and the Operator

A Tool is a bounded capability that can read, calculate, recommend, prepare or act.

The Operator is the orchestration layer that selects and uses Tools in service of an approved Move.

13.1 Tool families

Understand: Ingest a source · Extract claims · Resolve entities · Build or update a Profile · Identify missing evidence · Explain a metric or relationship

Diagnose: Calculate unit economics · Detect a constraint · Identify a breakpoint · Find margin or revenue leakage · Compare performance by customer, SKU, location or unit · Surface operational anomalies

Explore: Generate adjacent Shapes · Simulate a Shape · Compare scenarios · Value a current or target Shape · Map new products, channels or acquisitions · Test sensitivity and downside

Decide: Propose a Move · Rank Moves · Build an approval package · Check policy · Request approval · Record a decision

Execute: Conduct research · Prepare customer interviews · Create an offer or SKU brief · Build a landing-page experiment · Prepare outreach · Generate a proposal or quote · Update an approved system record · Issue a Work Call · Prepare a financing package

Learn: Collect an outcome · Reconcile expected and realized results · Explain variance · Propose a coefficient or policy update · Back-test a proposed change · Update the Profile after approval

13.2 Tool contract

Every production Tool defines:

• Purpose
• Input and output schema
• Required sources
• Deterministic calculations
• Permitted models
• Confidence requirements
• Permission scope
• Approval requirement
• Side effects
• Idempotency behavior
• Failure and escalation behavior
• Rollback path
• Ledger event
• Evaluation suite

13.3 Tool principles

• Use deterministic logic when the answer is knowable.
• Use specialized models when a repeated pattern can be learned reliably.
• Use frontier models for ambiguity, synthesis and novel scenarios.
• Use humans when consequences exceed delegated authority or evidence is insufficient.
• Never let model fluency substitute for provenance.
• Never ask a person to re-enter information already available from an authorized source.

13.4 Operator behavior

The Operator follows a stable sequence:

1. Read the active Profile and Target Shape.
2. Identify the highest-value gap or event.
3. Propose the next Move or task.
4. Select the smallest reliable Tool.
5. Apply policy and request approval where required.
6. Execute or prepare the work.
7. Observe the response and outcome.
8. Write the complete event to the Ledger.
9. Update the Profile and next recommendation.

The user should experience one accountable system, not a theatre of colorful agents.

13.5 The Pulse

The Pulse is the Understand family running continuously rather than on request. It reads every connected Source on its own cadence, extracts Claims with provenance, resolves entities, recalculates Coefficients when executed work lands, and raises the events that feed the action queue in §37. It is the product's heartbeat and the demon's fee: the reason the floor can be enforced at commitment time is that the Pulse has already resolved today's cost. The Pulse has no authority. It reads, resolves and raises; the Operator, the Policies and the humans decide.

14Execute, Commitments and Work Calls

Execution translates an approved Move into work.

14.1 Commitment

A Commitment is an authorized obligation of money, capacity, time, service or risk.

Examples include: sending a quote · signing a contract · ordering inventory · reserving a crew or asset · hiring an employee · buying advertising · opening a location · acquiring a company · drawing a credit facility.

The Commitment is no longer the atomic object of Allometry. It is a critical state transition through which many Moves become economically real.

14.2 Work Call

A Work Call is a structured request for an actor to perform bounded work.

The actor may be: a person · an internal team · an AI agent · a software system · a vendor · a contractor · a conventional machine · a robot.

A Work Call contains:

• Purpose and parent Move
• Scope and acceptance criteria
• Required capability
• Inputs and authorized context
• Economic envelope
• Capacity and timing requirements
• Risk and safety policy
• Approval authority
• Compensation or settlement terms
• Escalation and fallback
• Required outcome proof

14.3 Relationship between Move and Work Call

One Move may generate many Work Calls. A new product launch may require customer research, design, supplier qualification, pricing, production, sales enablement and measurement. Each unit of work remains connected to the original hypothesis and Target Shape.

This is the essential distinction: The Move explains why the company is changing. The Work Call specifies what an actor must do. The Commitment records what the company has authorized or promised. The Ledger proves what happened.

14.4 The Work API

The Work API is the eventual machine-readable interface through which authorized systems can:

• Describe capability
• Check capacity
• Evaluate work
• Request or prepare a quote
• Check economic policy
• Request approval
• Record a commitment
• Issue a Work Call
• Attest an outcome

It begins inside one company. External agent-to-agent and physical-capacity coordination follow only after internal Tools and outcome records work reliably.

14.5 Agent-to-agent commercialization

Stripe collapsed "accept a payment" into one call with a scope, a price and a receipt, and every business became able to take money from software. Allometry collapses a harder question into one call: is this work worth doing for this company, at what price, and can it deliver. That is agent-to-agent (A2A) commercialization, and it is the Work API's purpose.

"Worth doing" is answered from the company's Shape, not from a rate card:

• Business model. Which Economic Unit the request maps to and what that unit's coefficients say about contribution after cost-to-serve at today's c(t).
• Intent. Whether the work moves the exponents toward the Objective, or bends them away from it. A profitable job that consumes the capacity a better customer needs is not worth doing.
• Capacity. Whether the real calendar, crews, machines, inventory or cash can deliver it in the window.
• Policy. Margin floor, customer risk, concentration limits, approval thresholds.

A buyer's agent sends a request with a spec, a deadline and a budget. The company's Node answers in seconds with a governed quote, a request for more information, a decline or an escalation to a human, and writes the exchange to the Ledger either way. The operators whose businesses can answer a call like that get found and get paid first. The floor, the Pulse and the Objective are how a business becomes able to answer.

Sequence discipline holds: A2A runs inside one company first (the company's own agents asking its own Node), then between a company and its known partners, and only then in the open network described in §45.

15The Allometry Ledger

Allometry has one unified Ledger with multiple views.

The Ledger is the durable record of what the company knew, believed, decided, authorized, did and learned.

Each complete entry binds:

• Trigger or signal
• Relevant Profile version
• Evidence and sources
• Diagnosis or opportunity
• Alternatives considered
• Hypothesis
• Expected outcome and range
• Decision and rationale
• Policy and approval
• Move and resulting Commitments
• Tools and Work Calls used
• Execution record
• Realized outcome
• Variance explanation
• Learning and approved updates

15.1 Ledger views

The single underlying event history can be viewed as:

• Evidence view: What source or claim changed?
• Decision view: Why did management choose this path?
• Commitment view: What money, capacity, time or risk was authorized?
• Execution view: What work was performed and by whom?
• Outcome view: What happened operationally and economically?
• Learning view: What should change in the Profile, Template or policy?

This replaces separate branded ledgers. The Hypothesis Ledger and Attested Operating Ledger remain useful conceptual views, not separate products.

15.2 The Ledger is the institutional memory

Conventional systems record transactions and task status. Allometry records the decision lineage behind them.

A CRM may record that an opportunity closed. Accounting may record the invoice. Allometry records why the customer was pursued, which price and capacity assumptions were used, who approved the commitment, what outcome was expected, what happened and how the next decision should change.

15.3 Learning requires reconciliation

No Move is complete when the action is launched. It is complete when the measurement window closes, the outcome is reconciled and the resulting learning is accepted or rejected.

The system must distinguish variance caused by:

• Incorrect facts
• Incorrect assumptions
• Weak causal reasoning
• Poor execution
• Insufficient time
• External change
• Measurement failure
• Randomness

15.4 Controlled improvement

The learning sequence is: Record → Detect variance → Explain → Propose update → Back-test → Shadow-test → Approve → Promote → Monitor → Roll back if required

Allometry may automatically update low-risk descriptive coefficients when evidence is strong. It may not silently change accounting logic, capital hurdles, margin policy, risk limits or delegated authority.

16Templates

The kernel is horizontal. Usefulness arrives through Templates.

Agnostic does not mean generic. One company language, specialized economic grammar.

A Template provides priors, questions, calculations, diagnostics, Shape possibilities, Move playbooks, Tools and outcome measures for a particular context.

16.1 Template stack

Templates can compose across four levels:

A company may use several Templates simultaneously. A hotel with events, food service and branded products should not be forced into one vertical box.

16.2 Shape Template contract

Every Shape Template contains:

• Economic units
• Ontology extensions
• Required and optional evidence
• Financial and operating calculations
• Typical revenue and delivery architectures
• Common breakpoints
• Constraints and failure modes
• Best practices by breakpoint
• Common adjacent Shapes
• Valuation methods and quality adjustments
• Move playbooks
• Relevant Tools
• Policies and approval defaults
• Outcome metrics
• Confidence and applicability rules
• Known exceptions

16.3 Priors are not conclusions

A Template begins with industry and business-model priors. The company Profile replaces those priors with local truth as evidence accumulates.

The Template supplies a useful starting point. The company's Ledger earns the final coefficient.

16.4 Template quality

A Template is good when it:

• Produces an accurate first Profile faster
• Identifies a material breakpoint or constraint
• Generates relevant rather than generic Shapes
• Reduces implementation time
• Improves Move outcomes
• Reuses the canonical objects without forking the architecture

If a vertical repeatedly requires a separate product architecture, the horizontal thesis has been falsified or the core ontology is incomplete.

16.5 Initial Template families

Allometry may describe many Templates publicly while building depth only where evidence and design partners exist.

