Handwriting OCR

The API supports PDFs and common image formats, including JPG, PNG, and TIFF. The web workflow accepts document photos and PDFs, including batches of scans.

Standard transcription requires no setup or training: upload the document, wait for processing, and export the result. Custom extractors require you to define fields and test them against a sample first.

Accuracy depends on the original handwriting, language, and scan quality. Very faded pages, poor scans, or rare scripts may need closer review. Check the transcription and extracted fields against the original document before using them.

Documents are used for requested processing, not to train the service’s own AI models. Uploads are encrypted in transit, with access controls for stored files. Necessary hosting and operational providers can receive data, and optional connectors return results to the connected service, which may keep its own copy. Documents and results are deleted after seven days by default; settings allow ten minutes to fourteen days. Account, billing, and security records have separate retention.

A free trial is available without a credit card. API access is included on all plans, including the trial.

Each page or image costs one credit. Pay-as-you-go is $15 for 100 pages, valid for one year. Monthly subscriptions include Starter at $19 for 250 pages, Pro at $59 for 1,000 pages, and Business at $499 for 10,000 pages. Table extraction and custom extractors are included on Pro and Business. Subscription credits remain valid while the subscription is active.

The current API limit is two requests per second per account, shared across all endpoints. Webhooks can deliver results without repeated polling. Higher throughput is available to Enterprise customers by agreement.