The pitch sounds almost too convenient: automate the most tedious parts of value-based care without forcing doctors to learn new software, adopt new workflows, or even log into a separate dashboard. Just install the agents inside your existing electronic health record system and let them handle the rest.
That's what Beacon Health is offering, anyway. The New York startup, which participated in Y Combinator's Winter 2026 batch, went live with its first deployment this February at an independent physician association managing care for 40,000 patients. Two founders. One customer so far. And a technology they say can autonomously close quality measure gaps, surface missed risk-adjustment codes, and manage patient outreach—all without a physician lifting a finger.
Whether that promise holds up beyond a single IPA deployment is another question entirely.
Agents That Live Where Doctors Actually Work
Beacon's core bet is deceptively simple: physicians won't adopt another tool. They're already drowning in alerts, portals, and administrative systems that fragment their workflows. So instead of asking doctors to come to the software, Beacon built software that goes to them.
The company's AI agents operate directly inside EHR platforms, learning to navigate the systems "like a human would," according to the startup's materials. That means no new logins. No parallel dashboards. The agents identify patients overdue for preventative screenings, manage outreach campaigns, queue up lab orders, and document results—all within the EHR environment clinicians already use daily.
They also handle the unglamorous work that value-based contracts demand: after-hours patient support, prior authorization requests, referral coordination, follow-up triage. For practices bearing financial risk under capitated contracts, the agents hunt for missing hierarchical condition category codes—the diagnosis documentation that directly affects how much a practice gets paid per patient.
It's administrative automation aimed squarely at the pain points of value-based care, packaged under what Beacon calls "Value-Based Care On Autopilot." The company structures its pricing around success-based fees, though it hasn't disclosed specific commercial terms.
A Single Customer, So Far
Beacon's first live deployment appears to be with MetroWest Healthcare Alliance, a Massachusetts IPA participating in the Tufts Medicine Integrated Network. MWHCA manages care for roughly 40,000 lives—matching the patient count Beacon cited in February when announcing its launch. The startup highlights MWHCA's involvement on its website, though no independent performance data has surfaced publicly.
The company says it's "trusted by physicians across the country" and displays several practice logos beyond MetroWest on its site. Team size remains small. Y Combinator's profile lists two employees as of March, while Beacon's LinkedIn suggests somewhere between two and ten. Either way, it's a lean operation for software claiming to automate workflows across multiple health systems.
Beacon also asserts HIPAA compliance, though like most early-stage health tech companies, detailed security documentation isn't readily available to outsiders.
Two Founders With Adjacent Experience
CEO Mark Pothen didn't stumble into primary care by accident. He grew up watching his mother run her own practice, absorbing the operational chaos that comes with it. Before launching Beacon, he spent six months embedded inside another clinic—ostensibly doing operations work, but really testing whether AI could handle the grunt work physicians hate most. Prior to that, he held product and go-to-market roles at earlier-stage health tech startups.
His co-founder, Obinna Akahara, studied physics at UT Austin before building production AI systems at LinkedIn and Conversion.ai (the company that later rebranded as Jasper). He serves as Beacon's CTO, responsible for the technical architecture that allows these agents to operate autonomously inside third-party EHR platforms without breaking things.
The two founded Beacon in 2025, then joined Y Combinator's Winter 2026 cohort. For now, they remain a two-person team—or close to it.
A Crowded Space Getting More Crowded
Beacon is hardly alone in betting that AI agents can profitably absorb healthcare's administrative mess. The market has moved fast, perhaps faster than many expected.
Innovaccer, already a significant player in population health management, unveiled its Agents of Care platform in February 2025. Notable Health has been automating faxes and phone workflows—tedious tasks, but ones that consume surprising amounts of staff time. Even the EHR vendors themselves are building agent capabilities. Greenway Health launched something it calls an "Agentic AI Factory" in January 2026. Oracle's Clinical AI Agent posted measurable time savings in deployments throughout 2024.
Meanwhile, the underlying demand for value-based care enablement continues growing. Aledade—one of the larger networks helping independent practices navigate risk-based contracts—added more than 700 primary care organizations for the 2026 performance year alone. That's a lot of small practices entering arrangements where incomplete coding or missed quality metrics directly hit their bottom line.
What sets Beacon apart, at least in theory, is its narrow focus. It's not trying to be a comprehensive clinical documentation tool or a patient engagement platform. It targets the specific administrative workflows that value-based contracts require: quality reporting, risk adjustment, care gap closure. The customers it's after—IPAs, ACOs, independent practices bearing financial risk—are organizations that lose real money when these tasks fall through the cracks.
But focus cuts both ways. A narrow product might be easier to sell and deploy, or it might struggle to justify its cost when larger platforms bundle similar features into broader suites.
What We Don't Know Yet
Beacon hasn't disclosed which EHR systems its agents support. That's not a trivial detail. Integration depth varies wildly across platforms, and some vendors are more permissive than others when it comes to third-party automation. The company also hasn't specified what success metrics it guarantees under performance-based pricing, or how it defines "success" in the first place. Improved quality scores? Higher risk-adjustment revenue? Reduced staff hours?
Scaling beyond a single 40,000-patient IPA will require answering those questions—and proving the technology works across different EHR environments, payer contracts, and practice types. For now, Beacon represents another wager that autonomous software can relieve the tedious burdens of value-based care, and that small independent practices will pay for that relief.
Perhaps they will. The administrative load is real, and it's getting heavier as more practices enter risk arrangements. Whether two people in New York can build the solution remains very much an open question.
Practices interested in exploring Beacon can contact CEO Mark Pothen at [email protected].
