A self-learning sensitive data detection and redaction engine combining deterministic rules (regex) with LLM-based evaluation and feedback loops.
Auto-Dedact is intentionally named with dual meaning:
- Automated Detect & Redact — the system automatically detects sensitive data and redacts it using deterministic rules and learned patterns.
- A deliberate sound-pun on autodidact — reflecting that the engine learns by itself through LLM-driven feedback loops, validation, and rule refinement.
These screenshots reflect early development runs executed locally. Output is intentionally raw to demonstrate decision logic and failure handling.
