AI Doctor Notes – Private doctor visit notes app for patients
On-device AI processing keeps sensitive medical data off the cloud.

Yet another AI document summarizer in a crowded space.
Consumers dealing with complex bills and official notices
Jina AI · Humata · ChatPDF
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.
On-device AI processing keeps sensitive medical data off the cloud.
Transcript-first chat beats one-shot summaries from Eightify and SummarizeTech.
AI receipt scanning + auto-generated warranty claim letters citing local law—solves real consumer friction.
Blinkist for YouTube—clean SaaS, but Opus, NotebookLM, and dozens of AI summarizers already do this.
Native protobuf storage in Postgres with 65% savings over JSONB.
Student-built extraction API competing directly with established players like LlamaParse.