Read on-device
Self-host the parser and its PDF/OCR assets. The browser reads the selected file; there is no document-upload endpoint or hosted AI model in this SDK.

Build credit education into your product. Turn a PDF or screenshot into structured fields—without sending the document to an external parsing service.
No credit card. One server key. Self-hosted browser SDK.
↓ Local PDF reading + OCR
{
"kind": "credit_card",
"balanceCents": 240000,
"limitCents": 600000,
"reviewRequired": true
}Illustrative output · no name, DOB or account number
Self-host the parser and its PDF/OCR assets. The browser reads the selected file; there is no document-upload endpoint or hosted AI model in this SDK.
Account types, balances, limits, status, month-level dates and page references. Missing values stay unknown instead of being guessed.
Show the original alongside the extracted fields. Corrections and clear warnings come before any education or downstream use.
This fictional screenshot runs through the real OCR reader in your browser. No personal report is needed, and the example does not consume your allowance.

Ready to read the fictional screenshot.
First OCR use downloads the self-hosted reader and English language model. Speed depends on your connection and device.
CA$0
100 authorized parse starts per UTC calendar month.
CA$19.99 / month
1,000 authorized starts per monthly billing period.
CA$49.99 / month
5,000 authorized starts per monthly billing period.
Custom
Discuss higher volume, deployment or procurement requirements.
Subject to a separate written agreement. No service level or custom integration is included unless agreed.
Discuss your integrationUp to 3 active server keys share one account’s allowance. Starts include later reading failures or cancellations; a retry with the same request ID in the same allowance period is not counted twice. Paid plans renew monthly; cancel renewal in your console. Review the paid SDK terms before checkout. Browser code can be modified; this is access metering, not tamper-proof offline usage tracking.
Conservative English-label extraction for Equifax, Experian and TransUnion, one bureau at a time. PDF, PNG and JPEG up to 20 MB, with up to 30 PDF pages. Layouts vary: extraction is partial, human review is required, and tri-merge reports are not supported. Scores and complete payment histories are not extracted.
No. The SDK extracts on the user’s device. The server API authorizes access using your secret key and a random request ID. It never needs a file, filename, report text, financial result or end-customer identifier.
No. It reduces external document exposure, but does not provide a regulatory exemption. Your app controls consent, page scripts, access and any later storage. Assess your deployment with your compliance team.
Yes—use the free AI Credit Analyzer. You do not need an API key or developer account.