Physical and asset-enabled services

• Economic units: job, route, crew-day, asset-hour, contract
• Current Shape questions: true job contribution, utilization, backlog quality, capacity and working capital
• Typical Moves: reprice work, improve estimating, change customer mix, add recurring maintenance, acquire capacity

Professional and business services

• Economic units: account, engagement, deliverable, person-hour, retainer
• Current Shape questions: founder dependence, utilization, scope leakage, recurrence and concentration
• Typical Moves: productize a service, add recurring revenue, license methods, build software or data products

Product, food and CPG businesses

• Economic units: SKU, batch, order, channel, customer, production line
• Current Shape questions: landed contribution, trade spend, returns, inventory and channel profitability
• Typical Moves: add or remove SKUs, private label, co-manufacturing, direct channel, own production

Hospitality and multi-location businesses

• Economic units: room-night, table, event, visit, location, franchise unit
• Current Shape questions: occupancy, rate, contribution by experience or channel, labor model and property economics
• Typical Moves: reposition, add experience or product revenue, franchise, acquire, convert space or centralize operations

B2B software and recurring businesses

• Economic units: account, contract, seat, usage, workflow
• Current Shape questions: retention, expansion, service burden, channel efficiency and product concentration
• Typical Moves: change packaging, add usage pricing, enterprise motion, services layer, marketplace or embedded finance

Real estate, healthcare and institutional assets

• Economic units: asset, lease, bed, procedure, visit, location, operating contract
• Current Shape questions: utilization, service mix, regulatory constraints, capital intensity and operator-property separation
• Typical Moves: reposition assets, change use, add services, separate operating and property economics, consolidate or partner

Template order follows customer evidence, not intellectual completeness.

17The horizontal commercialization model

Commercialization is how a company converts an asset, capability, product or idea into repeatable economic value.

It is broader than outbound sales or CPQ.

Commercialization can mean:

• Repricing an existing offer
• Packaging a service
• Creating or removing a SKU
• Entering a channel
• Changing a customer segment
• Adding a recurring layer
• Launching a location
• Licensing an operating method
• Selling through partners
• Adding private label
• Franchising
• Acquiring distribution or capability
• Converting an asset to a higher-value use
• Adding financing at the point of demand
• Building a marketplace or network

The system first understands the company's Shape and economic unit. It then selects the relevant commercialization logic through Templates.

This produces one horizontal question: Given what this company already knows, owns, reaches and can deliver, what is the highest-value credible way to improve or reshape it?

17.1 Optimize versus reshape

Allometry must make the distinction explicit:

• Optimize: Make the current Shape stronger.
• Reshape: Change the configuration through which the company creates value.

Optimization can create the evidence, cash and capacity needed for reshaping. Revenue Recovery, margin improvement and working-capital gains are valuable partly because they fund and de-risk the next Shape.

17.2 Build, buy, partner or stop

Every material opportunity should consider more than internal development:

• Build organically
• Buy an existing company, product, capability or distribution channel
• Partner or license
• Automate or augment
• Separate into a new entity
• Wait for more evidence
• Stop or divest

Allometry is agnostic to the vehicle. It evaluates the Shape and the Move.

18Capital

Capital is a resource required by some Moves, not a disconnected product category.

Every proposed Move should state:

• Capital required
• Timing of cash outflow
• Working-capital burden
• Expected return and range
• Time to first evidence
• Time to payback
• Downside recovery
• Collateral or security, if applicable
• Dependencies and milestones
• Effect on liquidity and resilience
• Most appropriate capital type

18.1 Capital follows an evidence-backed Move

The sequence is: Profile → Constraint or opportunity → Shape → Move → Evidence package → Capital decision → Execution → Outcome

The capital layer should help management decide:

• Whether the Move deserves funding
• How much capital it can responsibly absorb
• Whether internal cash, debt, equity, leasing, revenue share or a partner is the best instrument
• Which milestones should release additional capital
• What evidence a capital provider requires

18.2 Financing package

When a validated Move meets a capital constraint, Allometry can prepare:

• Source-backed company Profile
• Move hypothesis and economic rationale
• Base, upside and downside cases
• Contract, backlog or demand evidence
• Capacity and execution plan
• Historical estimate-to-outcome accuracy
• Cash-conversion schedule
• Repayment or return source
• Policies, approvals and covenants
• Measurement and reporting plan

18.3 Capital provider interface

Over time, permissioned Profiles and Ledger evidence can make companies and specific Moves more legible to:

• Banks and alternative lenders
• Equipment lessors
• Revenue-based financiers
• Insurers
• Investors
• Acquirers
• Public programs
• Strategic partners

Allometry begins as the intelligence, preparation and monitoring layer. It does not become a balance-sheet lender before the Ledger earns underwriting credibility.

18.4 Capital is outcome-accountable

The Ledger connects capital deployed to the Move it funded and the outcome it produced. This creates a living capital-allocation history rather than a collection of disconnected budgets and investment memos.

19Node and Index

19.1 The Allometry Node

The Node is the sovereign economic intelligence environment for one company.

It contains: Knowledge Graph · the Pulse (continuous reading and Coefficient recalculation) · current and historical Profiles · Decision Graph · Shape and Scenario models · Templates and local coefficients · policies and permissions · Tool registry · Operator · Ledger · model router · APIs and MCP interfaces.

The Node can be deployed in Allometry's cloud, a customer-controlled cloud environment or a private environment when customer needs and economics justify it.

19.2 The Allometry Index

The Index is the permissioned intelligence layer across companies, portfolios and markets.

It may contain:

• Public and claimable company Profiles
• Template priors
• Aggregated breakpoint patterns
• Shape transition patterns
• Price, cost and capacity ranges
• Move performance benchmarks
• Capability and supplier discovery
• Payment and execution signals
• Acquisition and partnership opportunities
• Capital needs and financing performance

The Index does not require centralizing raw company data. Nodes may contribute authorized, aggregated or privacy-preserving outputs.

19.3 The compounding loop

The Index supplies priors. The Template supplies context. The Node supplies company truth. The Decision Graph selects a Move. The Ledger supplies the outcome. The outcome improves the next prior.

19.4 Portfolio use

For a portfolio, the Index can compare companies without erasing their local Shapes. It supports:

• Underwriting and diligence
• Thesis-to-outcome tracking
• Cross-company benchmarks
• Shared procurement and distribution
• Capacity allocation
• Acquisition integration
• Management intervention
• Capital allocation
• Internal markets for demand or capability

One Node represents each company. The portfolio Index compares and coordinates them.

20Model, system and interface agnosticism

Allometry must be agnostic where the underlying component is replaceable and opinionated where economic meaning must remain stable.

20.1 Model-agnostic

Use the most reliable and economical method for each task:

• Deterministic calculations for known financial truth
• Rules for stable policy
• Statistical models for repeated forecasts
• Small or open models for private, frequent or narrow work
• Frontier models for ambiguous synthesis and novel reasoning
• Humans for consequential judgment and exceptions

No foundation model is the moat.

20.2 Source-system-agnostic

Allometry reads from and writes to existing accounting, CRM, ERP, FSM, ecommerce, banking, document and workflow systems. It does not require a company to replace them before receiving value.

20.3 Interface-agnostic

The same governed capability can appear through: Allometry's control room · Allo · Slack or Teams · email · CRM or ERP extensions · API · MCP · voice · partner applications · authorized buyer or supplier agents.

Allometry is headless by architecture but not faceless by product.

20.4 Actor-agnostic

A Work Call can be executed by a human, agent, vendor, system or machine. The economic intent, policy and outcome record remain consistent.

20.5 Capital-agnostic

Allometry evaluates the suitability and consequences of capital. It does not privilege debt, equity, leasing, internal cash or a particular provider unless the company's policies and evidence justify it.

20.6 Industry-agnostic, Template-specific

The canonical objects and loop remain stable across industries. Economic units, diagnostics, Shapes, policies and Tools become specific through Templates.

This is the design rule: Stable kernel. Composable Templates. Governed Tools. Local truth.

21AI architecture

Allometry combines deterministic systems and probabilistic intelligence.

21.1 Truth hierarchy

1. Source-backed transactional facts
2. Deterministic transformations and calculations
3. Verified management assertions
4. Statistical estimates
5. Model inferences
6. Template priors

Lower-ranked evidence may fill a gap. It may not silently override stronger evidence.

The market inlet (§11.4) has its own hierarchy, and no rung of it outranks the sovereign inlet on a fact about the company itself:

1. Attested outcomes contributed to the Index by other Nodes
2. Licensed or contractual datasets with known method and date
3. Public filings and regulatory records
4. Verified counterparty Profiles (claimed and attested)
5. Scraped and unverified public sources
6. Model inference about the market

21.2 Routed intelligence

For every task, the router considers: required accuracy · consequence of error · data sensitivity · latency · cost · context requirement · tool-use reliability · need for explanation · available evaluation evidence.

The economic router chooses the cheapest reliable path, not the cheapest model.

21.3 Company memory

Persistent memory belongs in the Knowledge Graph, Decision Graph and Ledger, not inside a model conversation.

Prompts are replaceable. Company context, source lineage, policy, decisions and outcomes are durable.

21.4 Evaluation

Every important model or Tool is evaluated against: extraction accuracy · calculation agreement · citation and provenance quality · tool-selection accuracy · policy compliance · scenario calibration · decision usefulness · outcome prediction · recovery and escalation behavior.

21.5 Self-correction without self-government

Allometry can propose improvements to models, coefficients, Templates and policies. Material changes require evidence, back-testing, shadow mode, approval, versioning and rollback.

The system may improve itself. It may not grant itself authority.

21.6 The Oracle rule

The Oracle is the conversational interface to the company: ask it anything, get a module back. It is an interface to Tools, not a model with opinions. Every answer is a governed Tool run with provenance, confidence and a Ledger event, or it is not an answer. Token optimization is the router's job (§21.2: the cheapest reliable path), never the answer's: the Oracle may choose a smaller model, a cached calculation or a deterministic rule to reply cheaply, but it may not trade provenance for fluency. A free-form model reply never establishes a fact about the company. This is what separates the Oracle from the generic advice §2.4 condemns.

22System of record and workflow

Allometry does not initially replace the systems companies already use.

• CRM records customers and opportunities.
• ERP or FSM records orders, work and operational transactions.
• Accounting records invoices, payments and financial consequences.
• Allometry records how the company is shaped, why it chose to change, what it authorized and what it learned.

Allometry becomes: The system of record for Business Shape, economic decisions and transformation outcomes.

22.1 Zero duplicate entry

Never ask a person to re-enter information that already exists in an authorized source.

