Document OCR

Classify supported Bhutan documents, return verified fields only after strict eligibility, and show parser candidates separately as an unverified preview.

Use a phone photo of the physical document and send it exactly as the phone saved it: no cropping, resizing or re-saving. Screenshots, downloaded images and anything under 640 × 400 px are sent to manual review.

const ocr = await client.ocr.scanDocument(frontImage, { backImage, type: 'cid' });
— run something above and the SDK call appears here —