> ## Documentation Index > Fetch the complete documentation index at: https://docs.uselayerup.com/llms.txt > Use this file to discover all available pages before exploring further. # The Insurance Ontology > The canonical, version-controlled object model. 28 objects across seven groups: Parties, Master data, Underwriting, Claims, Workflow, Evidence, and Governance. # 05 — The insurance ontology. The Ontology is the canonical, version-controlled object model that every plane projects onto. It defines 28 objects across seven groups: Parties, Master data, Underwriting, Claims, Workflow, Evidence, and Governance. Every Property, Decision, Action and AuditEvent the platform produces is typed against the Ontology. ## 5.1 Object index | # | Group | Object | Anchor | Purpose | | -- | ------------ | -------------------- | ------------- | ---------------------------------------------- | | 1 | Parties | **Insured** | o-insured | Named insured / policyholder party | | 2 | Parties | **Broker** | o-broker | Distribution intermediary party | | 3 | Parties | **Claimant** | o-claimant | Party making a claim against a policy | | 4 | Parties | **Underwriter** | o-underwriter | Carrier-side decision principal | | 5 | Parties | **Adjuster** | o-adjuster | Carrier-side claims handler | | 6 | Master data | **Policy** | o-policy | Insurance contract | | 7 | Master data | **Coverage** | o-coverage | Typed risk-transfer scope on a Policy | | 8 | Master data | **Endorsement** | o-endorsement | Mid-term contract amendment | | 9 | Underwriting | **Submission** | o-submission | Inbound risk for consideration | | 10 | Underwriting | **Quote** | o-quote | Carrier-side priced response | | 11 | Underwriting | **RiskAssessment** | o-risk | Structured risk evaluation against an exposure | | 12 | Underwriting | **Pricing** | o-pricing | Premium calculation result | | 13 | Underwriting | **BindingAuthority** | o-binding | Delegated authority record | | 14 | Claims | **Claim** | o-claim | Reported loss case | | 15 | Claims | **LossEvent** | o-loss | Real-world loss occurrence | | 16 | Claims | **Exposure** | o-exposure | Per-coverage manifestation of a loss | | 17 | Claims | **Reserve** | o-reserve | Held estimate against a loss | | 18 | Claims | **Payment** | o-payment | Outflow against a claim | | 19 | Workflow | **Task** | o-task | Unit of operator work | | 20 | Workflow | **Exception** | o-exception | Typed deviation requiring resolution | | 21 | Evidence | **Document** | o-document | Content-addressed file | | 22 | Evidence | **EmailThread** | o-email | Inbound / outbound conversation | | 23 | Evidence | **Attachment** | o-attachment | Document bound to a thread | | 24 | Evidence | **EvidenceSpan** | o-evspan | Typed citation back to source bytes | | 25 | Governance | **Decision** | o-decision | Typed verdict, agent or human | | 26 | Governance | **Action** | o-action | Typed effect intent on a SoR | | 27 | Governance | **AgentRun** | o-agentrun | Bounded execution of an agent | | 28 | Governance | **AuditEvent** | o-audit | Hash-chained audit record | ## 5.2 Object cards Each object card defines: a one-line description, the field set (a subset; full schema in the Accordion), key links, default permissions, and the JSON Schema. Every field is implicitly accompanied by a provenance record. ### Group A · Parties **ontology · party** | marking: `pii.medium` | scope: `tenant` The named insured / policyholder party on a Policy. May be an individual or organisation. Always content-addressed against an external party-master identifier where one exists. | Field | Type | Req | Notes | | --------------- | ----------------------- | --- | ------------------------ | | `insuredId` | id | req | Internal canonical id | | `kind` | enum: `person` \| `org` | req | | | `legalName` | string | req | | | `externalIds` | map\ | opt | e.g. tax id, registry id | | `jurisdictions` | iso3166\[] | opt | One or more countries | | `contact` | ContactRef | opt | Email, phone | **Links**: `policiesOf → Policy[]`, `claimantOf → Claimant[]`. **Permissions**: read — `insured.read`; mutate — `insured.write`; PII fields require purpose `underwriting.review` or `claims.adjustment`. ```json theme={null} { "$id": "layerup://ontology/v1/Insured", "type": "object", "required": ["insuredId", "kind", "legalName"], "properties": { "insuredId": { "type": "string", "pattern": "^ins_[A-Z0-9]{10,}$" }, "kind": { "enum": ["person", "org"] }, "legalName": { "type": "string", "minLength": 1 }, "externalIds": { "type": "object", "additionalProperties": { "type": "string" } }, "jurisdictions": { "type": "array", "items": { "type": "string", "pattern": "^[A-Z]{3}$" } }, "contact": { "$ref": "layerup://ontology/v1/ContactRef" } } } ``` **ontology · party** | marking: `tenant` The distribution intermediary that placed a Submission or that is bound to a Policy. Typically also has BindingAuthority records on certain LOBs. | Field | Type | Req | Notes | | ---------------- | --------------------------- | --- | ----- | | `brokerId` | id | req | | | `name` | string | req | | | `licenseIds` | map\ | opt | | | `commissionTier` | string | opt | | | `contact` | ContactRef | opt | | **Links**: `submissionsOf → Submission[]`, `policiesOf → Policy[]`, `bindingAuthorityOf → BindingAuthority[]`. ```json theme={null} { "$id": "layerup://ontology/v1/Broker", "type": "object", "required": ["brokerId", "name"], "properties": { "brokerId": { "type": "string", "pattern": "^brk_[A-Z0-9]{10,}$" }, "name": { "type": "string" }, "licenseIds": { "type": "object", "additionalProperties": { "type": "string" } }, "commissionTier": { "type": "string" }, "contact": { "$ref": "layerup://ontology/v1/ContactRef" } } } ``` **ontology · party** | marking: `pii.medium` A party making a claim against a Policy. Distinct from the Insured: a third-party claimant on a liability coverage is not the policyholder. | Field | Type | Req | | ------------------- | ---------------------------------------------- | --- | | `claimantId` | id | req | | `kind` | enum | req | | `legalName` | string | req | | `relationToInsured` | enum: `self` \| `third_party` \| `beneficiary` | req | | `contact` | ContactRef | opt | **Links**: `claimsOf → Claim[]`. **ontology · party · internal** A carrier-side decision principal authorised to bind, decline or refer Submissions within their scope. Has typed authority limits (line, jurisdiction, premium ceiling). | Field | Type | Req | | ---------------- | ---------------- | --- | | `underwriterId` | id | req | | `principalRef` | PrincipalRef | req | | `authority` | UWAuthorityScope | req | | `lineOfBusiness` | string\[] | req | **Links**: `quotesAuthored → Quote[]`, `decisionsAuthored → Decision[]`. **ontology · party · internal** A carrier-side claims handler authorised to assess, reserve, settle or refer Claims within their scope. Authority is typed by line, severity band, and jurisdiction. | Field | Type | Req | | -------------- | -------------------- | --- | | `adjusterId` | id | req | | `principalRef` | PrincipalRef | req | | `authority` | ClaimsAuthorityScope | req | | `specialism` | string\[] | opt | **Links**: `claimsAssigned → Claim[]`, `decisionsAuthored → Decision[]`. ### Group B · Master data **ontology · master** | soR: `policy admin` An insurance contract. The canonical record lives in the carrier's policy administration system; Layerup's Policy is a projection augmented with derived properties. | Field | Type | Req | | ---------------- | ----------------- | --- | | `policyId` | id | req | | `policyNumber` | string | req | | `insuredRef` | Ref\ | req | | `brokerRef` | Ref\ | opt | | `productCode` | string | req | | `lineOfBusiness` | string | req | | `currency` | iso4217 | req | | `effective` | date | req | | `expiry` | date | req | | `status` | enum | req | | `coverageRefs` | Ref\\[] | req | **Links**: `coverages → Coverage[]`, `endorsements → Endorsement[]`, `claims → Claim[]`. ```json