Healthcare organizations are increasingly turning to AI to accomplish tasks like record filing, standardizing patient data, and extracting patterns from large datasets. However, one of the areas that remains largely overlooked in this modernization is claims processing.
Despite being essential to determining how providers get reimbursed, most claims are still handled manually using outdated systems — an error-prone workflow that’s prone to errors and offers little visibility into the approval process. As a result, payments are often delayed, underpaid, or flat-out denied, and hospitals end up eating the cost, leading to reduced budgets and fewer resources for care.
Engineer Sally Ragab witnessed the effects of these inefficiencies while working as a healthcare auditor, which is why in 2024, she founded Neunetix to bring automation and structure to the claims process. The company’s flagship product, ClaimStreamAI, is designed to securely process claims, ensure compliance with billing rules and industry standards, and provide a transparent, audit-ready record of how each claim is handled.
A Firsthand Look at the Challenges of Medical Billing
Before founding Neunetix, Sally Ragab worked as a product manager for companies like iFIT and Atlas Health, where she led the launch of data platforms that serve over 15 million users. But her insights into hospital billing practices came from her time as a freelance auditor, which is when she witnessed firsthand the substantial losses caused by outdated billing infrastructure.
Sally traced the problem to the processes behind revenue cycle management, the system hospitals use to submit claims and receive reimbursement from payers such as Medicare. She noted that many of these workflows are still handled manually, often through fragmented legacy systems that give staff frequently little insight into why claims are approved or denied — making it hard to correct errors or properly navigate the appeals process.
She also discovered the significant financial losses caused by these inefficiencies, as missed submissions and unresolved denials result in underpaid or rejected claims (estimated at over $262 billion annually). In one instance, Sally witnessed an orthopedic clinic lose $280,000 in reimbursements after missing multiple appeal deadlines.
These experiences made the need for a more reliable system clear. In Sally’s words, “Watching surgeons debate layoffs over an avoidable paperwork error pushed me to create technology that gets providers paid on the first submission.”
Inside ClaimStreamAI: Neunetix’s Layered Claims Processing System
In 2024, Sally founded Neunetix to address these inefficiencies. Bootstrapped with her own savings, the company set out to build an automated claims processing system that could easily be integrated into existing healthcare workflows.
That system is ClaimStreamAI, which processes billing claims through a structured four-stage pipeline designed to enforce policy, interpret documentation, and closely validate outputs.
First, billing specialists enter claims and chart notes into the program through a secure gateway, where a tokenizer-level filter immediately removes all patient-identifiable data so that it never reaches the AI layer.
Next, a reasoning mesh engine runs each claim through established billing rules, including National Correct Coding Initiative edits, local coverage determinations policies, and payer-specific requirements. Claims that fail these validations are filtered out early, preventing AI hallucinations from influencing billing.
Once a claim passes these validations, it’s processed through a private GPT-4o instance that interprets clinical text like operative notes and standardizes it into a structured format. This output is then analyzed by models that are trained on 12 million claims records, enabling them to predict billing codes, assign confidence scores, and automatically approve claims scoring 92% or higher.
Finally, any claim that doesn’t meet this threshold gets flagged for human review. These manual decisions are recorded and used in nightly reinforcement learning cycles, helping the system improve accuracy over time. Building on this, a physician-led advisory board reviews every model update before deployment, and 5% of all claims (including high-confidence ones) are blind-audited weekly to catch errors early and prevent model drift. The system also includes protections to maintain fairness and avoid misuse, flagging potential upcoding (the submission of inflated billing codes for higher reimbursement) and undergoing quarterly bias audits to ensure claims are handled consistently across patient demographics.
To ensure maximum transparency, each processed claim includes information that demonstrates how the system interpreted each part of the input. These are then logged in an immutable ledger stored in a database, allowing regulators to trace the reasoning behind each claim from a unified dashboard. This commitment to transparency extends to Neunetix’s pricing model, as clients only pay a fee if collections exceed a pre-agreed threshold, ensuring they only pay for real value.
The result is a system that minimizes manual errors, maintains human oversight, and integrates smoothly into hospital operations. It’s a platform that Sally refers to as an “exportable framework”: a model for building AI systems that are not only technically sound but also built for accountability.
A New Infrastructure for the Economics of Healthcare
Since its launch last year, ClaimStreamAI has delivered strong early results. In a five-site cohort pilot, providers recovered over $3 million in legacy write-offs and achieved a 99.4% clean-claim rate, while a provider group in California facing an eight-month denial backlog recovered more than $450,000 and improved its clean-claim rate from 89% to 99.2% in just two months.
As Neunetix expands, Sally sees ClaimStreamAI becoming a foundational part of daily health operations, allowing for claims to be scrubbed, coded, and adjudicated within minutes of a procedure and with payments already in motion. She sees the platform greatly reducing administrative overhead, freeing up more budget for nursing staff and frontline care, and giving clinicians more time to focus on patients.
With ClaimStreamAI, Sally Ragab is creating a path to faster reimbursement, smarter resource allocation, and a healthcare system that runs more efficiently for providers and patients alike.
The post How Sally Ragab Is Using AI to Save Money in Healthcare appeared first on The Village Voice.