CamoText: Offline anonymizer

CamoText was born from my firsthand experience as a practicing lawyer with the challenges of maintaining client confidentiality while leveraging generative AI and LLMs for text analysis, research, and form generation. It's essential to protect sensitive text before it's absorbed and potentially regurgitated by these models. Every existing anonymization solution I found either required external API calls or other cloud reliance of some kind (third party access to sensitive text, again). This posed extensive privacy risks and privilege issues—a new tool was necessary to make local text anonymization fast, intuitive, and secure.

So, I built CamoText to:

- be private and compliant by design, with zero external dependencies or connectivity, and zero user data retained upon close
- ensure its anonymizations were irreversible by third parties
- save time and resources, and be useful even if the auto-detection feature is subjectively inaccurate
- perform well on an average laptop
- have a familiar user interface and experience, and
- ensure a human-in-the-loop can easily review and anonymize text after the auto-detection or revert false positives.

All packaged in a pay-once, use-forever desktop app. Customers receive free updates (new versions of the software) for the year following their purchase, and custom text recognizers and other features are available for enterprise. Let me know what you think!