Founder here. I built this because many everyday notices are hard to understand: utility bills, medical bills, insurance letters, landlord notices, government letters, etc.

The app extracts a few practical fields from an uploaded PDF/image: who sent it, who it is for, deadlines, amount owed, what it seems to mean, and what should be checked against the original.

I know this is a sensitive area because people are uploading private documents. Current design choices: - users control retention/deletion - summaries are explicitly not legal/financial advice - failed or low-confidence model output should fail visibly rather than show fake certainty - duplicate uploads reuse prior successful results without reprocessing

The technical parts I’ve been working through are queue reliability, provider failures, document retention, quota accounting, and making the UI avoid “confident nonsense” when extraction fails.

I’m the builder and can answer implementation/privacy questions.