Sword Health

AI-powered virtual musculoskeletal care (digital physical therapy) sold to employers, health plans, and national health systems, now a multi-vertical "AI Care" platform: pelvic health (Bloom), injury prevention (Move), mental health (Mind for enterprise; Dawn direct-to-consumer, launched Mar 2026), surgical prehab (Surgery Hero), cardiometabolic/GLP-1 support (Pulse, launched Mar 2026), and pulmonary rehab/COPD via Kaia. Core differentiator is Phoenix, an AI care specialist combining computer-vision motion tracking, real-time voice AI, and clinician oversight — plus Dawn, a proprietary mental-health foundation model trained on hundreds of thousands of hours of clinician data with the MindGuard safety-classifier suite. Scale to date: 800K+ members treated, 11M AI care sessions, 58 peer-reviewed clinical studies.

Stated: become the world's leading "AI Care" company across many clinical verticals, with an IPO around 2028 once multiple verticals reach scale (CEO calls an IPO 'boring' for now). Inferred from 2026 moves and hiring: (1) consolidate the digital MSK category via M&A (Surgery Hero, Kaia for up to $285M; Bloomberg reports a planned $500M raise for more deals); (2) diversify beyond the crowded US employer market into government-payer contracts across Europe (Portugal SNS nationwide, Greece, UK NHS, German DiGA); (3) verticalize the AI itself — build proprietary multimodal foundation models in-house (Dawn for mental health; active hiring of multimodal post-training research scientists and clinical video annotators) so the margin engine is owned, not rented; (4) open a direct-to-consumer channel (Dawn) that bypasses employer benefits consolidation and monetizes the AI directly.

Proprietary longitudinal motion + outcomes dataset from 11M+ AI-guided care sessions across 800K+ members (markerless computer vision plus Kaia's sensor biofeedback), now being converted into owned foundation models — Dawn is trained on hundreds of thousands of hours of clinician data, and a clinician-staffed video-annotation pipeline keeps enriching training data. Phoenix/Dawn lower cost-per-session versus human-therapist rivals; MindGuard safety classifiers and 58 peer-reviewed studies create regulatory and trust barriers; three new US AI patents (personalized communications/recommendations, CV-driven full-body motion tracking); deep enterprise and national-government contracts (~95-100% claimed client retention) and outcomes-based pricing that newcomers cannot easily underwrite.