Prepare documents for AI use or external review. Automatically find names, contact details and identifiers, review every change, and save a separate copy — entirely offline on your computer, even when it is not connected to the internet.
In development · macOS & Windows first · Linux later
Prepared for Morgan Ellis,
on behalf of Willow Advisory.
Project fee: 2,400
The scope and context stay right here.
Illustrative workflow · Fictional data · Review before sharing
Prepare a useful copy for the next task.
You decide what leaves your workspace.
Open your files, choose the relevant country and domain, and start a local review. Your original stays intact.
Inspect grouped findings. Redact, use consistent placeholders or keep a value. Manually mark anything the detector missed.
Export a new document. Check the result and any unresolved content before sharing it with an AI service or another person.
Remove sensitive details offline.
Then share only what the next task needs.
Need a summary, explanation or analysis? Redact names, contact details and account numbers before uploading a report, contract or statement to an AI tool. Keep the context it needs to help.
Sending a document to a consultant, vendor or external reviewer? Remove client, employee or customer information they do not need, then send a reviewed copy while keeping your original.
Using a website to merge, compress or convert a document? Remove sensitive information on your computer first. Check the saved copy, then upload that version to the online service you choose.
You decide whether sharing is appropriate. Unusual facts, amounts and relationships can still identify someone after direct identifiers are removed.
A focused desktop workflow.
No automatic upload to an AI service.
Rules and a local language model help find structured identifiers and entities such as names and organizations. Document processing happens on your device.
Use recognizable placeholders such as PERSON_1 and ORG_1 to retain context. Review repeated occurrences together instead of treating every match in isolation.
Choose an action for a finding, keep relevant text, or mark missed details. A detector’s suggestion is the start of a decision, not the last word.
Combine common recognition with country-specific identifier coverage and domain choices. Select the scope that fits the document and its intended use.
Keep the source document separate from your redacted output. Review the new copy before deciding where to send it.
Scans, protected files and complex embedded content need specific support. Check format coverage and unresolved regions rather than assuming an empty finding list means a clean file.
The initial scope is English-language documents. Exact tested coverage and system requirements will accompany each installer.
Parseable, unencrypted documents. Scanned or mixed-image PDFs require a separately tested processing path.
Ordinary text, tables, headers and footers. Complex structures and embedded content need explicit support.
UTF-8 documents and pasted text, with useful text and line structure preserved.
A 40-second walkthrough: add documents, choose your settings, redact, review and save.
No document upload for processing. The desktop workflow processes the source locally. This website does not accept document uploads.
No automatic handoff to AI. You choose whether to share the reviewed copy and which service or person should receive it.
No promise of anonymity. Detection can miss details, and context can reveal identity. Review the whole document, not just highlighted matches.
Offline Document Redactor is a desktop application that automatically finds sensitive information in documents, lets you review and redact or replace selected details, and saves a separate copy locally. It is designed for preparing documents before AI use or external review, starting with accounting and tax workflows.
Document redaction is designed to run on your computer. There is no automatic upload of the source or the redacted copy to an AI service. This website does not accept document uploads. Initial installation and model setup may require a download; release instructions will specify the setup requirements.
The app is in development, with macOS and Windows planned first and Linux later. Public installers are not available yet. Join the waitlist to request a launch notification; release details and system requirements will be shared when ready.
The first-release focus is text-based, unencrypted PDF, ordinary DOCX and UTF-8 TXT or pasted text in English. Scanned PDFs, mixed-image pages, protected files and complex embedded content need explicit support. Spreadsheets and standalone images are outside the initial scope.
The workflow combines local rules and entity recognition for names, organizations, contact details, addresses and structured identifiers such as tax IDs and account numbers. Country and domain choices affect coverage. Detection can miss values, so manual review and corrections remain essential.
No. Consistent placeholders can preserve useful relationships, but amounts, dates, locations and unusual events may still identify a person or organization. Review the entire document and decide whether its intended use or destination is appropriate.
The workflow creates a separate redacted copy and preserves the original. Check the exported document before sharing it. Keeping the original separate also lets you return to the source if you need to revise your decisions.
No. The initial product is a desktop document workflow that you deliberately start. It does not intercept browser uploads, prevent you from sending originals elsewhere or automatically submit documents to an AI service.
Pricing and license details have not been published. They will be provided alongside release information before a purchase or download decision.
Offline Document Redactor is in development. Join the waitlist to hear when the desktop app is ready to try.
Signup is handled by Brevo. By signing up, you request a launch notification. You can unsubscribe from our emails. How we use your email.
Form not loading? Open the signup form ↗
Linux is planned later. Public installers are not available yet; release details and system requirements will be shared when ready.