Allometry should:

• Read existing records
• Resolve and normalize their meaning
• Ask only questions that materially reduce uncertainty
• Prepare the action
• Request approval
• Write approved results back
• Preserve its own decision lineage

22.2 Authority progression

Allometry earns authority progressively:

1. Observe: Build the Profile and detect issues.
2. Recommend: Suggest a Shape or Move.
3. Prepare: Assemble the work and approval package.
4. Approve: Become a required review point for defined actions.
5. Delegate: Execute routine actions inside explicit policy.

The defensibility transition is: Insight → Habit → Workflow → Authority → Infrastructure

23Product and brand architecture

23.1 Allometry

Allometry is the company and full Business Shape platform for established companies, management teams, portfolios and capital partners.

It provides: connected company Profile · the Pulse · Knowledge and Decision Graphs · Shape Templates · Scenario Studio · Move planning · governance and approvals · Operator and Tools · Ledger · Node and Index · Capital and Work Call extensions.

23.2 Allo

Allo is the friendlier, lower-friction product for solo founders, creators, freelancers, agencies and small teams.

It provides: website-first Profile · guided founder questions · current Shape and breakpoint · three adjacent Shapes · founder-fit and lifestyle considerations · one Move at a time · lightweight workspace · conversational guidance · optional execution credits.

Allo uses the same core objects and Templates. It hides enterprise governance and infrastructure complexity.

23.3 Shape Scan

Shape Scan is the acquisition and activation experience.

Inputs may include: company URL · industry and business model · revenue and team range · products, services and channels · owner or management aspiration · diagnostic questions · optional financial and operating files.

The output is: current Profile summary · current Shape · primary economic unit · next breakpoint · primary constraint · three adjacent Shapes · recommended Move · data-confidence score · evidence required next.

23.4 Language hierarchy

Customer-facing language should remain simple:

• See your business
• Find what is holding it back
• Explore what it could become
• Choose the next Move
• Put it into action
• Learn from the result

Knowledge Graph, Decision Graph, Node, Index, MCP and model routing appear only when the audience needs architectural detail.

24The first production product

The first product proves one complete learning loop. It does not imitate the final platform.

24.1 Initial experience

1. Connect a company URL and answer structured questions.
2. Add financial and operating evidence manually or through limited integrations.
3. Generate a source-backed Profile.
4. Confirm or correct key claims.
5. Diagnose the current Shape, constraint and breakpoint.
6. Generate three credible Shapes.
7. Compare scenarios using explicit assumptions and ranges.
8. Select one Move.
9. Generate the hypothesis, execution plan and approval package.
10. Execute one bounded experiment.
11. Record the result and reconcile variance.
12. Update the Profile and recommendation.

24.2 Minimum production objects

• Workspace and company
• Source and Claim
• Entity and Relationship
• Economic Unit
• Profile and Profile Version
• Breakpoint and Constraint
• Shape and Scenario
• Hypothesis and Move
• Policy and Approval
• Tool Run
• Outcome and Ledger Entry

Commitment, Work Call, capital package and Index contribution can remain limited extensions until a real customer Move requires them.

24.3 Minimum product promise

In less than ten minutes, see a credible first model of your company. Within one working session, identify a meaningful breakpoint and select one evidence-producing Move.

The deeper paid product connects evidence, executes the Move and measures what happens.

24.4 Production discipline

• Financial truth is calculated deterministically.
• Material claims show sources and confidence.
• Every scenario exposes assumptions.
• No consequential external action occurs without explicit authority.
• Every Tool Run is logged.
• Every Move has a measurement window.
• Every result can update the Profile only through a governed event.
• Tenant data remains isolated and exportable.

25Initial commercialization

Horizontal product identity does not require horizontal distribution on day one.

Allometry launches Template by Template, through a concrete problem that exposes the larger Shape opportunity.

25.1 Door versus destination

• Door: A costly, urgent problem with available evidence and measurable value.
• Destination: The full Business Shape loop.

The Door may vary. The product loop and ontology do not.

25.2 Revenue Recovery

Revenue Recovery remains a strong initial offer where it matches the Template: Find value hiding between contracts, operations and accounting, then help convert it into invoices and cash.

It can surface:

• Completed but unbilled work
• Missed change orders
• Unapplied contractual increases
• Usage above contracted limits
• Missing travel, equipment, fuel or rush charges
• Old renewal pricing
• Duplicate supplier payments
• Unclaimed credits or rebates
• Receivables blocked by missing evidence

Its strategic value is not only recovered cash. It reconstructs the company's economic model and creates trust for future Moves.

25.3 The expansion sequence

Find value → Build the Profile → Diagnose the Shape → Choose a Move → Execute → Prove the outcome → Expand authority

25.4 Primary buyers

Founder or CEO · owner-operator · CFO or finance leader · COO · commercial or strategy leader · PE operating partner · holding-company principal.

The first buyer must own both the economic problem and enough authority to act on it.

25.5 The Door names

A Door needs a name an operator will recognize in a cold email. The category name never appears on a Door.

Only the first Door is open. The others are named so that the second and third Template families can open without a rename.

25.6 The distribution machine

Distribution is a Move like any other: bounded, measured, reconciled.

• The list. A curated set of operators inside the first Template, ranked by fit, with one judged contact per company, enriched and verified. The top of the list is worked by the founder; the remainder runs on sequenced email from warmed domains. Nothing goes out that the founder would not send by hand.
• The cadence. Sixty-five new companies a day. Day 0 is the story: former investor turned revenue operator, a hundred-plus conversations, three pains, one Door. Day 3 is free work on the operator's own quotes. Day 6 is the insight and the terms. Day 9 is the binary close. LinkedIn on day 2 for the silent. No price before day 6.
• The target. One thousand operators contacted by October 1, 2026. Ten design-partner slots. Three to five paying by the December gate.
• The partner channel. A channel partner runs exploratory conversations with operators in its own book and hands over the ones with a live pricing or margin problem. Those conversations are reported as partner-run, never as Allometry's own, and count as Stage 0 engagements for the first Template.
• The Labs. Twelve narrow hypothesis pages, one per pain cluster from the interview grid, each tagged so that a scan request can be traced to the lab that produced it. A lab that produces no qualified scan in eight weeks is closed. A lab that does becomes a Door candidate for its Template.
• The funnel ladder. Contacted → replied → conversation → data-readiness call → data access → scan delivered → design partner → paying → expansion. Every step is counted in one place. Nothing is reported that is not counted.
• What is not done. No paid acquisition before the funnel is instrumented end to end. No benchmark, index or performance claim before ten operators and twelve attested months. No public price for enterprise. No named design partner.

25.7 Why now

Three things changed in the last eighteen months, and the Door only works because all three did.

1. Agents can hold a setpoint. Control loops that once needed a control engineer now run on a language model with tools: read the sensor, compare to the target, act, repeat. A margin floor is a setpoint. Holding a price to a floor while costs drift is the same loop, run on the books.

2. Software is about to buy physical work by API. It already buys compute, payments and intelligence as a call with a scope, a price and a receipt. The next thing a customer's software will ask for is a pallet run, a crew, a machine-hour, with a spec, a deadline and a budget, expecting an answer in seconds. Operators whose businesses can answer get found and get paid first.

3. The systems of record have stopped being the moat. ERP, CRM and accounting are now cheap to read from and write to. The value moved from recording the company to deciding for it. The vendor that owns the decision lineage owns the account.

4. Capital is funding systems of reality. Large seed rounds are now going to companies whose pitch is that the system of record is not enough and that leaders need a live picture they can act on. That thesis is being validated in revenue teams and SaaS. Allometry is the same thesis for the physical economy, with two differences the market has not yet funded: the law that says what the picture should be, and enforcement at the moment of commitment rather than a report after it.

None of this required a new model. It required the loop, the law and the door.

25.8 The ninety-day calendar

From this revision to the December gate, by week. Each line is counted in the funnel ladder.

25.9 The objection map

The five objections that end conversations, and the answer that keeps them open.

26Packaging and business model

Pricing should increase as Allometry moves closer to connected data, governed execution and realized value.

26.1 Pricing principles

• Do not charge only for seats when value comes from decisions and outcomes.
• Do not promise pure outcome pricing when attribution cannot be governed.
• Separate recurring platform value from variable execution cost.
• Use caps, approval limits and transparent work units.
• Allow deferred, lower-cost processing for non-urgent work.
• Price increased authority only after trust and evidence exist.

26.2 Possible value dimensions

Companies, entities or locations modeled · revenue or capital under analysis · active Shapes and scenarios · Tool runs and execution work · decisions governed · outcomes reconciled · value recovered or protected · capital packages monitored · API or Work Call volume.

The business model may combine subscription, implementation, execution and outcome components without changing the product identity.

26.3 Pricing hypothesis under test

The physical-services Template's Door carries the first concrete pricing hypothesis. It enters live test on October 1, 2026 and is measured against the December gate.

These are hypotheses, not canon. The Ledger decides which survive. Every other Template's Door prices from its own economic unit when its Stage 0 engagements produce evidence.

26.4 The business-model math

Stated so it can be falsified.

• Contract value. The floor at $2,500 a month is $30,000 of annual contract value per operator. The read tier is $9,600. Enterprise and portfolio are custom and are not in the plan until one closes.
• The ladder. One hundred operators on the floor is $3M of annual recurring revenue. That is the end-2027 milestone. It requires no enterprise deal, no second Template and no outcome pricing. Anything above it comes from expansion, not from the base case.
• Cost to serve. The floor is software plus a weekly founder session during the design-partner period, then software alone. The binding constraint is deployment time per operator, not compute. The falsifier in §29.4 (deployment never compresses) is the falsifier of this ladder.
• Gross margin. Model routing (§21.2) keeps inference below ten percent of contract value at the floor tier; deterministic calculation carries the rest. If inference exceeds that, the router is wrong, not the price.
• Expansion. Read tier → floor → recovery engagements → portfolio. Each step is a Move the operator authorizes after the previous one reconciled. Net revenue retention is a lagging measure of whether the loop closes.
• What compounds. Every operator on the floor adds attested coefficients to the first Template. The second operator deploys faster than the first because the priors are better. That is the only mechanism by which deployment compresses, and it is measured.

26.5 The counted market

The market is counted bottom-up from named operators, not modeled top-down from a category.

