AI Takes Center Stage in Healthcare: Agents, Robots, and $3.7B in Investment
From AI chatbot clinical trials to robotic surgery and record funding, this week's health AI news signals a major inflection point.
By Dr. Asher Knippel
Health AI had an unusually dense week. A landmark clinical trial, conference data showing AI doing real diagnostic work, a funding round for surgical robotics, a new industry spending report, and a live deployment of autonomous AI agents in hospital administration — each story on its own would be notable; together they sketch an industry crossing a threshold.
Apple and OpenEvidence launch the first large-scale trial of AI health chatbots
Most evidence for AI-assisted health care comes from retrospective studies or narrow pilots. Apple and OpenEvidence are trying to change that. The two companies announced a 200,000-person prospective study designed to measure whether conversational AI actually changes what patients do — whether they seek care sooner, follow medication regimens more closely, or make different lifestyle decisions after an AI-assisted interaction.
Participants use a dedicated app built on OpenEvidence's clinical AI platform, which is trained on peer-reviewed medical literature and designed to give evidence-based answers rather than generic health information. The study will track outcomes over time and is intended to generate the kind of randomized, longitudinal data that regulators and health systems need before making AI-assisted guidance standard of care.
The scale matters. At 200,000 participants, this is large enough to detect small but meaningful effects that smaller pilots would miss. It is also notable that Apple is running the study under its Research app infrastructure, which has previously supported heart and hearing health studies, suggesting the company sees health AI as a long-term platform investment rather than a one-off product feature.
AI trims unnecessary genetic testing in lung cancer by a third
Rapid EGFR mutation testing is a critical step in deciding whether a lung cancer patient can benefit from targeted therapies. It is also expensive and time-sensitive, which creates pressure to order it broadly even when pathology findings would not support it. New data presented at the Association for Molecular Pathology (AMP) 2026 annual meeting suggest AI can help.
Researchers reported that an AI-assisted triage system reduced unnecessary rapid EGFR orders by approximately one-third, by identifying cases where tumor cell content or tissue quality made the test unlikely to yield an actionable result. A second AMP study examined AI-assisted review of germline cancer genetics reports and found it reduced error rates in final sign-out, a step where mistakes carry direct consequences for patients and families.
Both findings land at the same conceptual point: AI is most useful in molecular diagnostics not when it replaces pathologist judgment but when it filters the caseload before that judgment is applied. The economic implications are significant — EGFR testing costs hundreds of dollars per case, and lung cancer is among the most commonly diagnosed malignancies globally.
Remedy Robotics raises $62 million to bring AI-guided endovascular surgery anywhere
Cardiovascular interventions like stent placement and thrombectomy require specialized expertise concentrated in major medical centers. Remedy Robotics is building a robotic platform intended to close that gap. The company announced a $62 million funding round to advance its AI-powered endovascular robot toward a pivotal clinical trial and eventual FDA clearance.
The system is designed to perform procedures on-site under physician supervision, with the longer-term capability to operate remotely or with increasing autonomy as regulatory frameworks catch up. Remote operation could allow an interventional cardiologist in one city to guide a procedure at a hospital hours away — expanding access to time-critical cardiovascular care in rural or underserved settings.
Endovascular robotics is a competitive space, but Remedy's emphasis on AI-guided navigation rather than purely mechanical assistance marks a distinct approach. The $62 million round positions the company to run the trials needed to demonstrate safety and efficacy at scale.
Healthcare AI spending hits $3.7 billion — and the mix is changing
Menlo Ventures released its 2026 State of AI in Healthcare report this week, and the headline figure is striking: healthcare AI spending has grown 2.5 times year-over-year to $3.7 billion. But the composition of that spending may be more important than the total.
Earlier AI healthcare investment was dominated by administrative automation — coding, prior authorization, documentation. That category still exists, but the report shows growth shifting decisively toward clinical applications: drug discovery platforms, clinical decision support, and patient-facing tools. Life-sciences AI, which covers drug development and biomarker discovery, grew 23-fold in a single year, suggesting that the pharmaceutical industry's bet on AI for research and development is now flowing through to investment.
Clinical copilots — AI tools that assist physicians at the point of care — lead the growth in direct-care applications, followed by tools for population health management and diagnostic support. The pattern suggests that health systems are moving past experimentation and toward procurement of AI tools with defined clinical roles.
OpenAI deploys autonomous agents with CVS Health and Cigna to cut prior authorization friction
Of all the administrative burdens in US healthcare, prior authorization may be the most disliked. Hospitals, physician groups, and patients cite it as a source of delay and frustration, and it is estimated to cost the industry billions in staff hours annually. OpenAI announced this week that it is deploying computer-use agents — AI systems that can navigate software interfaces autonomously — with CVS Health and Cigna specifically to handle referrals and prior authorizations.
The system, part of OpenAI's ChatGPT for Healthcare platform, can open electronic health record interfaces, extract relevant clinical information, complete payer portal forms, and submit requests without requiring a human to execute each step. Early partners also include Abridge and Ambience Healthcare, two companies already focused on AI-assisted clinical documentation.
Computer-use agents are a qualitatively different kind of AI deployment from a chatbot or classification model. They interact with software systems as a human would, which means they can operate across the fragmented mix of EHRs and payer portals that characterizes US healthcare infrastructure without requiring custom integrations. That flexibility may be what allows the approach to scale where previous automation efforts have stalled.