Avelis Health's Pre-Seed Bet Audits Every Medical Claim With an AI Agent

The YC-backed startup aims to intercept billing errors before they reach the patient, targeting a 2-7% savings on annual claims spend.

About Avelis Health

Published

In the sprawling, opaque world of medical billing, errors are a given. They are also expensive, adding billions in waste to the healthcare system each year. Avelis Health, a New York-based startup fresh from Y Combinator’s Summer 2025 batch, is making a pointed bet: that the best place to catch these mistakes is not after a bill is sent, but before a payment is ever made [Y Combinator, 2025].

The upstream wedge

Avelis is built for a specific, data-rich audience: self-insured employers and the health plans that administer their benefits. The company’s core proposition is to integrate with existing claims processing systems and apply machine learning models to audit 100% of incoming claims in real time [Y Combinator, 2025]. The goal is to flag billing, coding, and contract compliance errors before funds are disbursed. This upstream focus on payment integrity is a deliberate wedge. By preventing erroneous payments, Avelis argues it can save its clients between 2% and 7% on their annual claims spend [Avelis Health, 2025]. The recovery process is then automated through voice AI agents, which handle the work of retrieving medical records and initiating overpayment conversations with providers.

A team built for product

The founders, Angel Onuoha and Ahmad Shehu, have worked together for four years and lead a team of three [Y Combinator, 2025]. Onuoha, the CEO, is a Harvard graduate in economics and computer science and a former Google Associate Product Manager [Forbes, 2021]. Shehu, the CTO, brings a civil engineering background and prior software roles at companies like Zuvy Technologies [RocketReach, 2026]. It is a profile typical of Y Combinator’s early-stage bets: technically capable founders applying a modern software and AI toolkit to a large, entrenched problem, with the accelerator’s $500,000 pre-seed check providing the initial runway [Yahoo Finance, Jun 2025].

Founder Role Key Background
Angel Onuoha CEO Harvard University (Economics & CS), Former Google APM [Forbes, 2021]
Ahmad Shehu CTO Eastern Mediterranean University (Civil Engineering), Former software roles [RocketReach, 2026]

The competitive and credibility gap

The ambition is clear, but the path is crowded with established players and unproven execution. Avelis enters a payment integrity market with competitors like Alaffia Health and Anomaly. The startup’s primary differentiator appears to be its emphasis on full automation and real-time prevention, but these are claims yet to be validated with named enterprise customers. The public record shows no disclosed deployments, a common pre-seed reality but a significant gap when selling into the risk-averse, compliance-heavy world of employer health plans.

The risks for Avelis are practical hurdles any new entrant must clear in a sector defined by legacy systems and long sales cycles:

  • Clinical validation. The accuracy and audit trail of AI determinations will be scrutinized by plan medical directors and legal teams.
  • Integration depth. True real-time auditing requires deep, reliable API connections to claims adjudication platforms.
  • The sales motion. Convincing a self-insured employer to install a new AI layer into a core, regulated financial process is an enterprise sale.

What success looks like for patients

For the individuals whose care is at the center of these transactions, the impact of a tool like Avelis is indirect but meaningful. The patient population here is anyone covered under a self-insured employer plan. Today, the standard of care for addressing these errors is almost entirely reactive. Patients or employer benefits administrators must notice a discrepancy, gather evidence, and initiate a protracted appeals process after the money has already changed hands. Avelis is betting that intercepting the error at the source is not just more efficient for the plan, but a more humane outcome for the member who never sees the mistake in the first place.

Read on Startuply.vc