The a16z Show Deploying AI in Healthcare
406 snips
Mar 4, 2026 
Nikhil Buduma, CEO and co-founder of Ambience Healthcare and former Stanford deep learning PhD, talks about applying AI to clinical workflows. He recounts running a medical practice to learn real-world EHR pain points. Conversations cover building EHR-agnostic layers, clinician adoption at academic centers, rapid prototyping, and how AI product and data architectures must evolve to deliver CFO-grade ROI.
AI Snips
Chapters
Transcript
Episode notes
Lock In Deep Health System Partnerships For Fast Iteration
- Build deep, long‑term partnerships with health systems to move from prototype to production and enable live learning loops in under 30 days.
- Rapid in‑system iteration and trust are required to deploy and learn on real clinical workflows.
A New AI‑First Layer Can Disrupt The EHR Stack
- Rebuilding the EHR stack is plausible now because Ambience built a layer that normalizes EHR data and makes AI product development dramatically cheaper.
- That infrastructure took years of R&D but lowers marginal cost to ship many new AI use cases.
Point Of Care Intelligence Can Disrupt Revenue Cycle
- Embedding clinical and RCM expertise into models can change operating margin by making the right coding/billing actions obvious at point of care.
- Real ROI combines improved coding accuracy, throughput, and reduced back‑office fixes.