• The pool. Roughly 17,800 North American physical operators in the first Template's revenue band, identified by name, enriched and ranked. This is the list the distribution machine works from.
• The curated set. About 2,300 of them fit the Door precisely: quoting-heavy, capacity-constrained, running an ERP or field system, with a named commercial or finance owner. The end-2027 milestone of one hundred operators on the floor is roughly four percent of this set.
• The Door's ceiling. At the floor's contract value, the curated set alone is a $69M annual revenue ceiling for the first Door, before enterprise, portfolio, recovery engagements or any second Template. The ceiling is not the plan; it is the reason the plan does not require a second Template to reach $3M.
• What the second Template adds. Each Template family is counted the same way before it opens. No family opens on a modeled market.

26.6 What to diligence

An investor should be able to verify the thesis without trusting the founder. Everything below is either in the Ledger or in the counted funnel.

1. The design partner's ninety-day scan: the surfaced leakage patterns, and the Coefficients that reproduce them (Appendix F test).
2. The floor's shadow-run log: every quote, the floor it was checked against, and the miss rate.
3. The funnel ladder, end to end, with dates: contacted, replied, conversations, data access, scans, slots, paying.
4. The Labs' verdicts and the scan requests attributable to each.
5. The partner-run conversation log, labeled as partner-run.
6. The Template contract for the first family and the object schema in Appendix H, as code.
7. The December gate result, reported separately: verified recovered cash, recoverable invoiced value, identified leakage, annualized opportunity.

Anything not on this list is narrative, and narrative is not evidence.

27Defensibility

The moat is not access to a foundation model. It compounds through:

27.1 Company truth

The source-backed, longitudinal representation of how each company actually works.

27.2 Decision history

The record of alternatives, assumptions, approvals, Commitments and reasons behind action.

27.3 Expected versus realized outcomes

The proprietary dataset connecting company context and management choices to economic results.

27.4 Template calibration

Breakpoints, coefficients, Shape adjacencies and Move playbooks that improve with attested evidence.

27.5 Governance and authority

Policies, permissions, approvals and embedded workflows that make Allometry a trusted decision point.

27.6 Tool evaluations

Evidence about which model, workflow, Tool or human performs a business task reliably under which conditions.

27.7 Network priors

Permissioned cross-company learning that improves cold-start diagnosis without exposing raw customer data.

The compounding unit is: Profile context + Move + expected result + authorized execution + realized outcome + variance explanation

28Roadmap and earned expansion

Stage 0: Assisted discovery. Manually deliver Profiles and Shape sessions · test the object model across three Template families · record what evidence is missing · measure whether management recognizes the diagnosis · execute small Moves manually. Gate: Customers select and act on recommended Moves.

Stage 1: Profile and Shape Scan. Source ingestion · Claim provenance · core Knowledge Graph · Profile generation and attestation · breakpoint diagnosis · three adjacent Shapes · scenario comparison. Gate: Profiles are accurate enough to earn connected data and payment.

Stage 2: Move and Ledger. Move contract · Decision Graph · approval workflow · bounded Tools · execution support · outcome reconciliation · local learning. Gate: One complete loop repeatedly produces measurable value.

Stage 3: Template system. Standard Template contract · three deep Template families · reusable onboarding and calculations · Template-specific Tool packs · cross-company evaluation. Gate: Deployment becomes repeatable without custom rebuilding.

Stage 4: Node and portfolio. Connected systems · company policy engine · multi-entity Profiles · Portfolio Index · capital-allocation views · private deployment options. Gate: Allometry becomes a recurring management and approval layer.

Stage 5: Capital and external Work Calls. Evidence-backed capital packages · monitoring interfaces · capability and capacity Profiles · structured Work Calls · external APIs and agent coordination. Gate: Internal Ledger evidence predicts external execution and repayment well enough to support third-party reliance.

Stage 6: Economic network. Permissioned market intelligence · supplier and partner discovery · agent-to-agent economic coordination · capacity reservation · attested settlement and financing integrations. Gate: A trusted two-sided network exists. This stage is not assumed.

29Current validation ledger

The v2.0 thesis must be tested against explicit questions.

29.1 Core hypotheses

1. Management teams recognize a source-backed Profile as a more accurate representation of their business than existing planning artifacts.
2. Breakpoint diagnosis creates more urgency than generic benchmarking.
3. Companies want to compare alternative Shapes, not merely optimize current operations.
4. A common object model can support multiple vertical Templates without losing relevance.
5. Customers will authorize Allometry to execute at least one bounded Move.
6. Expected-versus-realized reconciliation improves future recommendations.
7. A narrow Door can expand into the full Business Shape loop.

29.2 First validation program

Complete twenty assisted Shape engagements across at least three Template families.

Measure: Profile accuracy as rated by management · percentage of material claims supported by evidence · time required to produce a useful Profile · whether the breakpoint was recognized as material · whether adjacent Shapes felt specific and credible · percentage selecting a Move · percentage authorizing execution · percentage paying for connected analysis or execution · time to outcome evidence · reuse of the common schema and Tools across Templates · measurable economic or strategic outcomes.

29.3 Recommended initial thresholds

• At least 70% rate the Profile as materially accurate.
• At least 50% identify one insight they could not previously see clearly.
• At least 50% select a Move for deeper evaluation.
• At least 25% authorize execution or pay for continued work.
• At least one measurable outcome is produced in each active Template family.
• At least 70% of core objects remain shared across Templates.

These are operating targets, not external claims.

29.4 Falsifiers

Revise or narrow the thesis if:

• Profiles remain generic after access to reasonable evidence.
• Management enjoys Shape exploration but will not pay or act.
• The product cannot distinguish useful adjacency from generic ideation.
• Different verticals require fundamentally incompatible object models.
• Scenario outputs cannot become credible without consulting-scale manual work.
• Customers will share data but not delegate any Move.
• Moves cannot be measured within a useful decision window.
• The Ledger does not improve future prediction or decision quality.
• A narrow entry offer repeatedly fails to expand into Profile, Shape or execution.

29.5 Milestones and gates

Dates are commitments. Numbers are the evidence that opens the next gate.

29.6 Risks and what is done about them

30What not to do

• Do not position Allometry as an AI CEO.
• Do not lead with models, agents, MCP or ontology.
• Do not build a generic workflow builder.
• Do not rebuild CRM, ERP, accounting, email or document storage.
• Do not hardcode the platform around asset-heavy businesses.
• Do not claim horizontal usefulness without Template depth.
• Do not build ten shallow Templates at once.
• Do not generate Shapes without source evidence and explicit assumptions.
• Do not let an LLM establish accounting truth.
• Do not let agents change policy or authority silently.
• Do not expose private company data through the Index.
• Do not manufacture a marketplace before trusted supply and demand exist.
• Do not attach capital before the Move and repayment logic are credible.
• Do not control machines when an existing system should own motion and safety.
• Do not publish benchmarks before evidence is sufficiently comparable and attested.
• Do not present every long-horizon possibility in the current sales conversation.
• Do not let the product become a pageant of agents producing activity without economic purpose.
• Do not freeze the thesis against contrary evidence.

Internal doctrine: The company should not operate the software. The software should learn the company and prepare the work.

31Product constitution

1. Source before assertion. Material claims show where they came from.
2. Current reality before future possibility. A credible Shape begins with an accurate Profile.
3. Agnostic kernel, specific Templates. Horizontal architecture must produce contextual depth.
4. Ranges before false precision. Uncertainty is represented, not hidden.
5. One Move before a grand plan. Strategy earns commitment through evidence.
6. Policy before autonomy. Authority is explicit, bounded and reversible.
7. Tools before agent theatre. Every capability has a contract and evaluation.
8. Outcome before learning. Action without reconciliation does not improve the system.
9. Local truth before network prior. The company's evidence overrides generalized assumptions.
10. Zero duplicate entry. Read authorized records and ask only high-value questions.
11. Exportability and sovereignty. The company can inspect and export its Profile, decisions and Ledger.
12. The user controls consequential change. Allometry earns authority; it does not assume it.
13. Founder and management fit matter. The financially largest Shape is not automatically the best.
14. Capital serves the Move. Financing is attached to evidence-backed change.
15. Every Work Call returns evidence. Execution must close the loop.

32Canonical narratives

32.1 One sentence

Allometry helps companies see how they operate, decide what they should become and execute the next Move.

32.2 Customer narrative

Allometry builds a living Profile of your company from financial, operational, commercial and management evidence. It identifies what creates value, what is holding the business back and what will likely break next. It then explores credible products, services, channels, locations, acquisitions and operating configurations, compares their economics and helps execute the next approved Move. Every outcome updates the Profile, so the company learns from what actually happens.

32.3 Investor narrative

Allometry is the Business Shape intelligence layer for companies. The Pulse reads a company's systems continuously and resolves what is actually true. Its Knowledge Graph represents how a company works. Its Decision Graph models what the company could change and why. Templates provide contextual economic logic. Governed Tools execute approved Moves. The Ledger connects every hypothesis and commitment to its realized outcome. One Node learns each company; the Index compounds permissioned intelligence across companies and portfolios. Capital and Work Calls attach as the system earns authority.

32.4 Technical narrative

Allometry is a model- and system-agnostic economic intelligence architecture combining source-backed Knowledge Graphs, causal Decision Graphs, deterministic calculations, probabilistic models, policy-governed Tools and an append-only outcome Ledger. It exposes approved capabilities through product interfaces, APIs and MCP while preserving tenant sovereignty, provenance and human authority.

32.5 Allo narrative

Allo helps a founder understand the current Shape of their business, explore what it could become and take one smart Move at a time.

32.6 Homepage language

• Headline: See what your business could become.
• Supporting copy: Build a living Profile of how your company works, find what should change, explore credible future Shapes and put the next Move into action.
• Primary action: Build my Business Profile
• Alternative sharper headline: Find the highest-value Shape of your business.

32.7 The law in one line

Every business has a Shape. The Shape is its exponents. Allometry measures them, decides against them, and enforces the decision at the moment the company commits.

