> ## Documentation Index
> Fetch the complete documentation index at: https://docs.amlok.io/llms.txt
> Use this file to discover all available pages before exploring further.

# API correspondence

> Trace integration coverage, normalized fields, risk-package results and research concepts to their exact roles in the public API.

The API reference, Data Model and console Data Lookup describe one public API. Use this correspondence to choose the correct resource and interpret integration and research coverage. There are **57 public operations**, **64 normalized evidence variants**, **961 source field definitions**, **860 detailed verification fields**, and **149 risk metrics**. These are different counts; none guarantees a populated value for every borrower.

## From authorized source to API result

| Need | API operation or representation | Meaning |
| - | - | - |
| Register the institution's applicant | `POST /v1/borrowers`, `Borrower.subject_id` | Institution-supplied identity, not a verified conclusion |
| Launch a configured Hub | `POST /v1/embedding-sessions` | Borrower-bound authorization journey; configuration stays in the console |
| Inspect connection health | `Connection.status`, `data_status`, source freshness | Authorization and collection progress are separate |
| Retrieve collected and enriched records | `DataRecord.normalized`, `resource_type`, `resource_id`, `record_version` | Typed source observation or derived record, independently of an assessment |
| Retrieve permitted provider-native details | `DataRecord.source_payload` | Source payload availability and licensing are explicit; never credentials or unrelated data |
| Submit underwriting inputs | `POST /v1/originations` | Application, documents, exact product/policy versions and connection scope |
| Read underwriting and monitoring results | `RiskPackage.dimensions`, `analyses`, `verification`, policy results | Immutable assessment with explicit limitations and evidence |
| Verify the exact evidence used | `Evidence.record.data_record_id`, `Evidence.normalized` | Frozen input version, not an automatic substitute with the latest record |
| Retrieve reports and original files | `Document` and authorized document download | Access and disclosure rules apply to files as well as normalized fields |
| Monitor a funded exposure | `Monitor`, `MonitoringRun`, `Alert` and deal snapshots | New observations and packages preserve the baseline and earlier versions |
| Reconcile billing | Billing summary, transactions, payments, usage and exports | Institution workspace funding and consumption, distinct from a borrower's normalized payment records |

Use [Data access](https://docs.amlok.io/api-reference/data-access.md), [evidence traceability](https://docs.amlok.io/api-reference/evidence-traceability.md), [large results](https://docs.amlok.io/api-reference/large-results.md) and the [endpoint reference](https://docs.amlok.io/api-reference/overview.md) for exact request and response shapes.

## Names do not imply identical schemas

`Connection.status` is product state, while `normalized.object=connection` describes source evidence with `NormalizedEnum_connection_status`. Likewise, normalized `consent`, `document`, `monitor`, `alert`, `application` and `risk_snapshot` are not replacements for their product response schemas. Use [enumerations](https://docs.amlok.io/data-model/enumerations.md) for the correct vocabulary. Read-only processing identifiers are provenance, not orchestration commands.

`DataRecord.id` identifies an immutable data version. `normalized.id` and `DataRecord.resource_id` identify the logical normalized record. `Borrower.id` and its business/person `subject_id` are also distinct. Exact data-record IDs connect frozen evidence to those versions.

## Integration coverage

The [integration catalog](https://docs.amlok.io/integrations/status.md) has **238 entries: 22 Active and 216 In Review**. Preserve country, Beta, product and entitlement conditions. The **2,036 provider-to-resource associations** describe candidate data coverage. The **71 documented field mappings** cover selected upstream paths; they are not a complete adapter conformance certificate for all 238 entries. All referenced normalized resources belong to the current contract.

Internal verification and registry sources supplement borrower-connected integrations. They are not additional customer-selectable Hub integrations. Their authorized results use the [KYC/KYB model](https://docs.amlok.io/api-reference/verification.md), source records, evidence and document references. No provider's entire API is promised, and denied or unavailable information stays explicit.

## Financial and verification definitions

The [source dictionary](https://docs.amlok.io/data-model/field-dictionary.md) documents normalized fields and partial-source behavior. The [verification dictionary](https://docs.amlok.io/api-reference/verification-fields.md) documents 25 detailed observation sections. The [risk metric catalog](https://docs.amlok.io/api-reference/underwriting-dimensions.md) documents 24 dimension families and 149 approved metric keys. The 56 [source-derived attribute definitions](https://docs.amlok.io/data-model/derived-attributes.md) are supporting calculations, not an alternative risk metric enum.

An observed debit is not the complete debt obligation; a provider score is not an AMLOK-calibrated probability; a search candidate is not a verified identity. Preserve calculation version, metric basis, units, period, missing reasons and evidence. An empty collection needs its query/coverage status before it can mean no results.

## Research correspondence

| Research concept | Product representation | Boundary |
| - | - | - |
| Source identity and revisions | `DataRecord`, normalized `source_refs`, `field_provenance` | Research pseudonyms and training manifests are not new API fields |
| Economic reconciliation | normalized `reconciliation_link`, `RiskPackage.analyses.reconciliations` | Linked order, invoice, payment and deposit observations remain distinct |
| Financial features | `Metric.key`, `value`, `basis`, `inputs`, `coverage`, `calculation_version` | Only catalog keys are API metrics; proposed features require separate definition |
| Business/consumer study populations | Borrower type and metric applicability | Public-company or business benchmarks do not establish consumer performance |
| Identity and source reliability | `RiskPackage.verification` and metric/source coverage | A documented capability is not proof that a check completed |
| Change detection | Monitoring runs, immutable packages, alerts and the Data change feed | Model evaluation events are not webhook event names |
| Training and evaluation | Research datasets, labels and reproducibility artifacts | Consent to connect does not authorize shared training; benchmark results do not become borrower facts |

Read the [research data foundations](https://docs.amlok.io/research/data-foundations.md), [record model](https://docs.amlok.io/research/record-model.md), [financial analysis](https://docs.amlok.io/research/financial-analysis.md) and [evaluation](https://docs.amlok.io/research/evaluation.md). Proposed research models and measured public-data experiments remain distinct from the API contract and production performance claims.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.