theme={null} { "$id": "layerup://ontology/v1/Policy", "type": "object", "required": ["policyId","policyNumber","insuredRef","productCode","lineOfBusiness","currency","effective","expiry","status","coverageRefs"], "properties": { "policyId": { "type": "string" }, "policyNumber": { "type": "string" }, "insuredRef": { "$ref": "layerup://ontology/v1/Ref" }, "brokerRef": { "$ref": "layerup://ontology/v1/Ref" }, "productCode": { "type": "string" }, "lineOfBusiness": { "type": "string" }, "currency": { "type": "string", "pattern": "^[A-Z]{3}$" }, "effective": { "type": "string", "format": "date" }, "expiry": { "type": "string", "format": "date" }, "status": { "enum": ["quoted","bound","in_force","cancelled","expired"] }, "coverageRefs": { "type": "array", "items": { "$ref": "layerup://ontology/v1/Ref" } } } } ``` **ontology · master** A typed risk-transfer scope on a Policy. Each Coverage has its own limits, deductibles, perils, and exclusions. | Field | Type | Req | | ------------ | ------------ | --- | | `coverageId` | id | req | | `policyRef` | Ref\ | req | | `code` | string | req | | `limit` | Money | req | | `deductible` | Money | opt | | `perils` | string\[] | opt | | `exclusions` | string\[] | opt | **Links**: `policy → Policy`, `exposures → Exposure[]`. **ontology · master** A mid-term contract amendment to a Policy or Coverage. Has its own effective date, change set, and approval lineage. | Field | Type | Req | | --------------- | -------------- | --- | | `endorsementId` | id | req | | `policyRef` | Ref\ | req | | `changeSet` | JSONPatch | req | | `effective` | date | req | | `approvalRef` | Ref\ | opt | **Links**: `policy → Policy`, `approvedBy → Decision`. ### Group C · Underwriting **ontology · uw** Inbound risk presented by a Broker (or directly by a prospective Insured) for the carrier's consideration. Drives the underwriting workflow until quoted, declined, or referred. | Field | Type | Req | | -------------------- | ---------------------------------------------------------------- | --- | | `submissionId` | id | req | | `brokerRef` | Ref\ | opt | | `insuredRef` | Ref\ | opt | | `productCode` | string | req | | `lineOfBusiness` | string | req | | `requestedEffective` | date | req | | `exposureSummary` | ExposureSummary | opt | | `status` | enum: `open` \| `quoted` \| `declined` \| `referred` \| `closed` | req | **Links**: `quotes → Quote[]`, `riskAssessment → RiskAssessment`, `documents → Document[]`. **ontology · uw** A carrier-side priced response to a Submission. Bound by the issuing underwriter's authority. Either binds (becoming a Policy) or expires. | Field | Type | Req | | ---------------- | ---------------------------------------------------------------- | --- | | `quoteId` | id | req | | `submissionRef` | Ref\ | req | | `underwriterRef` | Ref\ | req | | `premium` | Money | req | | `terms` | QuoteTerms | req | | `validUntil` | datetime | req | | `status` | enum: `draft` \| `issued` \| `bound` \| `expired` \| `withdrawn` | req | **Links**: `submission → Submission`, `boundPolicy → Policy?`, `pricing → Pricing`. **ontology · uw** A structured evaluation of an exposure: hazards, prior loss history, geography, regulatory class, model-derived scores, agent-derived recommendations, and underwriter sign-off where required. | Field | Type | Req | | --------------- | -------------------------------------- | --- | | `assessmentId` | id | req | | `subjectRef` | Ref\ | req | | `hazardScores` | map\ | opt | | `verdict` | enum: `accept` \| `refer` \| `decline` | req | | `rationale` | string | req | | `evidenceSpans` | Ref\\[] | req | **Links**: `subject → Submission|Endorsement`, `decision → Decision`, `evidence → EvidenceSpan[]`. **ontology · uw** The premium calculation result for a Quote. Captures rating factors, base rates, modifiers, surcharges, taxes, and the model lineage of any non-deterministic rating step. | Field | Type | Req | | -------------- | -------------------------- | --- | | `pricingId` | id | req | | `quoteRef` | Ref\ | req | | `baseRate` | Money | req | | `factors` | map\ | opt | | `surcharges` | map\ | opt | | `taxes` | map\ | opt | | `finalPremium` | Money | req | | `modelLineage` | ModelLineage\[] | opt | **ontology · uw** A delegated authority record granting a Broker (or carrier office) the right to bind risks within a defined scope on the carrier's behalf. Has limits by line, jurisdiction, premium, and aggregate. | Field | Type | Req | | ------------------ | ------------ | --- | | `bindingId` | id | req | | `holderRef` | Ref\ | req | | `scope` | BindingScope | req | | `premiumCeiling` | Money | req | | `aggregateCeiling` | Money | opt | | `effective` | date | req | | `expiry` | date | req | ### Group D · Claims **ontology · claims** | soR: `claims` A reported loss case against a Policy. Aggregates one or more LossEvents and one or more Exposures (per coverage). Long-horizon entity with its own state machine. | Field | Type | Req | | ------------- | ------------------------------------------------------------------------------- | --- | | `claimId` | id | req | | `claimNumber` | string | req | | `policyRef` | Ref\ | req | | `claimantRef` | Ref\ | opt | | `reportedAt` | datetime | req | | `incurredAt` | datetime | req | | `severity` | enum | opt | | `status` | enum: `draft` \| `open` \| `investigating` \| `settled` \| `closed` \| `denied` | req | | `currency` | iso4217 | req | **Links**: `policy → Policy`, `lossEvent → LossEvent`, `exposures → Exposure[]`, `reserves → Reserve[]`, `payments → Payment[]`. **ontology · claims** The real-world loss occurrence — the underlying fact a Claim is reporting. Distinct from the Claim because one event may produce multiple Claims (third-party, multi-policy, multi-jurisdiction). | Field | Type | Req | | ------------- | ----------- | --- | | `lossEventId` | id | req | | `occurredAt` | datetime | req | | `locationRef` | LocationRef | opt | | `cause` | string | opt | | `description` | string | opt | **Links**: `claims → Claim[]`. **ontology · claims** The per-Coverage manifestation of a Claim. A single Claim against a Policy with three Coverages produces up to three Exposures, each reserved and settled independently. | Field | Type | Req | | ------------- | ---------------- | --- | | `exposureId` | id | req | | `claimRef` | Ref\ | req | | `coverageRef` | Ref\ | req | | `status` | enum | req | | `reserveRefs` | Ref\\[] | opt | | `paymentRefs` | Ref\\[] | opt | **ontology · claims** | soR: `claims · ledger` A held estimate of expected payout against an Exposure. Reserves are typed (indemnity, expense, recovery), have an estimator (model, adjuster, agent), and are versioned over the life of the Claim. | Field | Type | Req | | ------------- | -------------------------------------------- | --- | | `reserveId` | id | req | | `exposureRef` | Ref\ | req | | `kind` | enum: `indemnity` \| `expense` \| `recovery` | req | | `amount` | Money | req | | `estimator` | EstimatorRef | req | | `asOf` | datetime | req | **ontology · claims** | soR: `billing · gl` A typed outflow against a Claim Exposure. Always settled through the carrier's billing/GL system; Layerup records the intent, the approval, and the SoR confirmation. | Field | Type | Req | | ------------- | -------------- | --- | | `paymentId` | id | req | | `exposureRef` | Ref\ | req | | `amount` | Money | req | | `payeeRef` | PayeeRef | req | | `method` | enum | req | | `committedAt` | datetime | opt | | `sorReceipt` | string | opt | ### Group E · Workflow **ontology · workflow** A unit of operator work. Tasks are routed to roles or named principals via queues and have SLAs. Tasks may be created by agents, by humans, or by policy. | Field | Type | Req | | ------------- | ------------------------------------------------------------------- | --- | | `taskId` | id | req | | `kind` | string | req | | `subjectRef` | Ref\ | req | | `assigneeRef` | PrincipalRef | opt | | `queue` | string | req | | `dueAt` | datetime | opt | | `status` | enum: `open` \| `in_progress` \| `blocked` \| `done` \| `cancelled` | req | **ontology · workflow** A typed deviation from the expected path that requires human or policy resolution. Examples: low-confidence extraction, integration unavailable, conflicting evidence, authority breach, schema mismatch. | Field | Type | Req | | ------------- | ---------------- | --- | | `exceptionId` | id | req | | `code` | string | req | | `severity` | enum | req | | `subjectRef` | Ref\ | req | | `detail` | string | req | | `raisedAt` | datetime | req | | `resolution` | ResolutionRecord | opt | ### Group F · Evidence **ontology · evidence** A content-addressed file (PDF, image, spreadsheet, structured form). Always hashed; original bytes retained