CodaHx

CodaHx is a Baltimore-based healthcare technology company that applies AI to medical claims auditing for self-funded employer health plans. Its platform takes 837/835 claim batches streamed from a plan's third-party administrator (TPA) and audits 100 percent of claim lines — rather than a sample — before the TPA issues payment. Checks run in parallel across coding edits (NCCI and MUE), duplicates and coordination of benefits, summary plan description (SPD) terms and coverage, and published-rate benchmarking. Claims that clear proceed through the TPA as normal, while flagged items are packaged into correction packets with cited evidence for the plan's reviewers.

The product is organized into three modules. CodaHx Audit detects billing code errors, duplicate charges, unbundling, upcoding, coverage misapplications, and pricing discrepancies. Rate Atlas benchmarks billed amounts against state-published rates with coverage described across all 50 states and flags charges substantially above the benchmark. Member Advocacy explains bills to employees in plain language and disputes errors on their behalf. Additional functions include stop-loss threshold monitoring with alerts as a member's claims approach the limit, a charity-care eligibility checker that screens members against hospitals' financial-assistance policies based on household size and income, renewal and stop-loss modeling, and generation of TPA packets and renewal reports. Five specialist agents — medical coding, cost benchmarking, financial aid, renewal optimization, and report generation — work each batch in parallel and post findings with citations, with human reviewers retaining decision authority. The system also records adjuster decisions as plan-specific rules so it learns which findings a given plan pursues or waives.

In an earlier described form, the company's AI tool was aimed at patients: it extracted and interpreted charges from medical bills, translated them into accessible language, flagged potential errors and inconsistencies against national coding dictionaries, and surfaced financial-assistance programs a patient might qualify for.

Founding story

The platform was developed at Johns Hopkins University by Yoohyuk Chang, Raam Unadkat, and Oren Wei, initially as a tool to reduce the cost of medical bills by explaining charges to consumers, scanning bills for errors, and helping users apply for financial assistance they were eligible for. CEO Oren Wei has said the company was built with HIPAA compliance and data-privacy protocols developed in consultation with healthcare attorneys, billing specialists, and compliance experts.

Business model

CodaHx sells an AI claims-auditing platform to self-funded employer health plans and works alongside their third-party administrators and brokers. The platform ingests 837/835 claim batches from the TPA, audits them before the payment run, and returns correction packets and recovery documentation to the plan; an earlier consumer-facing product analyzed individual patients' medical bills.

Traction

Product materials cite illustrative plan-level figures including roughly $184,220 caught before payment in a quarter, a sample batch of 1,279 claims with 94.1 percent cleared and 5.9 percent flagged, and workspace figures of $109K flagged to recover and $221K prepared for recovery. In a customer example described by the CEO, a patient in Illinois who submitted bills after a car accident recovered over $400 in identified errors.