Every time a physician submits a prior authorization request, a claims adjudicator processes a denial, or a care manager reviews a treatment plan, someone — or something — has to interpret the rules. Those rules live in hundreds of pages of policy documents, provider contracts, and clinical guidelines that health plans update constantly but rarely make computable. For a $2.89 billion AI market targeting healthcare payers, that gap between written rule and actionable decision has become the defining business problem of the decade.
Chicago-based Boost Health AI, which launched in October 2025, is betting that solving it starts not with building another software platform, but with unlocking what the company calls the payer rulebook.
A Crisis Hidden in Plain Sight
The administrative burden on American healthcare is staggering by any measure. According to a 2024 survey by the American Medical Association, physicians and their staff spend an average of 13 hours per week processing roughly 39 prior authorization requests, with 40% of physician practices employing staff dedicated exclusively to that single task. Twenty-nine percent of physicians report that prior authorization delays have led to a serious adverse event for a patient in their care.
Meanwhile, health plans themselves are under acute financial pressure. The National Association of Insurance Commissioners reported that health plans’ aggregated net income fell 14% in the first half of 2024 compared to the prior year, while cash flow from operations dropped 86%. Hospital and medical expenses rose $35 billion in the same period. McKinsey estimates that payers embracing AI and automation could reduce administrative expenses by 13 to 25% — a margin that, for large insurers, translates into hundreds of millions of dollars annually.
The Rulebook Problem
The obstacle blocking faster, more consistent decisions is not a lack of data. It is that the logic governing payer decisions, what gets approved, denied, flagged, or escalated, is trapped in static documents that no machine can readily query or apply. Policies sit in PDFs. Contract terms live in spreadsheets. Clinical guidelines are interpreted differently by different staff members in different departments.
Boost’s product architecture is built around dissolving that bottleneck. Its AI Foundry contains healthcare-trained AI Agents that extract and structure the logic inside payer documents, converting policies, contracts, and clinical guidelines into what the company calls machine-readable, auditable, reusable intelligence. Those structured rules then feed into pre-built Solution Blueprints such as workflow templates covering claims processing, utilization management, care management, and compliance validation.
“Every payer we meet is managing enormous complexity,” said Raheel Retiwalla, Boost’s Chief Product Officer. “The underlying logic is there, it’s just scattered. We built Boost to unify that logic and make it actionable, consistent, and efficient.”
Explainability as a Regulatory Hedge
The company’s timing is deliberate. Health insurers are operating under growing legal and regulatory scrutiny over how AI influences coverage decisions. A 2024 Senate Permanent Subcommittee on Investigations report documented how Medicare Advantage insurers used algorithmic tools to deny patients access to post-acute care. A survey of 93 large health insurers by the NAIC found that 84% were already using AI for some operational purpose, yet governance has struggled to keep pace with adoption.
Boost’s architecture uses deterministic decision rules and DMN decision tables designed to make every outcome traceable. “Efficiency without clarity eventually creates risk,” Retiwalla said. “Our goal is to make efficiency and clarity work together so payers can see how every decision is made and continuously improve it.”
This audit-ready design, where each decision carries citations, versioning, and compliance monitoring, may prove increasingly valuable. Major insurers, through industry group AHIP, have pledged to deliver real-time prior authorization decisions on 80% of electronic requests by 2027, standardized on FHIR-based submissions. Meeting that deadline without explainable, documented decision logic carries substantial regulatory exposure.
Owning the Infrastructure
What distinguishes Boost from conventional SaaS competitors is its licensing model. Rather than charging recurring subscription fees for access to a hosted platform, Boost sells perpetual, extensible IP licenses — meaning the payer owns and controls the technology within its own environment. Clients can deploy on their own cloud, on-premise infrastructure, or through managed hosting.
“We believe payers should have sovereignty over their intelligence,” Retiwalla said. “When you own the framework, you can evolve it as technology changes. You’re not waiting on someone else’s roadmap.”
Whether that model scales commercially remains an open question. Boost has not disclosed named clients, revenue figures, or funding to date. The company is seven months old, and its early implementations have focused on care management and utilization management workflows. The global AI for healthcare payer market is projected to grow from $5.7 billion in 2025 to $46.67 billion by 2035, but it is also drawing competition from well-capitalized incumbents, including Oracle Health, Microsoft Nuance, and a growing field of funded startups. Boost’s proposition, that payers are better served by owning their AI infrastructure than renting it, is a coherent argument. Proving it at scale is the work still ahead.