33Final master narrative

Every business has a Shape: the configuration of products, customers, channels, people, assets, systems, capital and operating constraints through which it creates value.

Most software helps a company operate inside its current Shape. Allometry helps the company understand and change the Shape itself.

Allometry begins by constructing a living, source-backed Profile. Its Knowledge Graph represents what the company is and how it works. Its diagnostic system identifies the primary constraint, next breakpoint and underused capabilities. Shape Templates combine universal business objects with contextual economic logic for services, products, software, hospitality, physical operations, real estate and other business models.

The system then explores credible future Shapes. It compares how changes to offers, SKUs, channels, customers, locations, capacity, ownership and capital could affect cash, risk, complexity and enterprise value. Management chooses a Move: a bounded, measurable change with a hypothesis, expected outcome, authority, budget and kill condition.

The Decision Graph preserves why the Move was chosen, which alternatives were rejected and what management expected. Policies govern what can happen. The Operator selects Tools to prepare or execute the work through humans, agents, systems, vendors or machines. A Move may create Commitments and Work Calls, each connected to the original strategic intent.

The Ledger reconciles expectation with reality. It records the evidence, decision, authorization, work, result, variance and resulting learning. That learning updates the Profile, local coefficients, Templates and next recommendation.

One sovereign Node understands each company. A permissioned Index improves priors across companies and portfolios without requiring raw data to become public. As the Ledger earns trust, capital can attach to evidence-backed Moves and Work Calls can coordinate external capacity. Those are consequences of the core loop, not prerequisites for it.

Allometry is therefore not another CRM, planning dashboard or agent builder. It is the system through which a company understands its present configuration, chooses a better one, executes the next change and learns from what actually happens.

See the business. Shape what comes next. Make the change.

34Revision ledger: v1.1 to v2.0

34.1 Changes promoted into core architecture

• Business Shape becomes the company identity rather than a late ontology addition.
• Profile becomes the governed present-state object.
• Knowledge Graph and Decision Graph become distinct but connected systems.
• Move becomes the atomic transformation object.
• Authorize and Reconcile become explicit steps in the operating loop.
• Tools replace generic agent language as the governed unit of capability.
• The Ledger becomes one unified event history with evidence, decision, commitment, execution, outcome and learning views.
• Templates become composable across business model, vertical, breakpoint and Move.
• Allo becomes the lighter product expression of the same architecture.

34.2 Ideas retained but repositioned

• Margin intelligence becomes one diagnostic and Tool family.
• Revenue Recovery becomes a high-value Door for relevant Templates.
• Commercial Commitment becomes an important transactional state, not the atomic company object.
• Node remains the sovereign company environment.
• Index remains the permissioned network intelligence layer.
• Capital attaches to evidence-backed Moves.
• Work Calls translate Moves into bounded execution.
• The Work API begins internally before expanding to external and physical coordination.
• Physical-capacity infrastructure remains a long-horizon consequence of the architecture.

34.3 Ideas deliberately demoted

• Asset-heavy operators no longer define the universal product architecture.
• G2 category strategy does not define company identity.
• Physical AI does not lead current product or website positioning.
• Separate branded ledgers are consolidated.
• Foundry is treated as a future use of Shape intelligence, not a current product pillar.
• Financing, marketplaces and forward capacity products remain gated extensions.

34.4 Propagation rule

v2.0 governs doctrine immediately. Public and investor surfaces flip together, once, at the v6 site gate (the pilot floor live with real, separated numbers). Until then: the site stays margin-first, because prospects mid-sequence must keep finding the story their email started; decks, memos, the letter and the six-pager stay frozen on v1.1 language, because investors mid-wave must not watch the identity change under them; the v2.0 narrative is used only in conversations that begin after this date. The rule is one flip, not a drift.

Ruling, Sept 2 2026 (founder): the flip is executed now rather than at the v6 gate. The open letter, the six-pager, the memos, the decks and the site footer speak v2.3 from this date. The homepage swap remains gated on the pilot floor's real, separated numbers. This entry governs propagation, not doctrine.

Ruling, Sept 5 2026 (founder): the floor goes live on the design partner on October 1 rather than October 1, and the thousand-operator target moves with it. The reason is the shadow run: a one-week run with zero silent misses needs a full week, and the September date left three days. The December gate is unchanged. These two entries are the only changes admitted to the frozen v2.3 text before the first reconciled outcome, because both govern schedule and propagation rather than doctrine.

34.5 Governing hypothesis for v2.0

A source-backed, Template-specific system can represent the current Shape of materially different businesses using one stable core ontology, recommend a credible next Move, help execute it and improve through reconciled outcomes.

The near-term proof is not the elegance of the ontology. It is whether companies across several Template families recognize themselves in the Profile, select a Move, act and produce measurable learning.

34.6 Revision ledger: v2.0 to v2.1 (Sept 2 2026)

Logged under §50.7. Six changes, each answering a specific weakness found in the v2.0 review:

1. The law restored (§3.5). v2.0 said "local coefficients" and never stated Y = a·M^b or the physics of profit. Without the law, "allometric" was a name and Shape distance was a checklist. Now Coefficient, SKU morphology, adjacent possible and Shape distance are defined by the exponents, and the floor and V* target give every Move its constraint and its goal.
2. Objective first-class (§4.2, §05). Goal was buried in the Profile and in Appendix A. Choosing the Objective is what makes Shape exploration specific. It is now a loop step and a canonical object.
3. Agent-to-agent commercialization named (§14.5). The Stripe analogy: one call that answers whether work is worth doing for this company, from business model, intent, capacity and policy. Internal first, partners second, open network last.
4. Two inlets to the Decision Graph (§11.4, §11.5, §21.1). Sovereign and market evidence ranked separately; Moves priced as changes in exponents toward V*.
5. The Oracle rule (§21.6). Conversational interface to Tools, never a model with opinions; token optimization belongs to the router.
6. Pricing hypothesis stated (§26.3). v2.0 packaging was entirely indicative. The physical-services Door now carries concrete numbers under test from October 1, measured at the December gate.

Nothing in the loop, the objects or the gates was removed. Actor was added as an object so that Work Calls and A2A have a named counterparty.

34.7 Revision ledger: v2.1 to v2.2 (Sept 2 2026)

Logged under §50.7. v2.1 was scored against four dimensions (go-to-market, investability, inspiration, readiness to become a PRD) and fell short of the standard on each for a nameable reason. v2.2 answers each:

1. Go-to-market. Doors had no names and distribution was implied. §25.5 names one Door per Template, §25.6 states the machine with its cadence, targets and prohibitions, §25.7 states why now.
2. Investability. Packaging was priced but the model was not stated. §26.4 gives contract value, the ladder to $3M, cost to serve, margin and the compounding mechanism. §29.5 puts dates on gates. §29.6 names risks and mitigations. §50.9 states use of capital as a Move.
3. Inspiration. The document opened with a status block. It now opens with why the company exists.
4. PRD readiness. The objects had definitions but no schema, the law had no calculation, Tools had a contract list but no template, and the build order lived in Appendix D as prose. Appendices F through I supply the Coefficient specification, the Tool contract template, the object schema and the Stage 0 to Stage 2 build order with acceptance tests.

No doctrine changed. Everything added is either a name, a number, a schema or a date, and each is falsifiable.

34.8 Revision ledger: v2.2 to v2.3 (Sept 2 2026)

Logged under §50.7. Four additions, no doctrine changed.

1. The Pulse named. v2.2 had the demon's fee but no name for the layer that pays it. The Pulse is now the continuous reading layer in §3.5, §13.5, §19.1, §23.1 and the investor narrative. Node, Index and Pulse are the three named platform parts.
2. The counted market and the diligence list (§26.5, §26.6). Investability was capped by a modeled market and a founder's word. The market is now counted by name and the evidence an investor should demand is listed.
3. The ninety-day calendar and the objection map (§25.8, §25.9). Go-to-market was complete as a machine but not as a schedule. It is now dated by week to the December gate, and the five conversation-ending objections have answers.
4. A fourth why-now (§25.7). Capital is funding systems of reality in revenue teams and SaaS; Allometry is the same thesis for the physical economy, with the law and enforcement at commitment.

Part II. Earned extensions and recovered strategic detail

Sections 01 through 34 define the integrated Business Shape system. Sections 35 through 49 recover the major product and infrastructure ideas from v1.1, now subordinated to that system. Section 50 enters the complete v2.0 thesis into its own Ledger.

35The living morphology interface

Business Shape should eventually feel like a living object, not a spreadsheet with decorative AI.

The visual representation is a navigational and explanatory layer over the Profile and Scenario model. It must never replace the underlying evidence.

35.1 Visual grammar

35.2 A financial model management can touch

The user can manipulate the Shape: add or remove a product lobe · change pricing · add a channel or location · merge an acquisition · add capital · remove a major customer · convert service revenue into recurring revenue · add production or delivery capacity · automate part of the operating model · franchise or license a capability · sell a division.

The system updates revenue, contribution, cash, capital requirement, risk, complexity, enterprise value, founder involvement and time to evidence.

35.3 Evidence before aesthetics

Build in this order: structured Profile → economic model → breakpoint diagnosis → adjacent Shape logic → scenario comparison → Move execution → outcome reconciliation → living morphology.

A beautiful visualization built on generic assumptions becomes a personality quiz. A living Shape grounded in company evidence can become the signature interface.

36The economic workspace

Allometry should create a useful workspace as a consequence of understanding the company. It should not build another manual software suite.

36.1 Workspace objects

People · companies · customers · opportunities · offers · products and SKUs · capabilities · Commitments · projects and Work Calls · invoices and payments · Sources and Claims · Shapes and Scenarios · Moves and Plans · approvals · outcomes.

36.2 Economic CRM

A conventional CRM asks: What stage is this opportunity in?

Allometry also asks:

• Is this the right customer for the Target Shape?
• What should the company sell?
• What will it cost to acquire and serve?
• What capacity will it consume?
• Does it improve cash, resilience or option value?
• Which policy governs the offer?
• What did similar Commitments produce previously?

For Allo and small companies without a CRM, the workspace may function as a lightweight economic CRM. For established companies, Allometry syncs with the incumbent record and adds Shape, decision and outcome intelligence.

