NeuDocIQ
Request a Demo
API CategoryFlagship

AutoExtract

AutoExtract is NeuDocIQ's flagship intelligent pipeline. It parses incoming documents, classifies document subtype, routes to the right schema, and extracts structured outputs in a single workflow.

Business Use Case

Use AutoExtract when incoming documents are mixed, misnamed, or unknown. Teams in ops, BFSI, and government onboarding can process high-volume streams without manually pre-sorting files.

When To Use

  • Mixed inbound queues where document type is unknown in advance.
  • High-throughput batch ingestion with parallel auto-routing.
  • You need classification confidence with extraction response.
  • You want optional fallback schema behavior for low-confidence classification.

Why AutoExtract Exists

AutoExtract is designed for real-world intake pipelines where documents arrive unlabeled, mixed, and inconsistent. It removes the need to classify, route, and extract in separate steps by turning a single API request into a full document intelligence workflow.

Step 1

Ingest

Accept raw files or stored document references.

Step 2

Classify + Route

Infer document family and choose the extraction schema.

Step 3

Extract

Return normalized structured output for downstream systems.

Common Use Cases

  • Mixed enterprise inboxes where document type is unknown ahead of time.
  • Onboarding flows that need one API path for many document families.
  • Back-office review systems that require traceable extraction output and confidence checks.
  • Government and BFSI programs that need a durable audit trail from source file to structured record.

End-to-End Process

  1. Upload or reference the source document.
  2. Trigger AutoExtract with client and workflow context.
  3. Allow the service to parse, classify, and select the right extraction path.
  4. Read structured output and confidence indicators.
  5. Escalate uncertain results into human review and retain document identifiers for traceability.

Troubleshooting

  • Low confidence: tighten routing inputs and review fallback schema behavior.
  • Unexpected field gaps: inspect parse quality and schema coverage for the document family.
  • High volume latency: use batch AutoExtract for queue-based ingestion.

Integration Guide

  • Persist document IDs from upload flows so the same source asset can be reused.
  • Store classification and extraction outputs together for auditability.
  • Design downstream review queues around confidence thresholds and validation outcomes.

Best Practices

  • Use batch mode for bulk ingestion and single-call mode for synchronous operator workflows.
  • Track fallback behavior explicitly so degraded classification paths remain observable.
  • Keep source document identifiers and extraction results linked in your data model.

Endpoint Directory

POSTAutomatic Classification and Extraction

/api/v1/extract/auto

POSTBatch Auto-Classification and Extraction

/api/v1/extract/auto/batch

Endpoint Details

POST

Automatic Classification and Extraction

Automatic Classification and Extraction This endpoint is documented for production integration usage.

auto-extract

Endpoint URL

https://api.neudociq.com/api/v1/extract/auto

Overview

Automatic Classification and Extraction This endpoint is documented for production integration usage.

Purpose

Automatic Classification and Extraction

Use Case

Use this endpoint as part of your document processing workflow where the operation path is required.

Authentication

  • APIKeyHeader: API key in header (X-API-Key).

Header Example

http
X-API-Key: ndq_test_api_key

Headers

No parameters.

Query Parameters

No parameters.

Path Parameters

No parameters.

Request Fields

FieldTypeRequiredDescriptionExample
client_idstringRequiredClient identifier for schema routingclient_id_value

Request Default

Copy-ready examples with real endpoint paths and authentication placeholders.

javascript
const url = "https://api.neudociq.com/api/v1/extract/auto";
const response = await fetch(url, {
  method: 'POST',
  headers: {
    'X-API-Key': 'ndq_test_api_key',
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
  "document_id": "string_value",
  "document_url": "string_value",
  "client_id": "client_id_value",
  "document_type": "string_value",
  "confidence_threshold": 0.7,
  "use_llm_fallback": false,
  "fallback_schema_id": "string_value",
  "select_pages": 1,
  "maintain_format": true,
  "enable_citations": true,
  "return_raw_on_failure": true,
  "skip_failure_reextraction": false
}),
});

const data = await response.json();
console.log(data);

Best Practices

  • Validate required fields client-side before sending requests.
  • Log request identifiers and HTTP status codes for supportability.
  • Retry only when the operation is safe or your workflow is idempotent.
  • Route low-confidence classifications into review queues instead of silently accepting them.
  • Use batch AutoExtract for high-volume ingestion rather than sending many single requests concurrently.
POST

Batch Auto-Classification and Extraction

Batch Auto-Classification and Extraction This endpoint is documented for production integration usage.

auto-extract

Endpoint URL

https://api.neudociq.com/api/v1/extract/auto/batch

Overview

Batch Auto-Classification and Extraction This endpoint is documented for production integration usage.

Purpose

Batch Auto-Classification and Extraction

Use Case

Batch endpoints are asynchronous; create the job first, then use status and list endpoints to monitor completion.

Authentication

  • APIKeyHeader: API key in header (X-API-Key).

Header Example

http
X-API-Key: ndq_test_api_key

Headers

No parameters.

Query Parameters

No parameters.

Path Parameters

No parameters.

Request Fields

FieldTypeRequiredDescriptionExample
client_idstringRequiredClient identifier for schema routingclient_id_value

Request Default

Copy-ready examples with real endpoint paths and authentication placeholders.

javascript
const url = "https://api.neudociq.com/api/v1/extract/auto/batch";
const response = await fetch(url, {
  method: 'POST',
  headers: {
    'X-API-Key': 'ndq_test_api_key',
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
  "storage_path": "string_value",
  "query": {},
  "documents": [
    {}
  ],
  "client_id": "client_id_value",
  "document_type": "string_value",
  "confidence_threshold": 0.7,
  "fallback_schema_id": "string_value",
  "file_pattern": "string_value",
  "use_regex": false,
  "max_documents": 1,
  "max_concurrent": 1,
  "select_pages": 1,
  "maintain_format": true,
  "enable_citations": true,
  "priority": 0,
  "webhook_url": "string_value",
  "skip_if_already_extracted": false,
  "enable_quality_gate": true,
  "quality_null_threshold": 0.7,
  "quality_min_fields": 2
}),
});

const data = await response.json();
console.log(data);

Best Practices

  • Validate required fields client-side before sending requests.
  • Log request identifiers and HTTP status codes for supportability.
  • Retry only when the operation is safe or your workflow is idempotent.
  • Route low-confidence classifications into review queues instead of silently accepting them.
  • Use batch AutoExtract for high-volume ingestion rather than sending many single requests concurrently.
  • Poll status endpoints with backoff and stop polling once the batch reaches a terminal state.