per retention policy. Ingest produces a Document; OCR / VLM / extraction produce derived properties that link back via EvidenceSpan. | Field | Type | Req | | ------------- | ---------- | --- | | `documentId` | id | req | | `contentHash` | sha256 | req | | `mediaType` | mime | req | | `byteSize` | int | req | | `pageCount` | int | opt | | `language` | bcp47 | opt | | `ingestedAt` | datetime | req | | `storageRef` | StorageRef | req | **ontology · evidence** An inbound or outbound conversation thread. Holds participants, message order, and a set of bound Attachments. Drives many ingest paths (submissions, loss notices, endorsement requests). | Field | Type | Req | | ---------------- | ----------------------------- | --- | | `threadId` | id | req | | `subject` | string | req | | `participants` | EmailParty\[] | req | | `messages` | EmailMessage\[] | req | | `attachmentRefs` | Ref\\[] | opt | | `direction` | enum: `inbound` \| `outbound` | req | **ontology · evidence** A Document bound to an EmailThread (or a portal upload). Carries the parent thread reference, the part name, and content disposition. | Field | Type | Req | | -------------- | ----------------- | --- | | `attachmentId` | id | req | | `threadRef` | Ref\ | req | | `documentRef` | Ref\ | req | | `partName` | string | opt | | `disposition` | string | opt | **ontology · evidence** A typed citation linking a derived property (or a Decision) back to the exact bytes that justify it. The substrate's primary mechanism for evidentiary trace. | Field | Type | Req | | --------------------- | ---------------------------------------------------- | --- | | `spanId` | id | req | | `documentRef` | Ref\ | req | | `locator` | oneOf: byteRange \| bbox \| timestampRange \| rowCol | req | | `retrievalSnapshotId` | string | opt | | `extractor` | ToolRef | opt | | `confidence` | number \[0,1] | opt | ```json theme={null} { "$id": "layerup://ontology/v1/EvidenceSpan", "type": "object", "required": ["spanId", "documentRef", "locator"], "properties": { "spanId": { "type": "string" }, "documentRef": { "$ref": "layerup://ontology/v1/Ref" }, "locator": { "oneOf": [ { "type": "object", "required": ["kind","start","end"], "properties": { "kind": { "const": "byteRange" }, "start": { "type": "integer" }, "end": { "type": "integer" } } }, { "type": "object", "required": ["kind","page","x","y","w","h"], "properties": { "kind": { "const": "bbox" }, "page": { "type": "integer" }, "x":{"type":"number"},"y":{"type":"number"},"w":{"type":"number"},"h":{"type":"number"} } }, { "type": "object", "required": ["kind","start","end"], "properties": { "kind": { "const": "timestampRange" }, "start": { "type": "string", "format": "date-time" }, "end": { "type": "string", "format": "date-time" } } }, { "type": "object", "required": ["kind","sheet","row","col"], "properties": { "kind": { "const": "rowCol" }, "sheet": { "type": "string" }, "row": { "type": "integer" }, "col": { "type": "integer" } } } ] }, "retrievalSnapshotId": { "type": "string" }, "extractor": { "type": "string" }, "confidence": { "type": "number", "minimum": 0, "maximum": 1 } } } ``` ### Group G · Governance **ontology · governance** | first-class: `core` A typed verdict produced by either an agent or a human reviewer against a typed input. Holds linked evidence, model lineage, rationale, and any consequent Actions. Decisions are immutable; corrections are new Decisions with a `supersedes` link. | Field | Type | Req | | --------------- | --------------------- | --- | | `decisionId` | id | req | | `kind` | string | req | | `subjectRef` | Ref\ | req | | `verdict` | VerdictValue | req | | `rationale` | string | req | | `authoredBy` | PrincipalRef | req | | `runRef` | Ref\ | opt | | `evidenceSpans` | Ref\\[] | req | | `modelLineage` | ModelLineage\[] | opt | | `supersedes` | Ref\ | opt | | `committedAt` | datetime | req | ```json