Sefton has joined the NVIDIA Inception Program

Sefton is now a member of the NVIDIA Inception Program. Inception is NVIDIA’s program for startups building with AI, and membership gives us access to their developer resources, training, and the GPU tooling we need as we build out Sefton’s real-time pipeline.

Sefton is an operating system for allied health. The platform is live and we are onboarding our first practice. The part we are still building is the hard part, a single pipeline that interprets a therapy session in real time across languages, writes the clinical note from that same session, and suggests the ICD-10 and CPT codes with a confidence score the clinician confirms. Nothing gets coded or billed on its own.

Real-time speech and translation in a clinical setting is a compute problem and an accuracy problem at the same time, because general models are inaccurate on clinical terms and disordered speech and they are often confidently wrong. We are evaluating NVIDIA Riva for the speech layer and NeMo for adapting models to clinical language, and Inception gives us a shorter path between what we are testing and what we can ship.

The membership does not change how we work. The AI assists, the clinician decides, and we build to a standard where done means tested. It gives us better tools to get there, and I think that shows up fastest in the thing families actually feel, which is being understood in their own language during a visit.