About the Sword AI Research Team
The Sword AI Research Team brings together AI researchers and clinical scientists working on one of the most important problems AI can take on: delivering safe, high-quality healthcare to everyone who needs it. Most of our researchers hold PhDs and bring years of experience across reinforcement learning, computer vision, speech, mathematics, large language models, and clinical research. We train large-scale models and build the agents that power Sword’s AI Care, supporting every step of the patient journey, from triage and waitlists to prehab, rehab, and daily care across musculoskeletal, women’s, mental, and cardiometabolic health.
We publish regularly at leading AI venues, including NeurIPS, CVPR, COLM, and the *CL conferences, as well as in peer-reviewed clinical journals. Publishing in both worlds lets us connect technical advances to the clinical outcomes they are meant to improve.
What makes our approach different
Clinicians aren’t reviewers here. They’re builders. They join our team from day one, defining what safe care means and designing the evaluations our models must pass.
We grade our models on whether the care holds up. Real care unfolds over weeks and months, so we evaluate consistency over time, not just one good answer.
We build safety systems tailored to each clinical domain. Pregnancy, anxiety, and injury treatment each carry their own risks, and our guardrails reflect that.
Open research for more human care
At Sword, we’re building AI to heal billions and unlock humanity’s potential. We believe, perhaps paradoxically, that AI will make healthcare more human, freeing clinicians to focus where they matter most and giving people access to world-class care without waiting weeks for an appointment. Because no company can get there alone, we share our work openly, with a focus on evaluation and safety, to help establish frameworks that fairly measure clinical AI and earn the trust of patients, clinicians, and healthcare organizations.
