Your extraction pipeline,
fully under your control.
Define schemas, classify documents, split and chain operations — all from a single platform. Build, test and deploy with full visibility.
Design your data model,
teach the AI what to look for.
Create custom schemas and classifiers without writing ML code. Define the fields you need, the categories that matter, and let the platform handle the rest.
Schema editor
Define exactly what data to extract with a custom schema. Specify field names, types, and descriptions — receive structured JSON matching your definition.
Classifier builder
Create document categories with just a label and a short description. The AI learns to sort incoming documents by type automatically.
Split, chain, and orchestrate
complex document workflows.
Handle multi-document PDFs and build processing pipelines that combine classification, partitioning and extraction into a single flow.
Partitioner
Upload a PDF containing multiple documents and get each one separated automatically. Page-level detection identifies where each document starts and ends.
Composer
Chain multiple operations into a single pipeline. Classify first, then route each category to its own extraction schema — all in one API call.
One API,
three ways to process.
Choose the processing mode that fits your use case. From real-time single documents to bulk batch jobs — same API, different throughput.
View API referenceNormal
Synchronous processing. Send a document, get structured data back instantly. Ideal for real-time integrations.
Async
Asynchronous with webhook callback. Submit the document and receive results via webhook when ready.
Batch
Bulk processing. Submit hundreds of documents in a single request. Perfect for nightly imports and migrations.
Measure extraction quality,
field by field.
Run evaluations against ground truth datasets. Track precision and recall per field and catch regressions before they reach production.