theme={null} { "$id": "layerup://ontology/v1/Decision", "type": "object", "required": ["decisionId","kind","subjectRef","verdict","rationale","authoredBy","evidenceSpans","committedAt"], "properties": { "decisionId": { "type": "string" }, "kind": { "type": "string" }, "subjectRef": { "type": "string" }, "verdict": { "type": "object" }, "rationale": { "type": "string" }, "authoredBy": { "type": "string" }, "runRef": { "type": "string" }, "evidenceSpans": { "type": "array", "items": { "$ref": "layerup://ontology/v1/Ref" } }, "modelLineage": { "type": "array", "items": { "$ref": "layerup://ontology/v1/ModelLineage" } }, "supersedes": { "type": "string" }, "committedAt": { "type": "string", "format": "date-time" } } } ``` **ontology · governance** | first-class: `core` A typed intent to mutate a system of record. Lives on the Action Plane (§14) with state machine *proposed → staged → approved → committed → reverted/rejected/failed*. Always carries an idempotency key. | Field | Type | Req | | ---------------- | ------------ | --- | | `actionId` | id | req | | `kind` | string | req | | `targetSor` | string | req | | `targetRef` | Ref\ | opt | | `payload` | Json | req | | `idempotencyKey` | string | req | | `state` | enum | req | | `proposedBy` | PrincipalRef | req | | `approvedBy` | PrincipalRef | opt | | `committedAt` | datetime | opt | | `compensatedBy` | Ref\ | opt | **ontology · governance** One bounded execution of an agent against an input object. Has a unique runId, a scoped permission set, a token / wall-clock / cost budget, and a deterministic seed for replay. | Field | Type | Req | | -------------- | ------------------------------------------------------------------------------------ | --- | | `runId` | id | req | | `agentRef` | AgentRef | req | | `agentVersion` | semver | req | | `subjectRef` | Ref\ | req | | `scopes` | string\[] | req | | `seed` | string | req | | `budget` | Budget | req | | `state` | enum: `queued` \| `running` \| `verified` \| `handed_off` \| `completed` \| `failed` | req | | `decisions` | Ref\\[] | opt | | `actions` | Ref\\[] | opt | | `startedAt` | datetime | req | | `finishedAt` | datetime | opt | **ontology · governance** | hash-chain: `tamper-evident` The hash-chained record of every governance-relevant action in the platform: tool dispatches, decision commits, action commits, policy rulings, identity changes, configuration changes. The chain is anchored periodically (§17). | Field | Type | Req | | ------------- | ------------ | --- | | `eventId` | id | req | | `seq` | int | req | | `kind` | string | req | | `actor` | PrincipalRef | req | | `subjectRef` | Ref\ | opt | | `payloadHash` | sha256 | req | | `prevHash` | sha256 | req | | `thisHash` | sha256 | req | | `at` | datetime | req | ```json theme={null} { "$id": "layerup://ontology/v1/AuditEvent", "type": "object", "required": ["eventId","seq","kind","actor","payloadHash","prevHash","thisHash","at"], "properties": { "eventId": { "type": "string" }, "seq": { "type": "integer", "minimum": 0 }, "kind": { "type": "string" }, "actor": { "type": "string" }, "subjectRef": { "type": "string" }, "payloadHash": { "type": "string", "pattern": "^[a-f0-9]{64}$" }, "prevHash": { "type": "string", "pattern": "^[a-f0-9]{64}$" }, "thisHash": { "type": "string", "pattern": "^[a-f0-9]{64}$" }, "at": { "type": "string", "format": "date-time" } } } ``` ## 5.3 Marking inheritance & derivation Markings (§4.2) propagate along derivations: any property derived from a source carries the union of the source's markings unless the derivation explicitly de-classifies (and the de-classification is itself audited). EvidenceSpans inherit the marking of the underlying Document. Decisions inherit the union of all evidence markings. ## 5.4 Default permission set | Object | Read scope | Write scope | Default markings | | ---------------------------------------------------------------- | ----------------- | ---------------- | ------------------------------- | | Insured / Claimant | `party.read` | `party.write` | `pii.medium` | | Policy / Coverage / Endorsement | `policy.read` | `policy.write` | `tenant` | | Submission / Quote / RiskAssessment / Pricing / BindingAuthority | `uw.read` | `uw.write` | `tenant` | | Claim / LossEvent / Exposure / Reserve / Payment | `claim.read` | `claim.write` | `pii.medium` (where applicable) | | Task / Exception | `workflow.read` | `workflow.write` | `tenant` | | Document / EmailThread / Attachment / EvidenceSpan | `evidence.read` | `evidence.write` | inherits from source | | Decision / Action / AgentRun / AuditEvent | `governance.read` | append-only | `tenant` | ## 5.5 Per-LOB ontology specialization The 28 canonical objects defined above are *line-of-business agnostic*. Specific LOBs need additional fields, additional relationships, and LOB-specific Decision and Action types. Layerup models this as **ontology branches** (§6.3): the canonical trunk holds the cross-LOB shape; each LOB branch adds typed extensions that inherit from the trunk. This means a carrier deploying P\&C, Specialty, and Life on the same Layerup instance does *not* negotiate a single super-schema; each branch is governed independently and replay still works against any branch version. **Anatomy of a branch** * **Extends**: which trunk objects the branch extends (e.g. `Claim`, `Coverage`). * **New properties**: LOB-specific typed fields (e.g. `vehicle.vin` on Personal Auto Claims). * **New relations**: LOB-specific links (e.g. `Claim → Vessel` on Marine). * **New Decision / Action types**: LOB-specific reasoning outputs and effect intents. * **LOB markings**: e.g. `phi.high` on Health, `treaty.ceded` on Reinsurance. * **Pinned tools**: per-LOB tool versions in the registry (§9). **P\&C / Personal** — Extends `Claim`, `Coverage`, `Exposure`. Adds `Vehicle` (VIN, year/make/model, garaging), `DriverInvolvement`, `PoliceReport`, `RepairFacility`. Decision types: *FastTrackEligibility*, *TotalLossDecision*, *SubrogationCandidate*. Markings: `pii.medium`. **P\&C / Personal** — Extends `Claim`, `Coverage`, `Exposure`. Adds `Property`, `Peril`, `Inspection`, `ContractorEstimate`. Decision types: *CoverageApplicability*, *CATAccumulation*. Markings: `pii.medium`. **Commercial** — Extends `Submission`, `Policy`, `Claim`. Adds `InsuredEntity`, `Location`, `OccupancyClass`, `SchedulesOfValues`, `BusinessInterruption`. Decision types: *UnderwritingDecision*, *ReservingTier*. Branch governance is heavier (broker-mediated submissions). **Casualty** — Adds `Employer`, `JurisdictionalRules`, `InjuryCode` (ICD-10), `WageStatement`, `MedicalProvider`, `RTWPlan` (return-to-work). Decision types: *CompensabilityDecision*, *IndemnityRate*, *IMETriggers*. Markings: `phi.medium`, `pii.medium`. **Specialty** — Adds `Vessel`, `Voyage`, `BillOfLading`, `SurveyorReport`, `HullParticulars`, `CargoManifest`. Decision types: *GeneralAverageDeclaration*, *SalvageCoverage*, *JettisonAdjudication*. Often multi-jurisdictional. **Specialty** — Extends `Submission`, `Quote`, `BindingAuthority`. Adds `BrokerOfRecord`, `RiskAppetiteRule`, `FacultativeReinsurance`, `SLAInsurer`. Decision types: *AppetiteFit*, *FacRequired*, *BindingAuthorityCheck*. **Life & Annuities** — Adds `InsuredLife`, `Beneficiary`, `RiderSchedule`, `UnderwritingClass`, `MedicalEvidence`. Decision types: *RiskClassification*, *BeneficiaryDispute*, *ContestabilityCheck*. Markings: `phi.high`, `pii.high`. **Health** — Adds `Member`, `ProviderNetwork`, `Authorization`, `EOB`, `ClaimLine` (CPT / HCPCS). Decision types: *MedicalNecessity*, *NetworkAdjudication*, *AppealDisposition*. Heavy `phi.high`; HIPAA / HITECH governance applied at the marking level. **Treaty** — Adds `Treaty`, `CededLayer`, `CessionStatement`, `BordereauLine`, `FacRiskFile`. Decision types: *CessionDetermination*, *ReportingTrigger*, *TreatyExceptionFlag*. Often dual-tenant (cedant / reinsurer). An LOB branch is a typed *extension* of the trunk, not a fork. Branches inherit trunk versioning, lineage, audit, and governance. Cross-branch operations (e.g. a Commercial Property claim that triggers Reinsurance cession) traverse the trunk via the canonical objects both branches share. There is no "Marine instance" of Layerup separate from the "Auto instance" — there is one substrate, with branches. From a global CTO standpoint, this is the property that matters: **the cost of adding a new LOB is the cost of authoring that LOB's branch and its tools, not the cost of standing up a new platform.** Identity, ingestion, runtime, control, action, telemetry, audit, deployment topology, and SRE all carry over unchanged.