The engineer who taught your iPhone to recognize your face now wants to teach AI to recognize when your brain is broken.
The Summary
- Gidi Littwin, a core inventor of Apple's FaceID, launched Hemispheric to build AI-powered diagnostic brain scans for depression, PTSD, and Parkinson's
- The company's goal: make brain diagnostics as accessible and affordable as a blood test, not a $3,000 MRI
- This is consumer hardware expertise meeting foundation models for medical diagnosis, a pattern we'll see more of as agents move from screens into bodies
The Signal
Littwin spent years making Face ID work on hundreds of millions of devices. The technical challenge there was building a system that could authenticate identity across billions of facial variations, in different lighting, from different angles, fast enough that people wouldn't notice the computational load. Now he's applying that same pattern recognition discipline to brain scans, except instead of unlocking phones, he's trying to unlock early diagnosis for neurological and psychiatric conditions.
The business model assumption here is bold: most people who need brain diagnostics never get them because the friction is too high. MRIs are expensive, require specialist interpretation, and involve waiting weeks for appointments. Blood tests, by contrast, happen at CVS. You walk in, they stick you, you get results. Hemispheric is betting there's a massive unmet market of people who would get brain health screening if it were as simple as rolling up a sleeve.
"The engineer who taught your iPhone to recognize your face now wants to teach AI to recognize when your brain is broken."
The technical lift here is a frontier AI model trained on brain imaging data. Not a fine-tuned GPT wrapper. A purpose-built model that can look at whatever imaging modality Hemispheric uses (the article doesn't specify if it's EEG, fMRI, or something cheaper) and flag patterns associated with specific conditions. This is where the FaceID experience matters. Apple didn't just build a neural network, they built a system that had to work in the real world, on constrained hardware, with zero tolerance for failure. If Littwin can bring that same engineering rigor to medical AI, Hemispheric could actually ship something clinicians trust.
The implications:
- If brain diagnostics become as routine as cholesterol checks, we're looking at a category expansion, not just product iteration
- Mental health and neurodegenerative diseases are massively underdiagnosed because current diagnostic tools are slow and subjective
- For AI agents to truly augment human work, humans need to be functioning at baseline, which means accessible preventive brain health care might be infrastructure, not luxury
The question is whether Hemispheric can navigate FDA approval, insurance reimbursement, and clinical validation faster than the dozen other AI health startups trying to do the same thing in different organ systems. Littwin's edge is consumer product intuition. If he can make this feel like using Face ID instead of visiting a neurologist, he wins.