Overview
The Prep API turns large, messy clinical document bundles into cleanly split, classified, and NLP-ready documents. It is built on the emtelligent OCR engine and designed for the complex faxed and electronically transmitted chart bundles and APS reports common in medicine today.
A single multi-hundred-page fax or chart bundle becomes a set of discrete, individually addressable documents — each with a document-type classification, its extracted metadata, and its full extracted text.
Capabilities
The Prep API is modular. You select any combination of the following functions.
Input and output
Input — PDF and TIFF, including large multi-page bundles.
Output — for each document detected in a bundle: its document-type classification, its extracted metadata, and its full extracted text, delivered together as one structured result. When requested, the whole bundle is also emitted as a single NLP-ready .jsonl feed.
How a job works
The Prep API is asynchronous and reads from and writes to cloud object storage that you control. You grant it scoped access to an input bucket and an output bucket, then submit jobs that reference objects in the input bucket by key.
A job can be cancelled while it is still in flight. Requests authenticate with either an API key or a JWT.
Ways to use it
Process documents without writing code. The quickest way to try the API.
Call the submit, status, and cancel endpoints directly from any language.
Build the job, poll it, and read results from Python.
Using Prep with the other APIs
Prep can be used on its own, and it uses the OCR API internally, so you do not need to call OCR separately to process scanned or faxed pages. Prep is also used by the Extract API, which orchestrates Prep and the NLP API through managed pipelines.
If your goal is structured clinical data rather than prepared documents, the Extract API will do that in one call. Use Prep directly when you need the split and classified documents themselves, or want control over the individual stages.