YashPatel

There are about 14 psychiatrists for every 100,000 Americans. The wait time for a first appointment averages weeks. Insurance reimbursements have pushed most providers out-of-network. Yash Patel sat across this problem, did the math, and built a different system entirely.

In April 2026, Legion Health - the company Patel co-founded with his Princeton roommates Arthur MacWaters and Daniel Wilson - became the first startup in US history to receive regulatory authorization allowing AI to prescribe psychiatric medications. The state of Utah approved a mitigation agreement letting Legion Health's AI system assess stable patients and renew prescriptions for select maintenance medications, including fluoxetine and sertraline. The kind of quiet, paperwork-heavy renewal that takes up a significant share of every psychiatrist's day - now handled autonomously, covered by insurance, and available to the 700,000-plus Utahns living in designated mental health shortage areas.

That is not a small thing. Prescription authority for AI in medicine has been the third rail of health regulation for years. Patel walked up to it anyway.

The "people we love" line is not marketing copy. Patel's father was diagnosed with glioblastoma - a fast-moving, aggressive brain tumor - and Yash found himself navigating the American healthcare system from the patient's side for the first time. He already knew how it worked from the outside: as a healthcare analyst at the Congressional Budget Office, he had spent years analyzing Medicare and Medicaid payment policy, understanding which incentives were broken and why. Watching his father try to access care told him something no spreadsheet had: the system is not just inefficient. It is actively disorienting for the people who need it most.

So he went back to school - or rather, he went to Harvard Medical School as a research assistant, added clinical context to his policy fluency, and started building. In 2021, he and his co-founders - three Princeton graduates who had been friends for a dozen years before becoming business partners - entered Y Combinator's Summer batch. They won the Audience Choice Award at the Princeton-Harvard Startup Showcase that same year. And they started from a deceptively simple premise: insurance-covered psychiatric care should be available to anyone with a phone.

Legion Health launched as a marketplace connecting patients to board-certified psychiatrists, initially focused on Texas - a state with one of the country's largest unmet mental health needs. The company accepted insurance from the start, deliberately targeting the gap between people who need help and people who can afford boutique telehealth. Early rounds raised $2M in 2022, backed by Next Coast Ventures, Soma Capital, UpHonest Capital, Acequia Capital, and a roster of angel investors that included Erica Johnson, co-founder of Modern Health, and Ravi Shah, chief innovation officer of Columbia Psychiatry.

The pivot - or, more precisely, the evolution - came in 2024. Legion Health raised a $6.3M round led by Y Combinator and relaunched as an explicitly AI-native platform. Not AI-assisted. AI-native. The distinction matters. The platform automates 95% of the administrative work that currently consumes clinical time: scheduling, documentation, intake, prior authorization, referral management. Psychiatrists on the platform are left with what they were trained to do - evaluate, diagnose, and treat patients - instead of burning hours on paperwork that no one should be doing manually in 2026.

The clinical staff are board-certified. The insurance coverage is real. The AI handles the overhead. And now, for a defined subset of stable maintenance patients, the AI also handles renewal prescriptions - with pharmacist review, monthly regulatory reporting, and a hard stop: it cannot initiate new prescriptions, only renew those already established by a human clinician.

Patel studied pure mathematics at Princeton, graduating in 2018 with certificates in Applications of Computing, Statistics, and Machine Learning. It is the kind of combination that produces either quants or founders - people who see the world as a set of constraints to be optimized. Patel added a third dimension: policy. His years at the Congressional Budget Office gave him a working knowledge of how federal healthcare payment systems actually function - and how deeply the incentive structures embedded in Medicare and Medicaid shape what care gets delivered, to whom, at what price.

Arthur MacWaters (Co-Founder, President) and Daniel Wilson (Co-Founder, CTO) completed the team. Three Princeton graduates, twelve years of friendship, one company. Founders who have known each other that long skip a particular kind of risk: the one where incompatible work styles or undisclosed beliefs blow up a company in year two. They had already seen each other under pressure.

Patel also has a Google Scholar profile - suggesting a research thread that runs alongside his entrepreneurial one, a habit of treating the literature as something worth engaging rather than just citing in pitch decks. The platform's technology stack reflects the same instinct: Anthropic Claude, OpenAI, LangChain, and vector databases alongside a modern web stack (Next.js, TypeScript, Supabase, Tailwind). This is not a company that outsources its AI thinking to a single API call.

The Utah regulatory milestone, which came in April 2026, deserves its own paragraph. Regulators in most states have been cautious - sometimes rightly so - about AI in clinical settings. Utah's mitigation agreement with Legion Health is narrow by design: stable patients, maintenance medications only, no psychiatric hospitalization in the past year, monthly reporting, pharmacist involvement. But the regulatory principle it establishes is not narrow at all. For the first time, a US state has formally agreed that AI can participate in the prescribing loop for psychiatric medications. Legion Health plans to expand this model to all 50 states by the end of 2026.

The opponents of this will argue, correctly, that AI in prescribing requires caution. Brent Kious at the University of Utah School of Medicine called for greater transparency and more rigorous testing. That is a reasonable position. Patel's counter-position - implicit in the company's regulatory approach - is that the current system's failures have a body count too, and that rigorous AI oversight is still better than six-week waits for people in crisis.

Twenty-seven employees. A seed-stage company. And a regulatory authorization that every major health system in the country will be watching carefully. That is the current state of Legion Health in May 2026.