Skip to main content

Data Curation API Reference

The Data Curation API transforms raw files into structured, AI-ready content through a configurable processing pipeline: text extraction → normalization → optional PII detection/redaction → chunking → optional embedding generation.

Processing is asynchronous. Submit a job with POST /presign, upload the file, poll GET /status/{job_id} until the job reaches a terminal state (Done or Error), and then download results via the get_url.


Base URLs

EnvironmentBase URL
Production (US)https://knowledge-enrichment.ai.app.hyland.com/latest/api/data-curation
Production (EU)https://knowledge-enrichment.ai.app.hyland.eu/latest/api/data-curation

All of the following endpoints are relative to the base URL for your environment.


Authentication

All requests require a Bearer token in the Authorization header:

Authorization: Bearer <access_token>

Obtain a token using the client credentials OAuth 2.0 flow against the token endpoint for your environment:

EnvironmentToken endpoint
Production (US)https://auth.app.hyland.com/idp/connect/token
Production (EU)https://auth.app.hyland.eu/idp/connect/token

See Authentication for step-by-step instructions.


Pages in this section

PageContents
Configuration API OverviewList of all endpoints
PII HandlingPII entity types, redaction modes, output structure
EndpointsFull endpoint reference with schemas and examples
Code SamplesPython, C#, JavaScript

Limits and constraints

PropertyValue
Max upload size per single PUT5 GB
Supported file types600+ types (full list)
Rate limitsSubject to AWS API Gateway quotas
tip

As a best practice, large numbers of files should be sent in batches to the same presigned endpoint to avoid creating more endpoints than are needed. This can be handled either by sending an entire folder of files with the Upload Script or by iterating through each file.

If any given file exceeds the size limit, it must use a multipart upload or split into smaller parts before processing. Refer to Amazon S3 documentation for details.

When reviewing the supported file formats for the Data Curation API, also note the following:

  • Any document file with text layers is extracted as is.
  • The audio is transcribed from any audio or video files.
  • Image files with a high likelihood of text presence are processed through OCR.
  • Any files not encompassed by the preceding types are also processed through OCR.