Your bill is a document, not a data opportunity.

When you upload a bill or wallet screenshot, the extraction endpoint receives the image, validates its type and size, and runs text recognition. The checker processes detected coupon numbers, dates and optional bill details only for the requested review and check.

This implementation does not write uploaded images to a database or permanent file storage. Images are handled in request memory for extraction. Hosting providers may still maintain temporary operational logs or network buffers outside the application's direct control.

By default, bundled English and Nepali OCR runs within the application's server process; the image is not sent to a separate OCR company. Operators can explicitly configure a remote OCR endpoint, and must disclose that provider and its retention terms before doing so in production.

Detected 12-digit coupon numbers are sent from the browser to the Tax Prize server. The server retrieves the public winner list from the official IRD prize portal and compares them locally; it does not submit your bill or screenshot image to the IRD winner endpoint.

The API uses short-lived, in-memory request counters and server-side response caching. Depending on deployment, an IP-derived key may be held temporarily for rate limiting. It is not used to build a user profile.

Crop out personal details that are not relevant to extraction. You can avoid image processing entirely by selecting manual entry and entering only the prize coupon number.