I built SafeDOM.ai to generate AI prompt context directly from real UI state while automatically redacting PII.

It uses data-ai annotations to decide what to include, exclude, or redact, and produces structured fields, rawText with placeholders, and a redactions list. PII (email, phone, IBAN, credit card, SSN) is removed before sending data to a model, and placeholders can be safely reinjected on the backend.

Demo: https://jennifer-ha.github.io/SafeDOM.ai — Repo: https://github.com/jennifer-ha/SafeDOM.ai