(Portfolio Profile) Onix
Summary
Onyx founders Nicholas Nato and David Benahum discuss their AI platform, which creates custom models trained on world-class experts' private and public knowledge to assist users with health and wellness challenges. They emphasize high-fidelity, privacy-focused interactions using small language models on-device, avoiding the data scraping practices of larger AI firms. The episode covers their revenue-sharing business model, the importance of expert involvement in training, and how users can assemble personalized care teams for complex issues.
Topics discussed
Intro, sponsors, and guest welcome
Onyx overview: AI experts for health and wellness
App access, concierge, and expert landing pages
Training philosophy: fidelity and expert involvement
Domain-specific models vs. generic frontier models
Using 'dark data' to capture expert nuance
Expertise personified: facts, judgment, and taste
Revenue sharing and multi-expert group chats
Real-world use cases: parenting and nutrition
Why health and wellness is the first vertical
Privacy, encryption, and data sovereignty
Business model alignment and data rights
Nick's background: robotics and AI engineering
David's background: tech, media, and blockchain
Company origins and investor partnership
Startup lessons: de-risking and investor relations
The moat: trust and domain-specific efficiency
AI parallels to the 90s internet and closing
Listen ad-free on Castria