Health IT & hospital operations · Clinical data infrastructure

Eos AI

A solo-founder startup out of Y Combinator’s Winter 2026 batch, building software that stitches together a hospital’s scattered records into one timeline and then acts on what it finds. The product is live in a small number of unnamed clinics. The company is barely a year old.

A referral with a missing MRI

On the Eos AI website, the demo scenario is small and specific. A doctor tells the software, “Rivera’s referral is missing the MRI - can you find it and attach it?” The system answers back, “I’ll search the chart, attach what I find, and leave it for your review,” then walks through five steps: open the chart, search imaging records, locate the March 12 scan, attach it to the referral packet, mark it ready for review. A separate background mode reviews tomorrow’s new patients overnight and flags missing insurance information or a needed referral before anyone arrives. The company says these are demo examples built on synthetic data, not real patients.

That is the pitch in miniature: an assistant that lives inside the systems a clinic already has, does the searching and the paperwork, and hands a person the finished task instead of a score or an alert.

Data that cannot talk to itself

Eos AI’s framing of the problem, stated in its Y Combinator launch post, is that “hospitals have more data than any other industry in the world, but only 3% has been utilized to inform care decisions.” The reason, the company says, is fragmentation: a single patient’s record sits across the EHR, imaging systems, payer portals and specialty-specific software, recorded in different formats, and none of it links up. Eos’s answer is to leave the data where it is and build what it calls a centralized index over it, a compressed representation the company says lets it search and reason across scattered systems as though they were one database, then compare a new case against thousands of similar ones to predict how it will resolve.

It is a crowded idea. Health data interoperability and clinical workflow automation are categories with well-funded incumbents. Tracxn counts 386 active competitors against Eos AI, including Anterior, Tennr and Pair Team, and ranks Eos 186th among them on its scoring system.

One founder, a data-drift problem

Eos AI has a single named founder, Arya Khokhar. According to her speaker biography for the Precision Medicine World Conference, she is “building the essential data infrastructure layer for reliable healthcare AI,” driven by what she describes as “firsthand experience of seeing millions of dollars in AI research fail to reach the clinic due to data drift.” The same biography says she has “led research across medical imaging, multimodal learning, and clinical data harmonization, collaborating with academic medical centers and industry partners.”

In her own account, posted with the Y Combinator launch, she studied math and computer science at Caltech and did clinical AI and bio research at both Stanford and Caltech. “As long as I can remember, I have always wanted to be a doctor to help people,” she wrote, “and realized the best way to help as many people as possible is to build a tool that every doctor can use.” She named one collaborator by title only: “super excited to have Professor Kuo from Stanford on board as well.” No fuller identification of Professor Kuo was found in the sources reviewed for this page.

An independent write-up from The YC Tier List, a site that scores each Y Combinator batch, credits Khokhar with a research project called VERA, a medical imaging harmonization tool, done as a Y Combinator Summer Fellow before she started the W26 batch. Its verdict: “Strong researcher, uncertain operator,” noting she has “no co-founder, no industry relationships, no GTM experience” against a regulated market where hospital sales cycles are long and Epic, Oracle Health, Health Catalyst and Redox are already established.

From a battlefield application to a demo day launch

The earliest public record found for the company is a TechCrunch Startup Battlefield listing for “Eos.ai,” dated September 8, 2025, describing it as “Reimagining the way we interact with health data by building the infrastructure to unify fragmented records from hospitals, labs, and wearables” and naming Khokhar as founder at the website www.helloeos.ai. That predates the company’s Y Combinator batch by several months. By the W26 batch, which runs into spring 2026, Eos AI had a working product in what it calls early clinics, reporting internally measured results of roughly 3x productivity gains in admin workflows and 37% revenue recovery through earlier intervention. Neither figure has been independently verified, and no clinic has been named. Two data aggregators, Tracxn and PitchBook, separately record a $500,000 investment tied to the company: Tracxn labels it a Seed round dated January 1, 2026, while PitchBook labels the same amount an Accelerator/Incubator deal dated March 10, 2026 with Y Combinator as the sole listed investor. The amount matches Y Combinator’s standard investment for companies in its program, which suggests the $500,000 is the accelerator’s own check rather than a separate outside round, though neither aggregator states that explicitly. In early 2026 Khokhar also took a speaking slot at the Precision Medicine World Conference in Silicon Valley, presenting Eos AI in the event’s AI and Data Sciences Showcase.

What is proven, and what is still claimed

Read plainly, Eos AI has the markers of a very early company: a real accelerator placement, a founder with a relevant research background, a plausible problem statement, and no independently verified customer or outcome yet. The two funding figures agree on the dollar amount but disagree on what kind of check it was, and neither is confirmed by a public SEC filing.

What to watch

• Whether Eos AI names any of its early clinics or publishes an outcome study on the productivity and revenue figures it has claimed.
• A Form D or other SEC filing that would confirm the $500,000 figure and any investor beyond Y Combinator.
• Whether the company adds a co-founder or grows its headcount past one or two.
• A next round of funding, and whether it is priced or disclosed publicly.

In their words

Related companies

• Clarion: Similar focus. A New York startup building AI agents that answer the phone for clinics, then finish the task on the other end: booking, refills, referrals, billing questions, written
• Adialante: Similar focus. A mobile, per scan MRI service built to bring prostate cancer screening into urology clinics that cannot buy or house a scanner
• LunaBill: Similar focus. AI voice agents that call insurance companies on a hospital's behalf, chasing unpaid claims the way a biller would, around the clock
• Adentris: Similar focus. An AI compliance copilot that reads hospital and clinic EHRs the way a trained reviewer would, looking for documentation gaps before they turn into denied claims

Sources

1. CompanyEos AI website and platform descriptionhelloeos.ai · accessed Sep 25, 2026
2. CompanyEos AI company profileY Combinator · accessed Sep 25, 2026
3. CompanyLaunch YC: Eos AI, Autonomous OS for hospitalsY Combinator · 2026
4. IndependentEos AI batch verdictThe YC Tier List · 2026
5. IndependentEos.ai company listing, Startup BattlefieldTechCrunch · Sep 8, 2025
6. IndependentEos AI company profile and funding dataTracxn · 2026
7. IndependentEos AI company profile and funding dataPitchBook · 2026
8. IndependentArya Khokhar speaker profile and talk abstractPrecision Medicine World Conference · 2026
9. IndependentEos AI, autonomous operating system for healthcare, thread listingPortal of Robotics and Artificial Intelligence (pr.ai) · 2026
10. Public recordSEC EDGAR company search, terms Eos AI and Helloeos, Form DSEC · checked Sep 25, 2026
11. Public recordSEC EDGAR full text search, terms Eos AI and Arya KhokharSEC · checked Sep 25, 2026