Observed arrival · 2026-11-19
Fragmós puts a fence between sensitive data and AI
Fragmós presents a browser-based tool that replaces names, IDs, contacts, and secrets with tokens before sending a prompt to an AI model.
- For
- Educators and clinicians handling sensitive notes
- Worth noticing
- Its wire view is presented as a byte-for-byte preview of the payload sent to the model.
Field notes
The demo walks through detection, token replacement, sending, and restoration, with controls to inspect highlights or select missed words. Sample inputs include an IEP meeting note, a clinical intake, and a leaked API configuration; model choices shown are Gemini, Claude, and GPT. The site says its original-value vault exists only in tab memory and that the outgoing payload can be inspected, but those privacy claims are not independently verified by the page extract.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue