Science
How we read carefully.
A short note on the technical approach, what grounds it clinically, and how we treat the data you trust us with. We'll publish in more detail closer to launch.
01 · Approach
Multi-modal reasoning.
Our reading layer is a system of cooperating models — each trained on a single modality (blood, imaging, genetic, microbiome), each grounded in clinical reference libraries, each able to pass findings to the others. A pattern in one result is only as informative as the context the other results provide.
02 · Clinical grounding
Reviewed by practicing physicians.
Every model update is reviewed by board-certified physicians before release. Reasoning is grounded in established clinical reference libraries — UpToDate, NCCN, ACC, ADA, USPSTF, AAFP — and aligned with U.S. guidelines where applicable. We do not generate diagnoses. We produce a structured read that you and your clinician can work from together.
03 · Data
Anonymized, never sold.
Your results stay on encrypted infrastructure we operate. We don't sell data to advertisers, we don't share it with pharmaceutical companies, and we don't use it to train models you didn't opt into. Our U.S. pipeline launches HIPAA-compliant and independently audited.
Questions