Sully.ai

It is scheduling. It is answering the phone. It is checking insurance eligibility, typing up the visit into a note, translating that note into billing codes, and then chasing the claim when the code is wrong. A physician does not go to medical school to do any of this, and yet a large share of a clinician's day - and a large share of a clinic's payroll - goes to exactly this kind of work. The economic term for it is administrative overhead. The human term for it is why everyone is exhausted.

Sully.ai, founded in 2023 in Mountain View and backed by Y Combinator, sells software into this gap. What makes the company worth writing about is not that it built an AI scribe - lots of companies built an AI scribe - but the framing it chose. Sully doesn't describe its products as features. It describes them as employees. There is an AI Receptionist. An AI Scribe. An AI Medical Coder. An AI Triage Nurse. An AI Pharmacist. An AI Consultant. You do not install them so much as you hire them, and they show up inside the electronic health record system the hospital already runs - Epic, Athenahealth - and start doing the tasks.

This is a marketing choice, and you should be a little suspicious of marketing choices. But it is also a genuinely useful reframe, because it tells you where the company thinks the money is. If you sell a doctor a better tool, you are competing on features and price against every other tool. If you sell a clinic an employee, you are competing against the cost of hiring a human to answer the phone - a much larger, much more forgiving budget line. The pitch is not "our note-taker is 8% better." The pitch is "you don't need to post that job."

The wedge, as it usually is in this category, was documentation. Sully's AI Scribe listens to a patient-physician conversation in real time and turns it into a structured, HIPAA-compliant clinical note - the SOAP format doctors already write in, minus the typing. Ambient scribing is a crowded field; Abridge, Ambience, Suki, Nuance's DAX and others are all there. But scribing has a lovely property as a business: once your software is already sitting in the exam room, listening, the marginal cost of adding a second job - say, turning that same conversation into ICD-10 billing codes - is mostly onboarding. Sully claims its AI Medical Coder hits 93.8% accuracy on that coding, and pairs it against a human physician error rate it puts at around 40%. Those are the company's numbers, and headline accuracy stats in healthcare AI deserve a raised eyebrow, but the direction of the argument is clear enough: the boring tasks are where the leverage is.

The people building this are not the healthcare lifers you might expect. CEO Ahmed Omar is an Egyptian immigrant and, improbably, a three-time gold-medalist MMA fighter who previously co-founded Odiggo, an auto-tech marketplace in the Middle East. His co-founders, Ahmed Nasser and Chaitanya Gharpure, bring sales and engineering. This is a distribution-and-speed team more than a clinical-research team, which fits the strategy: Sully's edge is less "we invented a new model" and more "we got the model into your EHR and made it do the paperwork nobody wanted." The company says it went from roughly half a million to several million in annual recurring revenue in under nine months, and reached seven-figure ARR within ten months of launch. Those are self-reported and worth treating as approximate. But the January 2025 Series A - around $21.8 million, led by Amity Ventures with SemperVirens, the Sequoia Scout Fund and others - is a real check, and it says investors bought the framing too.

The obvious question is whether an autonomous agent should be answering triage calls or writing prescriptions at all. Healthcare is the one industry where "move fast and break things" is a lawsuit. Sully's answer, to the extent it has a public one, is the same answer everyone in the category gives: keep it HIPAA-compliant, keep the human in the loop, and aim the agents at the low-stakes, high-volume work first - the front desk, the coding queue, the after-visit summary. That is the sensible place to start, and it is also, not coincidentally, the place where the labor shortage is worst and the willingness to pay is highest. Sully.ai has bet that hospitals will hire an AI before they hire a person. So far, several hundred of them have.