36.3 Anti-suite doctrine

The business is not a collection of SaaS modules. The interface should emerge from the company's Profile, Templates and current Moves.

Commodity functions such as email, calendar, accounting, payments, storage and identity should be integrated or rented. Allometry owns the company model, decision lineage, transformation logic, authority and learning.

Software should adapt to the business. The business should not become administrative labor for the software.

37Reaction before planning

Planning sets the field. Events create the moments when action becomes valuable.

Allometry should operate as an economic action queue that responds to changes in demand, cost, capacity, cash and risk.

37.1 Trigger events

New lead, order, request or RFP · customer expansion or churn signal · supplier or input-cost change · capacity becoming constrained or available · work completed · missing or blocked invoice · late payment · contract renewal · margin variance · inventory accumulation · new location or asset opportunity · acquisition target appearing · financing need · policy exception.

37.2 Action card

Every important event becomes a concise management object:

• What changed?
• What does it mean economically?
• Which Shape or Move does it affect?
• What should happen next?
• What evidence supports the recommendation?
• What requires approval?
• What happened after action?

The initial product may be roughly 80% reaction and 20% periodic planning. The Target Shape keeps reaction from becoming random activity.

37.3 Interface placement

Action cards can appear in Allometry, Allo, Slack, Teams, email or an existing operating system. The control room holds the complete Profile, policy and Ledger. Daily work can occur where the team already operates.

38TAM Builder and the executable market

Market research becomes valuable when it produces an executable allocation of commercial effort.

The output is not a list of one thousand leads. It is: the finite set of accounts, segments, channels or acquisition targets most likely to advance the selected Shape, why now, what to offer and what the pursuit is expected to produce.

38.1 Market object

Each target can contain: Profile and footprint · current systems and business model · relevant needs or triggers · buying committee · existing relationship · recommended offer · expected contract or strategic value · acquisition and delivery cost · sales cycle and win probability · capacity and working-capital implications · opening angle · evidence and confidence · Target Shape alignment.

38.2 Market as part of the Decision Graph

The TAM Builder connects external market evidence to internal company economics. It can rank whether management should: pursue an account · enter a segment · launch a channel · develop a SKU · partner with an operator · acquire a company · ignore the opportunity.

38.3 Outcome accountability

Outreach, meetings, proposals, wins, losses, economics and follow-on demand return to the Ledger. Market selection improves through actual conversion and realized value, not engagement metrics alone.

39Sell completed economic work

Customers do not want agents. They want economically valuable work completed.

Purchasable outputs can include: verified company or account Profile · qualified opportunity · completed customer-research synthesis · approval-ready quote · recovered invoice · collected receivable · prevented supplier overpayment · new offer or SKU experiment · acquisition screen · financing package · profitable capacity filled · Shape scenario underwritten · commercial decision executed and reconciled.

39.1 Managed product before autonomous software

Allometry can begin as a managed outcome system. Humans and models complete work behind a standardized product surface. As evidence grows, repeatable steps become Tools and delegated workflows.

The customer receives one accountable result while the architecture progressively automates underneath.

39.2 Economic work credits

A future pricing mechanism may use prepaid credits for bounded economic work instead of charging only by seat. Credits should describe recognizable outcomes or Tool classes, not invisible token consumption.

Examples include: Profile verification · scenario construction · target research · quote preparation · RFP response · acquisition screen · outcome reconciliation.

Pricing increases as the output moves closer to realized economics. Attribution rules, quality standards, approvals and caps must be explicit.

39.3 The rule

Sell the result. Meter the work. Govern the action. Record the outcome.

40Revenue Recovery and the first economic Doors

Revenue Recovery remains a strong commercial Door for companies where contracts, operations and accounting do not reconcile cleanly.

The promise is: Find revenue already earned but not captured, then help convert it into invoices and cash.

40.1 Recoverable value

Completed but unbilled work · missing change orders · unapplied contractual escalators · usage above contracted limits · missing travel, equipment, fuel or rush charges · renewal prices that were not updated · duplicate supplier payments · unclaimed credits or rebates · receivables blocked by incomplete evidence.

40.2 Why it belongs in Business Shape

Revenue Recovery is not a separate company thesis. It creates four assets needed by the broader product:

1. A reconciled view of commercial and financial truth
2. A first economic model of the company
3. Immediate measurable value and trust
4. Evidence about recurring structural weaknesses

The expansion is: Recover value → Complete the Profile → Diagnose the Shape → Protect future economics → Choose the next Move

40.3 Other Template Doors

Revenue Recovery will not be the best Door for every Shape.

• Agency: engagement margin and founder dependence
• Food producer: true landed SKU contribution
• Hospitality: rate, occupancy and experience contribution
• B2B recurring: retention, packaging and service burden
• Asset operator: quote accuracy, capacity and job contribution
• Real estate: lease, utilization and highest-value-use scenarios

The Door is contextual. The destination remains the closed Business Shape loop.

41Foundry and the build, buy or transform decision

Once Allometry can represent a current company and compare future Shapes, it can ask a more fundamental allocation question: What business, product or capability should exist here, and what is the most efficient vehicle for creating it?

41.1 Available vehicles

Build organically · launch a new product or business line · spawn a separate company · acquire and scale · acquire and add AI · acquire and rebuild · acquire distribution or capability · partner or license · franchise · wait · ignore · sell, close or divest.

41.2 Foundry

Foundry is the future capital-and-effort allocation layer that compares these vehicles using: demand evidence · willingness to pay · competition and market structure · distribution advantage · unit economics · capability adjacency · build complexity · capital and working capital · time to evidence · integration risk · founder or operator fit · downside recovery · nonlinear upside.

Foundry is not a current product pillar. It is an earned application of Profiles, Shapes, Moves and outcomes.

41.3 Acquisition as a Shape Move

An acquisition is evaluated as a transformation of both buyer and target. The system models:

• Which capability, demand, distribution or cash flow is being acquired
• Standalone and combined Shapes
• Integration requirements
• Synergies and dis-synergies
• Management and founder dependence
• Capital structure
• Reversibility and downside protection
• Underwritten versus realized outcomes

This makes acquisition one vehicle among several rather than an isolated corporate-development workflow.

42The PE, HoldCo and portfolio operating system

Allometry can connect underwriting, ownership and realized value creation through one continuous record.

42.1 Before acquisition

Map and rank targets · build public and seller-supported Profiles · normalize profitability and cash conversion · identify operational breakpoints · compare acquire, build, partner and ignore · model standalone and combined Shapes · estimate integration and capital requirements.

42.2 During diligence

Validate claims against source evidence · reconstruct customer, product and unit economics · identify leakage and hidden obligations · measure concentration and owner dependence · assess systems, data and management readiness · create the initial Decision Graph and 100-day Moves.

42.3 During ownership

Deploy one Node per company · encode the investment thesis as Target Shapes and Moves · track approvals, capital and execution · compare underwritten and realized results · allocate resources across the portfolio · share capabilities without erasing local company truth.

42.4 At exit or recapitalization

Demonstrate decision and outcome history · show improvements in repeatability, recurrence and resilience · quantify the value bridge · separate market movement from operating intervention · provide a source-backed operating narrative.

The result is a living investment-thesis Ledger, not a static investment-committee memo.

43Capability, capacity and the Work Grade

The Knowledge Graph must represent what an actor can do and under which conditions.

43.1 Capability

A Capability describes the ability to perform a type of work. It can include: skill or service · equipment and tooling · certifications · geography and service radius · quality level · safety and insurance requirements · typical scope · historical performance · cost structure · dependencies.

43.2 Capacity

Capacity describes how much of that Capability is available over a time period, at a location and under defined constraints.

Capacity can belong to: a person or team · a production line · a room or property · inventory · a vehicle or machine · a vendor · a software or AI system · capital.

43.3 Work Grade

A Work Grade is a standardized, evidence-backed description of comparable capacity.

It can combine: capability · region · time window · volume · quality and acceptance rate · schedule reliability · insurance and certification · expected cost range · minimum commitment · verification standard · historical outcome confidence.

The Work Grade does not make all work identical. It makes relevant differences explicit enough for comparison, reservation, insurance, financing and agent purchasing.

43.4 Company value now, market infrastructure later

Inside one company, capability and capacity intelligence improves quoting, scheduling, hiring, outsourcing and capital allocation. Cross-company benchmarks and markets follow only when definitions and outcomes become sufficiently comparable.

44Physical capacity becoming callable

The long-horizon wave is not humanoids alone. It is physical capacity becoming increasingly discoverable, programmable and callable.

Companies already call APIs for compute, messages, payments and models. Future buyers and agents will increasingly request work from humans, vendors, conventional equipment, robots and mixed fleets through structured interfaces.

The scarce question becomes: Who or what should perform which work, for whom, at what price, under whose authority, using which capacity and with what proof of completion?

44.1 Allometry's role

• The equipment or execution system determines whether work can be performed safely.
• Allometry determines whether the work fits the company's economics, policy, priorities and Target Shape.
• The Work Call carries scope and authority.
• The Ledger carries proof and learning.

44.2 Start before autonomy

The same architecture begins with human teams, contractors, service vendors, assets, warehouses and conventional automation. A robot later becomes another Actor and capacity type inside the existing graph.

Allometry does not compete in motion control, machine safety or fleet communication. It owns economic context, selection, authorization and outcome attestation where customers grant that role.

44.3 Horizon discipline

No physical-market claim is required to validate the current Business Shape product. The long-horizon thesis is credible only if the internal loop first proves useful with existing human and software execution.

45External Work Calls and agentic commerce

The internal Work Call becomes external only after company Profiles, Tools, policies and outcome evidence are reliable.

45.1 Buyer request

An authorized buyer or agent may request: Find qualified providers capable of completing this defined work within the required geography, time, quality, capacity and economic constraints.

45.2 Supplier response

Supplier Nodes can evaluate: capability fit · available capacity · expected cost and margin · customer and payment risk · opportunity cost · strategic value · delivery confidence · required approval.

