The operating system for clinical research.
It's a Tuesday in San Francisco, and somewhere in a quiet Medeloop office, an AI agent is doing in nineteen minutes what a research coordinator used to do in six months. Nobody applauds. That is the point.
Most healthcare-AI demos sparkle. They promise miracles, then deliver a chatbot. Medeloop is doing something less photogenic and more useful: it is digesting the unglamorous middle of medical research - the data harmonization, the biostatistics, the IRB documentation, the grant boilerplate - and turning it into software that runs on its own.
The company calls this "the operating system for clinical research." Stripped of marketing, it is an agentic stack that touches every step from "I have a hunch" to "I have a peer-reviewed conclusion." Grant discovery? Agentic. Cohort identification across 300 million patient records? Agentic. Drafting the methods section? Also agentic. The biostatistician, for the first time in a generation, is not the bottleneck.
It started, as the most stubborn companies do, with a personal problem. Dr. Rene Caissie is a former surgeon and the founder of Medesync, an EMR company sold to Telus. His daughter has Complex Regional Pain Syndrome - a condition with limited treatment options and a research pipeline that moves at the speed of paperwork. Caissie noticed that the slowest, dumbest part of medicine wasn't the science. It was the process around the science. He left the operating room and started writing software.
Medeloop sells four products that look distinct on a pricing page and behave as one on a researcher's screen. Each module replaces a discrete kind of misery.
Plug in your hospital's data, then ask questions in English. Agents translate to SQL, run the query, return the chart, and footnote their work. The data scientist queue gets shorter.
For studies that need population-level scale. Pair internal records with claims data from a national footprint, and design outcomes research without negotiating eight data agreements first.
Surfaces relevant funding opportunities and helps draft proposals. Researchers report grant-writing season feels less like tax season.
Population health and care management tools that surface research-grade intelligence inside the clinical workflow - where the patient actually is.
Clinical research, broken down by where the calendar actually disappears. Medeloop's pitch is that the first three categories should not eat the last one.
Approximate distribution per industry interviews. Medeloop targets the top three rows.
AI is the key to solving healthcare's most pressing research challenges. - Medeloop manifesto
A surgeon, a Stanford AI researcher, a geneticist, and a special operations veteran walk into a clinical research startup. None of them blink.
Academic medicine is famously skeptical of vendors. The institutions willing to put their data through Medeloop suggest the company has cleared a high bar on compliance, IRB, and procurement.
Announces collaboration with Stanford Health Care - a deeper clinical footprint on top of the existing academic relationship.
Closes $15.5M Series A led by Inovia Capital with Icon Ventures, General Catalyst, and Maven Ventures. Total funding crosses $23.5M.
Hires a CTO from Innovaccer to scale the agentic platform; announces strategic research collaboration with Mila in Montreal.
Raises an $8M seed round. Begins onboarding academic medical center pilots.
Founded in San Francisco. Roots in Stanford's CS and medical communities.
Caissie was a practicing surgeon and Stanford-affiliated faculty before he traded the OR for a Git repo. He still teaches a digital health entrepreneurship class at Stanford.
Caissie's daughter has Complex Regional Pain Syndrome. The slow grind of rare-disease research is not a thesis to him - it is a household reality.
His previous EMR company, Medesync, sold to Telus. That exit is funding several of the calmer decisions you can feel in the product.
Healthcare-AI is crowded, but the operating-system-for-research framing has fewer takers than it sounds. Adjacent players include Komodo Health and Truveta on the real-world evidence side, Datavant on data plumbing, TriNetX on cohort discovery, Atropos Health on point-of-care evidence, and a horde of vertical research copilots. Medeloop's bet is that owning the workflow end-to-end - not just one bullet - is the durable position.
Return to that quiet office in San Francisco. The same Tuesday. An AI agent has just finished its nineteen-minute job. The protocol study is done. The grant for the next one is half-written. A biostatistician is, for once, drinking coffee instead of triaging tickets.
This is what Medeloop is selling. Not a chatbot, not a dashboard, not a pile of charts. Time - returned to the people who got into medicine because they wanted to do science, not paperwork. The unglamorous middle of research, finally automated. And somewhere, a clinician with a daughter who has CRPS gets to read one more paper that exists because the friction came down. That is not a moonshot. That is the product.