DOCUMENT TOOLS, WITH A HUMAN IN CONTROL
Privacy and document processing information
The application uses account name, email, optional phone, sign-in sessions, saved files, extracted text, reviewed fields, conversion results and corrections to provide selected tools. Processing admission records contain task type, account identifier and timing metadata, not uploaded document text.
Cloudflare hosts the application. The hosted setup uses Supabase for authentication, database and private file storage. Your browser requests Google Fonts. Authentication email uses the configured email provider. Optional AI handwriting, semantic extraction and translation send selected images or text to the configured AI provider after consent. An explicitly selected external converter receives its input file.
Camera preview frames are processed in your browser. Choosing Capture reviewed scan uploads the captured image. Saved content has no automatic expiry schedule in the application. Use Delete on saved documents or conversion results; successful document deletion removes the active original, document record and linked review records.
Application deletion does not manage provider backup snapshots. A backup retention and purge schedule has not been verified. Operator legal identity and dedicated privacy contact have not been supplied. No legal basis, region-specific rights process, jurisdiction or service terms are being invented here. These details must be completed before treating this as a complete operator privacy policy.
Use the deployment contact shown in the footer when configured. Do not upload records that require unverified retention, jurisdiction, legal or confidentiality guarantees. Automated outputs need human review before use.
Turn document images into information you can review.
EXPLORE IOCRIFYThe format you need. A workflow you can follow.
EXPLORE IOCRIFYMake your next PDF easier to use.
EXPLORE IOCRIFYConfidence starts with evidence.
EXPLORE IOCRIFYLess retyping. More time for the work that matters.
EXPLORE IOCRIFYTurn invoice documents into reviewed business data.
EXPLORE IOCRIFYMake receipts easier to read, review and organize.
EXPLORE IOCRIFYRead a document in its language. Translate when you choose.
EXPLORE IOCRIFYBring handwritten notes into a reviewed text workflow.
EXPLORE IOCRIFYMove structured records without losing the meaning inside a cell.
EXPLORE IOCRIFYKeep people in charge of document decisions.
EXPLORE IOCRIFYWhat works today, and what we are exploring next.
EXPLORE IOCRIFYUnderstand what happens to your documents.
Your private workspace
Checking OCR availability…
Upload a file or select a document to review.
Review extracted text. Your original file remains unchanged.
Choose a target language. Translation sends the saved extracted text to the configured AI provider.
Add values you have checked. Field names must be unique; up to 100 fields.
Choose your files, select a compatible output, and convert. Output formats reflect the available engines.
No files selected.
File conversion uses original contents. Use the OCR editor to export reviewed text. Image conversion uses the first image frame.
Saved to your account. Download a result or delete it when you’re finished.
PDF workspace
Upload, select an area, add edits, then save a new copy.
Omit pages to delete or split. List order controls sorting. Operations target the selected original page and follow it into the saved order.
Coordinates use unrotated page space. Draw on the preview or enter them here.
Typed signatures are visual marks, not cryptographic signatures. Redaction removes covered content. Editing signed PDFs can invalidate their signatures.
No edits queued.
Draw a rectangle over the page to position an edit. The preview shows the original; save and reopen to review the result.
Visual schema mapper
Use the open document’s reviewed fields or paste JSON. Drag a value into a mapping row, or click it for keyboard access. This maps values; it does not validate an ERP’s XSD.
Validate each extracted field against its source. Accept or correct a value, then choose whether to teach a label rule for future uploads.
Upload a document on the Documents & OCR page, then return here to validate extracted label-value fields.
Loading source…
SOURCE SNIPPET
OCR scores describe text recognition, not factual accuracy. Unscored fields need human review. Bounding boxes appear only where source words can be located exactly. Older uploads without stored evidence may need to be uploaded again.
Follow the document from source to structured output. Verify proposals and resolve warnings before using them downstream.
Choose a document to see its intelligence.
Focus on unscored or low-confidence fields and failed domain checks. Use Older documents to review earlier pages.
Import a vendor master snapshot from your ERP, Oracle or another system. Similarity suggestions never replace an extracted name automatically.
Your corrections inform future extractions through reusable label mappings. Inspect each rule and change its status. Learning stays within your account.
Enabled rules map source labels to field names on new uploads and provide label context to the optional Vision engine. Existing documents keep their original mapping.
The latest 100 decisions, including original and corrected values. Deleting a document also removes its review activity.
Future roadmap · Not yet available
The current release gives you OCR, conversion and reviewed editing. The next direction connects documents to structured commercial and consumer workflows.
These are planned capabilities. No compliance certification or automated legal judgment is provided.
Ready