Dual AI Architecture: On-Device Apple Intelligence Meets Claude's Cloud

Apple's on-device AI is excellent at synthesis. Claude's cloud AI is excellent at analysis. The trick is connecting them without exposing patient data.

I'm a solo founder who built HerDiabetes, a new category of diabetes management app designed for women, from concept to App Store in 6 months. The app connects menstrual cycle data with glucose patterns, something no other diabetes app does. Along the way I had to solve a hard problem: how do you use cloud AI in a health app without sending personal health information off the device?

The answer is a dual AI architecture. Apple Intelligence runs on-device, synthesizing raw health data into a clean, PHI-free narrative. Claude + PubMed MCP picks it up from there, delivering research-backed insights with real citations. The user stays in control at every step. Privacy isn't a constraint. It's the architecture.

This session includes a live demo of the working app and a walkthrough of the technical decisions behind it. I'll also cover the founder story, how a cold email to a diabetes tech conference led to three clinical advisors and an invitation to present at D-Data ExChange in June 2026.

• All levels
• Development
• Startups
• AI/ML