The Node can respond with a governed quote, request more information, decline or escalate.

45.3 Network sequence

Discover → Qualify → Evaluate → Quote → Authorize → Reserve → Execute → Attest → Settle → Learn

45.4 Economic API

The economic API normalizes meaning and governed actions rather than competing with generic connector infrastructure.

Illustrative capabilities include: get_company_profile · explain_breakpoint · simulate_shape · compare_shapes · propose_move · check_policy · check_capacity · evaluate_work · request_approval · record_commitment · issue_work_call · attest_outcome · reconcile_move.

The interface can be exposed through API, MCP or embedded applications. The Node remains the trusted economic context and authority boundary.

46Autonomous negotiation, settlement and attestation

Negotiation and settlement become possible only inside explicit economic envelopes.

46.1 Negotiation envelope

A company can define negotiable variables such as: price · timing · quantity · service level · payment terms · sequence · capacity reservation · cancellation · financing · insurance.

The envelope can constrain: minimum margin or return · maximum discount · customer risk · capacity exposure · working-capital exposure · required evidence · approval threshold · strategic-account exception.

Models may negotiate inside the envelope. They cannot redefine it.

46.2 Settlement

Allometry does not need to move money. Existing payment infrastructure can manage credentials, fraud, processing and settlement.

Allometry governs: what was purchased · whether the provider and capability were qualified · whether price and risk fit policy · whether capacity was committed · whether work was accepted · which evidence releases payment · how the result changes future decisions.

46.3 Attestation

Completion evidence may include: customer acceptance · time and location · images or documents · materials and energy consumed · system events · sensor or machine telemetry · quality results · human approval · invoice and payment status.

Attestation should be proportionate to consequence. The system does not manufacture elaborate proof for low-risk work.

47Ledger-powered financing

Traditional financial statements describe historical company performance. The Allometry Ledger can add forward-looking evidence about specific Moves and Commitments.

47.1 Evidence unavailable from accounting alone

Quality of contracted backlog · capacity available to deliver it · estimate-to-outcome accuracy · margin and cash conversion by work type · customer payment behavior · execution reliability · change-order behavior · utilization and bottlenecks · management approval history · actual results of previous capital allocation.

47.2 Finance the Move or Commitment

The financing unit can become more granular over time: Company → Business unit → Shape Move → Contract → Job → Asset → Work Call

Potential products include: contract or purchase-order financing · receivables financing · inventory and working capital · equipment leasing · capacity-expansion financing · acquisition financing · revenue-based finance · performance insurance.

47.3 Capital at the point of decision

When a credible Move is constrained by capital, the Node can prepare the evidence package and request terms from approved providers. Financing remains conditional on company policy and human authority.

47.4 Earn the rail

Relationships and assisted underwriting come before automated capital flows. Allometry should earn credibility as an intelligence, verification and monitoring layer before third parties rely on its data for autonomous decisions.

Finance the work and transformation, not merely the historical company.

48Open, free and private

Open, free and private are different promises.

48.1 Free

A low-friction hosted experience can include: initial Profile · Shape Scan · one active Shape · one Move or experiment · basic workspace · public-data enrichment · exportable company data.

Free distribution should create useful Profiles and qualified demand, not subsidize expensive unrestricted agent work.

48.2 Open

Allometry can open the interfaces that increase trust and ecosystem adoption: core Profile interchange schema · import and export formats · Tool and Work Call specifications · MCP and API interface · connector SDK · evaluation conventions.

Potentially proprietary compounding assets include: Template calibration · breakpoint models · scenario ranking · valuation adjustments · benchmark data · decision policies · cross-company learning · execution orchestration.

48.3 Private

Private deployments can provide: customer-controlled keys · dedicated model and data infrastructure · VPC or on-premise deployment · private Template extensions · local Tool execution · sovereign Ledger and Graphs · permissioned or no Index contribution.

The company should retain the right to inspect, export and govern its data regardless of deployment model.

49The benchmark and market-formation path

The Index becomes defensible through depth, verification and outcomes, not directory size.

49.1 Quality-gated expansion

An illustrative sequence is: Map broadly → Structure selectively → Verify deeply → Connect privately → Attest outcomes → Aggregate carefully

Public Profiles can create discovery. Private Nodes create economic truth. Ledger outcomes create benchmark credibility.

49.2 Legibility ladder

Legible → Comparable → Callable → Financeable → Tradable

Nothing becomes reliably callable or financeable until its identity, quality, availability, price and delivery can be understood.

49.3 Market-formation sequence

1. Measure: Normalize units, conditions, cost, price, capacity and outcomes.
2. Index: Create permissioned ranges and comparable performance classes.
3. Contract: Standardize bilateral Work Calls and reservations.
4. Finance and insure: Attach capital and risk products to evidence-backed Commitments.
5. Coordinate: Match demand and capacity under policy.
6. Market: Consider broader liquidity only if natural participants and regulation support it.

49.4 Benchmark restraint

Do not publish a benchmark until the underlying units are sufficiently comparable, the sample is useful, permission is clear and outcomes are attested.

The initial customer value is internal: better pricing · better capacity allocation · better planning · better underwriting · better supplier and partner selection.

External market infrastructure is an earned consequence.

50The thesis entered into its own Ledger

A company that sells evidence-backed Moves should govern its own strategy the same way.

50.1 Profile

Allometry is a solo-founder, pre-revenue company. It starts over from that position deliberately, with three things most pre-revenue companies lack: real operating data (a design partner's closed-job history and daily quoting, connected read-only), a pricing hypothesis (the margin floor) entering live test on October 1, 2026, and a thesis spanning company intelligence, strategy, execution and outcomes. Data access is not revenue. Nothing in this document counts as revenue until an invoice is paid.

The strongest current assets are:

• A distinctive Business Shape framework
• Experience diagnosing commercial and operating systems
• Access to companies across physical services, recurring B2B and product businesses
• An initial decision and outcome ontology
• A credible path from managed work to software
• A long-horizon view that does not need to be built immediately

The principal constraints are:

• Limited founder time and engineering capacity
• The risk of expanding architecture faster than customer evidence
• The amount and cleanliness of company data
• The need to prove that Shape exploration produces action and payment
• The need to choose a narrow initial Door without shrinking the company identity

50.2 Diagnosis

The primary breakpoint is moving from intellectually coherent thesis to repeatable customer outcome.

The system does not yet need more surface area. It needs proof that one company can move through the complete loop: Profile → Objective → Diagnose → Explore Shapes → Choose a Move → Authorize → Execute → Reconcile → Learn.

50.3 Target Shape

The near-term Target Shape is a productized, assisted Business Shape system that:

• Produces a credible source-backed Profile quickly
• Diagnoses a material breakpoint
• Generates specific adjacent Shapes
• Helps management select one Move
• Executes or prepares that Move
• Measures the outcome
• Reuses its core ontology across several Templates

50.4 Chosen Move

Run twenty assisted Shape engagements across at least three Template families. Build only the reusable product components required to complete those loops.

The engagements already in motion are the first of the twenty, not a parallel Move. The design-partner floor pilot and the partner-run operator conversations are the physical-services Template's Stage 0 engagements. The wedge-1000 outbound is the demand test for that Template's Door. None of these adds a second chosen Move; they are how the first Template family enters the count. The second and third Template families are opened only as the first produces reconciled outcomes.

For every engagement, record: Profile accuracy · missing evidence · breakpoint relevance · Shape selected · Move selected · willingness to pay · authority delegated · time to outcome · realized result · shared versus Template-specific architecture.

50.5 Expected outcome

Within the first validation cycle:

• Management recognizes the Profile as materially accurate.
• At least half of participants identify a valuable insight and select a Move.
• A meaningful subset pays for continued analysis or execution.
• At least one outcome is reconciled in each active Template family.
• Most core objects and Tool contracts remain reusable across Templates.

50.6 Falsifiers

The thesis must narrow or change if:

• Profiles remain generic after reasonable evidence access.
• Management likes exploring Shapes but does not act.
• Templates repeatedly require incompatible core architectures.
• Scenario quality requires consulting-scale labor indefinitely.
• Customers refuse to connect data or authorize bounded execution.
• Outcomes cannot be observed in a useful time frame.
• The Ledger fails to improve future recommendations.
• A concrete Door does not expand into the broader loop.

50.7 Revision policy

New ideas enter a backlog before entering canon. They graduate only when they:

• Change the product architecture
• Improve present customer acquisition or value delivery
• Reveal a compounding asset
• Carry credible customer or outcome evidence

The thesis is reviewed against its own Ledger. Vocabulary can remain stable while strategy changes. No narrative is protected from contrary evidence.

50.8 v2.0 commitment

The company will build the closed loop before building the network.

It will prove the Move before the Work API, outcomes before benchmarks, internal authority before external coordination and evidence before capital automation.

Keep the kernel stable. Let Templates become specific. Execute one Move. Reconcile the outcome. Let evidence earn everything else.

50.9 Use of capital

Capital is a resource required by this Move, and it is stated the way §18 requires every Move to state it.

• What it funds, in order. One full-stack engineer so the founder leaves the implementation loop; the Coefficient engine and Tool contracts in Appendices F and G as production code; depth in the first Template until deployment compresses; then the second Template family, chosen by evidence from the Labs and the partner channel.
• What it does not fund. Paid acquisition before the funnel is instrumented. A second product surface. Any Template without a Stage 0 engagement behind it. Benchmarks. Balance-sheet lending.
• Time to evidence. The December gate is the first reconciliation of this Move. Three to five paying design partners, with estimated-versus-realized margin reported separately, is the expected outcome. Fewer than three is the kill condition for the current Door, not for the thesis.
• Reversibility. Everything built for the first Template is reusable by the second, because the objects are shared. If the first Door fails, the loss is the Door, not the kernel.

