
Contracts, pleadings, judgments...
Before a contract goes to an external model, the app expects names, case numbers, and account numbers to be handled on the computer. Recognition covers eighteen kinds of sensitive items, including names, organizations, addresses, amounts, ID numbers, accounts, and seals. Review can delete mistakes, add misses, and check with search. Files in one project share aliases, later files reuse those rules, and progress sits in a local memory file. Export can keep the word-processor or PDF layout, or produce structured text for a model and restore names with authorization. There is no network module, and the recognition model ships with the installer. It fits law firms and in-house counsel preparing judgments, contracts, and evidence screenshots before a cloud model.
Lawbot Desensitization Cat, from Shenzhen Buyumao Technology, is a Windows and macOS app for legal documents. Contracts, pleadings, judgments, letters, chat screenshots, and payment records are scanned for sensitive information on the computer. After manual fixes, added items, and allow or block lists, you export aliases, asterisks, or masks. The same party keeps the same alias across a project, and progress is stored in a local memory file. Scans use a built-in recognizer. Recognition and export do not upload the file. Structured text can go to an external model, then aliases can be restored with authorization. The personal edition is a per-device purchase. Teams buy seats separately.
Difficulty: Intermediate
Detect sensitive items locally
Mark names, companies and short names, addresses, amounts, ID numbers, phones, credit codes, accounts, case numbers, email, plates, seals, URLs, and faces on the device.
Alias, mask, and review
Delete false marks, add misses, use allow and block lists, and choose aliases, asterisks, or masks. One party keeps one alias inside the project.
Project memory for aliases
Several documents share alias rules. Added files, a mask fix, or a pause resume from the local memory file.
Structured export and restore
Keep the original layout or export structured text for summary, translation, and risk review, then restore aliases when authorized.
Offline recognition for scans
Text, scanned, and mixed PDFs are supported, as are chat screenshots and payment images. Scans are recognized on the device by the bundled engine.
Redact a contract before a model
Counsel imports an investment, loan, or confidentiality agreement, applies consistent aliases, and exports structured text for a summary or risk list.
Batching court documents
Put judgments, rulings, and mediation papers in one project so party aliases stay consistent, then archive the masked copies.
Masking evidence screenshots
Mask or alias chat logs, transfer receipts, and invoice images so faces, chat IDs, and account numbers are not sent as-is.
Pricing may change. Please refer to the official website for the latest pricing details.
The site says it is desktop software with no network module, that it blocks outbound requests, and that the detection and text-recognition models download with the installer. Import, review, export, and restore stay local. If you later send the structured text to an external model, that step is outside this app. The app's own processing does not upload the original.
Judgments, contracts, and added evidence in one project share aliases. A newly added file reuses those rules, and a mask change can be re-exported. Progress, aliases, and allow or block lists live in a local memory file you can reopen. The mapping can be checked later. A different project does not automatically reuse the same aliases.
The page lists word-processor documents, PDF, and common images. PDFs may be text, scanned, or mixed. Material types include judgments, litigation papers, contracts, company filings, letters, and chat or transfer screenshots. Import routes each type to a recognition strategy, and you can change the class by hand. Scan speed depends on memory and page count and is slower on modest computers.
Real reviews and feedback from users