Over half of denied-claim appeals succeed. Fewer than 15% ever get filed - not because they'd lose, but because filing takes hours no billing team has. Aegis turns those hours into minutes.
A denied insurance claim is not a "no." It is a "not like this." Somewhere in the payer's letter is a policy clause, a missing code, or a document that would flip the decision - and roughly half the time, when a provider bothers to argue, it does flip. The catch is the bothering. A single appeal can eat two hours of a billing clerk's day: reading the denial, hunting through the chart, matching the right policy language, assembling the packet, and finding the right portal to file it. Multiply that across a hospital's daily denial stack and most of them never get answered. That gap - winnable money left on the table for lack of labor - is the entire business Aegis is built around.
Aegis is a San Francisco startup in Y Combinator's X25 batch. It sells software to healthcare providers, hospitals, and medical billing groups, and it does one narrow thing with unusual focus: it automates the insurance denial appeal from detection to resolution. A claim gets denied, Aegis catches it, pulls the supporting records, drafts the appeal, submits it, and tracks what happens next.
The numbers behind US medical billing read like a rounding error that grew teeth. Providers lose more than $260 billion a year to denied claims and spend another $20 billion trying to reverse them. And yet the appeal rate sits below 15%. Read those two facts together and the story writes itself: it isn't that appeals lose. It's that appeals are expensive to file and billing teams are perpetually underwater, so the rational move for an overwhelmed clerk is to triage, appeal the biggest claims, and quietly write off the rest.
Aegis's founders frame the denial not as a cost of doing business but as a solvable engineering problem. The care was delivered. The revenue was earned. What stands between the provider and the money is paperwork - and paperwork is exactly the kind of thing software is good at.
Aegis works less like a chatbot and more like an assembly line. It watches the flow of claims and Explanation-of-Benefits data coming back from clearinghouses, flags the denials, and ranks them by two things a busy team actually cares about: how much money is at stake and how likely the appeal is to win. The high-value, high-odds denials rise to the top of the queue.
The generation step is where the AI earns its keep. Aegis assembles a ready-to-submit appeal using the payer's Explanation of Benefits, the patient file, and records pulled from the EHR, producing letters tuned to the specific policy and payer. Then it files - one click to the payer portal - and keeps a live scoreboard of what's pending, won, or lost, feeding denial and revenue analytics back to the team so the next batch of claims goes out cleaner.
"Aegis automates the insurance appeals process end-to-end: from denial detection and compliant appeal generation to submission, tracking, and actionable analytics."
The buyer is a billing team - inside a provider practice, a hospital, or a third-party revenue-cycle-management firm. These are the people who already spend their days inside Availity and payer portals, re-keying denial data and chasing documents. Aegis's pitch to them is time: cut the time to file an appeal from over two hours to under two minutes, lower the cost per denial, and turn the write-off pile back into a work queue.
Plenty of companies say "AI for the revenue cycle." Aegis picked the narrowest, most measurable slice of that - the appeal - and built for it. Narrowness is the strategy. When you scope down to a single workflow with a clean outcome (the denial is either overturned or it isn't), you get a metric you can actually stand behind, and you get integrations deep enough to matter. The plumbing into Epic, Athenahealth, eClinicalWorks, and the payer portals is the unglamorous work that decides whether healthtech survives contact with a real hospital. The demo is easy; the integration is the moat.
There's a second, stranger edge to the framing. Insurers increasingly lean on AI to deny claims at volume. Aegis is AI pointed the other direction - built to appeal them. It's an arms race the founders are happy to name, and it explains why speed and volume matter: if denials are automated, the response has to be too.
Most denials aren't appealed because they'd lose. They're not appealed because filing takes hours nobody has. Remove the hours, and the economics invert.
Aegis was started by three friends from Carnegie Mellon University, each arriving from a different discipline. Krishang Todi, the co-founder and CEO, studied economics and math and previously did fixed-income risk modeling at a top Indian investment fund - a background in pricing uncertainty that maps neatly onto ranking which denials are worth fighting. Aarav Bajaj came from computer science and machine learning, with AI research experience and a stint at Palantir deploying data-driven systems at scale. Dhanya Shah studied information systems and computer science and shipped production software at three companies before this. Risk, AI, and production engineering - one founder for each seam of the product.
It is a four-person operation selling into one of the most regulated corners of the economy, which is why compliance shows up early rather than as an afterthought. The company built HIPAA-compliant infrastructure from the start, using tooling like Porter and Oneleet to keep its stack SOC 2 and HIPAA compliant while moving fast.
Aegis is part of Y Combinator's X25 (Spring 2025) batch, where it was noticed as one of the healthcare startups automating insurance appeals. Beyond YC, its reported backers include Rebel Fund, the Sequoia Scout Fund, Mana Ventures, and Maiora Ventures. The company is early - the kind of stage where the product's reported metrics (a 78% overturn rate, roughly 90% faster resolution) are promising signals rather than audited track records, and worth reading as such.
The wager underneath Aegis is that the biggest untapped revenue line in American healthcare isn't a new treatment or a new payer - it's the money providers already earned and never collected because the follow-up was too expensive to do by hand. If appeals really do win more than half the time and get filed less than 15% of the time, then the ceiling isn't a marketing number; it's arithmetic. Aegis is betting it can automate its way into that gap, one denial letter at a time.