Appendix A. Canonical relationship model

The following relationships form the minimum universal graph:

• Company has Profile
• Profile represents Shape
• Profile contains evidence about Entity
• Claim describes Entity or Relationship
• Claim originates from Source
• Company creates value through Economic Unit
• Shape is constrained by Constraint
• Shape approaches Breakpoint
• Template applies to Company, Shape, Breakpoint or Move
• Scenario models Shape
• Objective ranks Scenario
• Objective directs Move
• Coefficient describes Economic Unit
• Move changes Coefficient
• Actor answers Work Call
• Hypothesis supports Move
• Move transforms Current Shape toward Target Shape
• Move depends on Capability, Capital, Tool or another Move
• Decision selects or rejects Move
• Policy governs Decision, Tool, Commitment or Work Call
• Approval authorizes Decision or action
• Move creates Commitment or Work Call
• Tool executes or prepares Work Call
• Work Call is performed by Actor
• Execution produces Outcome
• Outcome is compared with Expected Outcome
• Variance updates Claim, coefficient, Template or policy proposal
• Ledger Entry binds the complete lineage

Appendix B. Shape and Move scoring

Allometry may rank Shapes and Moves using a transparent, configurable score rather than a universal hidden formula.

Shape dimensions: expected cash generation · enterprise-value potential · resilience and recurrence · capital requirement · time to evidence · time to scale · operational complexity · founder or management fit · strategic adjacency · risk · reversibility · option value · evidence confidence.

Move dimensions: expected economic value · information gain · urgency · cost and capital · required capacity · time to outcome · reversibility · dependency risk · policy fit · Target Shape alignment · confidence.

The score explains the ranking. Management can change objectives and weights. The system never hides a value judgment inside a model output.

Appendix C. Work Call schema

A minimum Work Call includes:

work_call:
  id: string
  parent_move_id: string
  purpose: string
  requested_by: actor_reference
  actor_type: human | team | agent | system | vendor | machine
  required_capability: capability_reference
  scope: object
  inputs: source_references
  acceptance_criteria: list
  economic_envelope:
    budget: range
    expected_value: range
    margin_or_return_floor: optional
  capacity:
    quantity: optional
    location: optional
    start_at: optional
    complete_by: optional
  policy_refs: list
  approval:
    required: boolean
    authority: actor_reference
  risk_and_safety: object
  compensation_or_settlement: object
  escalation: object
  rollback: object
  outcome_evidence_required: list
  ledger_entry_id: string

The schema may become more specialized through Templates. The parent Move, authority and outcome evidence remain mandatory for consequential work.

Appendix D. Horizon discipline

Build now: source ingestion and Claim provenance · core Knowledge Graph · company Profile · Shape and Breakpoint diagnosis · three adjacent Shapes · scenario assumptions and ranges · one Move contract · human approval · bounded execution Tool · outcome reconciliation · unified Ledger.

Build after repeated customer evidence: deep Template system · connected operational sources · automated Profile updates · multi-Move plans · more execution Tools · company policy engine · portfolio Nodes and Index · capital package preparation · internal Work Calls.

Build only after authority is earned: external agent-to-agent coordination · supplier and capability discovery network · capital matching · external Work API · capacity reservation · automated settlement triggers · physical-machine coordination · published benchmarks.

Preserve as long-horizon optionality: broad physical-capacity network · standardized Work Grades · machine commerce · granular job or asset financing · forward capacity commitments · market infrastructure · Foundry and autonomous company formation.

The discipline is simple: Profile before prediction. Move before platform. Outcome before network. Evidence before capital. Internal Work Call before external market.

Appendix E. Complete v1.1 to v2.0 migration map

Every numbered section from v1.1 remains traceable. "Reframed" means the underlying insight remains but no longer controls the company identity. "Gated" means it survives as an earned extension rather than a present build commitment.

Appendix F. Coefficient specification

The deterministic calculation behind §3.5. A Coefficient is the pair (a, b) for one Economic Unit and one scaled quantity (cost, capacity consumed, cash days, complexity proxy) as a function of volume M.

Inputs. Executed units only: closed jobs, shipped orders, delivered engagements, occupied room-nights. Each with realized volume, realized cost at cost-as-of-commitment c(t), realized revenue, capacity consumed, cash timing, and the Source and Claim ids behind each figure. Quotes and estimates are never inputs to a Coefficient; they are what the Coefficient is checked against.

Method.

1. Group executed units by Economic Unit and by the Template's declared segmentation (customer, product family, location) where the segment has enough sample. Segmentation pools upward until the minimum is met: SKU to product family to job family to unit type. A floor is enforced at the coarsest level that clears the minimum and reported at the finest level that exists.
2. Fit log Y = log a + b · log M by ordinary least squares over a trailing window the Template declares (default: twelve months or two hundred units, whichever is larger).
3. Report b with its standard error and the fit's residual spread. Report a in the unit's native currency.
4. Combine with the Template prior by inverse-variance weighting: the prior dominates below the minimum sample and vanishes as local evidence accumulates.
5. Flag regime changes: if the trailing quarter's fit differs from the window's fit by more than the Template's tolerance, mark the Coefficient as shifting and surface the breakpoint candidate.

Minimums. No Coefficient is enforceable below thirty executed units or a standard error on b above 0.15. Below that it is descriptive only and the floor runs in recommend mode for that unit.

Outputs. (a, b, standard error, sample size, window, freshness, prior weight, regime flag, source ids). Every Coefficient is a Claim with provenance and is versioned on every recalculation.

Uses. The floor computes c(t) for a new unit from the Coefficient and current input prices. Shape exploration computes the effect of adding or removing units on the company's aggregate exponents. Move scoring predicts the change in exponents; reconciliation re-measures them.

Tests. Synthetic companies with known exponents must be recovered within the standard error. The design partner's ninety days must reproduce the surfaced leakage patterns from the Coefficients alone. A Coefficient recalculation may never silently change an enforced floor; the change is a proposed update under §15.4.

Appendix G. Tool contract template

Every production Tool is declared in this form before it ships. The contract is a file in the repository and a row in the Tool registry.

tool:
  name: string                     # verb_object, e.g. check_floor
  family: understand | diagnose | explore | decide | execute | learn
  purpose: string                  # one sentence
  template_scope: [template_ids]   # or "kernel"
  inputs:
    schema: json_schema_ref
    required_sources: [source_types]
    minimum_confidence: 0.0-1.0
  outputs:
    schema: json_schema_ref
    claims_produced: [claim_types]
  computation:
    deterministic: [calculation_ids]   # never model-derived
    models_permitted: [model_classes]  # by class, never by vendor
    max_cost_per_run: currency
  authority:
    permission_scope: string
    approval_required: none | policy | human
    side_effects: [none | writes_record | sends_message | commits_money | commits_capacity]
    reversible: boolean
    rollback: string
  reliability:
    idempotent: boolean
    on_failure: retry | escalate | abort
    escalation_to: actor_reference
  ledger:
    event_type: string
    fields_recorded: [fields]
  evaluation:
    suite: path
    thresholds: {accuracy: 0.0-1.0, provenance: 0.0-1.0, policy_compliance: 1.0}
  owner: actor_reference
  version: semver

A Tool with side effects other than none cannot ship with approval_required set to none. A Tool that produces Claims must name the calculation or model class that produced each.

Appendix H. Object schema

The canonical objects of §05 as a minimum relational schema. Field lists are the minimum; Templates extend, never redefine.

Every table carries workspace_id, created_at and updated_at. Ledger Entry is append-only; corrections are new entries. Tenant isolation is at the workspace, and every export is complete.

Appendix I. Build order and acceptance tests, Stage 0 to Stage 2

Stated against the first Template. Each stage has one acceptance test and cannot be skipped.

Stage 0. Assisted discovery (now to October 1). Build: Source ingestion for the design partner's systems; Claim store with provenance; Coefficient engine per Appendix F over executed work, with Template segmentation pooled by job family so that most units clear the thirty-unit minimum; the floor as a Tool under Appendix G in recommend mode; one Ledger event type; the shadow-run log as a Ledger view and the funnel ladder as a counted view, both of which §26.6 promises to investors. Accept when: the Coefficients reproduce the surfaced leakage patterns from the ninety-day scan; the floor flags every below-floor quote in a one-week shadow run with zero silent misses.

Stage 1. Profile and Shape Scan (October 1 to December). Build: Profile generation and attestation; breakpoint diagnosis from the Coefficient regime flags; three adjacent Shapes computed from the Coefficient set; scenario comparison with ranges; Shape Scan intake for the Door. Accept when: the design partner attests the Profile as materially accurate; three of the ten cohort operators identify one insight they could not previously see; the floor runs live on daily quoting with a human approval path.

Stage 2. Move and Ledger (December to Q1 2027). Build: Move contract and Decision record with rejected alternatives; approval workflow bound to Policy; two more execute-family Tools; outcome reconciliation with variance classes; proposed-update flow under §15.4. Accept when: one Move per paying design partner has closed its measurement window and been reconciled; at least one Coefficient update has been proposed, back-tested and promoted through the governed flow; the second operator deployed in less time than the first, measured.

Nothing in Stage 3 and beyond is started until Stage 2 accepts.

v2.3 lock

• The law is: Y = a·M^b. Every business has a Shape; the Shape is its exponents. The floor is p ≥ (1 + f) × c(t) at the moment of commitment; the target is V*.
• The stable core is: Profile → Objective → Diagnose → Explore Shapes → Choose a Move → Authorize → Execute → Reconcile → Learn → Re-profile
• The architecture is: Knowledge Graph → Profile and Templates → Decision Graph → Policies and Tools → Ledger → Learning
• The platform structure is: One sovereign Node per company, read continuously by the Pulse. One permissioned Index across companies and portfolios.
• The execution structure is: A Move creates governed Commitments and Work Calls. Tools perform the work. Outcomes return to the Ledger.
• The commercialization structure is: One call answers whether work is worth doing for this company, at what price, and whether it can deliver. Internal first, partners second, open network last.
• The capital structure is: Capital attaches to evidence-backed Moves and is measured against realized outcomes.
• The company promise is: Allometry helps a company see how it works, decide what it should become and make the change.
• The standing instruction is: Keep the kernel stable. Let Templates become specific. Execute one Move. Reconcile the outcome. Let evidence earn everything else.

— Taylor Gendron, Founder · Montréal | New York · taylor@allometry.com