Nvidia is partnering with Abridge, the startup behind an AI note-taking app for doctors, to train an artificial intelligence model specifically designed for clinical conversations, according to the Wall Street Journal.
The two companies are building the model using de-identified patient data and Nvidia's Nemotron models, the Journal reported. The collaboration signals a push to move beyond general-purpose AI and create tools tuned to the particular demands of medical dialogue — including the precise, high-stakes language doctors and patients use during appointments.
Abridge, meanwhile, is broadening its footprint beyond clinical documentation. According to Fierce Healthcare, the company has also secured a strategic investment from pharmaceutical giant Eli Lilly and is expanding into payer and research workflows — suggesting ambitions well beyond its original focus on helping physicians take notes.
Stat News confirmed that Abridge has now inked deals with both Nvidia and Lilly, a notable double-announcement for a startup operating in the heavily regulated healthcare AI space.
The story matters because healthcare remains one of the most challenging frontiers for AI adoption — patient data is sensitive, stakes are high, and general models often stumble on clinical nuance. A purpose-built model trained on real (though anonymized) doctor-patient conversations could meaningfully close that gap, and the combination of Nvidia's chip and model infrastructure with Lilly's pharmaceutical reach suggests the industry is betting heavily that specialized medical AI is coming fast.