Healthcare revenue cycle management represents one of the most complex, fragmented, and regulation-heavy environments in modern commerce. Operating across 154 electronic medical record (EMR) systems, more than 25 practice management platforms, and handling approximately 100,000 cases weekly, Coronis Health finds itself at the epicenter of a massive operational transformation. Under the guidance of Chief Technology Officer Doug Marcey, the company is systematically replacing manual, repetitive administrative tasks with a sophisticated framework he terms "accountable intelligence." Rather than seeking to eliminate human oversight, Marcey’s strategy hinges on reserving human expertise exclusively for critical judgments, while leveraging advanced automation, traditional Robotic Process Automation (RPA), and intelligent document processing (IDP) for everything else.
The deployment of this automation architecture has evolved rapidly since Coronis Health initially formalized its platform agreement with UiPath in March 2025. What began as a parallel strategy utilizing traditional RPA alongside intelligent document processing has matured significantly over an 18-month period. Today, Coronis Health has constructed robust proprietary layers atop the UiPath platform, integrating core RPA, IDP, agentic automation, and orchestration capabilities to streamline workflows that were previously stifled by technological silos and interoperability gaps inherent in legacy healthcare systems.
Operational Realities and Technological Fragmentation
The financial mechanics of healthcare billing leave virtually no room for error. Operating primarily on contingency-based models, Coronis Health manages the entire lifecycle of medical claims: taking raw medical records from providers, translating them into coded claims, submitting them to insurance payers, and aggressively managing denials, appeals, and subsequent collections. Every transaction cost directly impacts the bottom line, rendering tool selection and operational efficiency matters of vital economic survival.
The scope of this challenge is particularly visible within the company’s anesthesia division, its most valuable and idiosyncratic service line, which processes roughly 60,000 documents weekly, including complex medical records, explanations of benefits, and lockbox feeds. Interoperability issues compound the difficulty. While programmatic data extraction is occasionally feasible, writing account information and claim data back into disparate client systems of record presents a formidable barrier. To bridge this gap, Coronis Health deploys agents and RPA bots for last-mile execution, ensuring a unified "single pane of glass" workforce approach that consolidates disparate systems into a cohesive operational workflow.
This last-mile capability has yielded substantial operational gains. Prior to the integration layer, staff members were siloed, trained exclusively on specific EMR systems, and unable to pivot to alternative teams during demand surges. The generic adapter layer supplied by UiPath removed these workforce mobility constraints. Furthermore, enterprise capabilities like UiPath’s Database Hub—which facilitates direct reads and writes to SQL Server, Oracle, and Databricks without requiring custom code—alongside the Model Context Protocol (MCP) Connector, have largely dismantled historical integration barriers.
Cost Optimization and Tiered Document Processing
Handling 100,000 cases every week makes transaction cost optimization just as crucial as extraction accuracy. To manage expenses effectively without compromising quality, Marcey’s engineering team designed a highly sophisticated, tiered document processing pipeline.
At the center of this architecture is a homegrown discriminator model that evaluates incoming medical records, assesses their complexity and structural integrity, and routes them to the most cost-effective processing engine available. The tiers span from Apple’s built-in Optical Character Recognition (OCR) running on localized Mac Studio hardware—capable of processing roughly 30 pages per second compared to the standard three pages per second yielded by Poppler—to standard IDP, Tesseract-based extraction, open-source OCR models, and high-performance commercial visual language models housed on Coronis Health’s own physical infrastructure for exceptionally complex records. This pipeline also incorporates UiPath’s IDP paired with Action Center for human-in-the-loop validation and correction.
This engineering approach directly mirrors deterministic-versus-cognitive operational principles: routing workflows to deterministic, tokenless execution wherever feasible, and escalating to cognitive, token-based processing strictly when necessitated by document complexity. Medical billing inherently features a high volume of straightforward records—roughly 50 to 75 percent of the daily workload—complemented by highly intricate edge cases, such as Medicare beneficiaries holding multiple secondary coverages or trauma cases lacking prior authorization. By routing these tasks to the cheapest effective tool, Coronis Health protects its thin operating margins.
While deploying IDP across challenging divisions like anesthesia revealed that roughly 14 to 20 percent of records still arrive entirely handwritten—bypassing standard IDP capabilities—the impact on digitally sourced and scanned documents proved transformative. Interestingly, average handle times decreased by only about 30 seconds. However, the operational reality shifted profoundly as standard deviations tightened significantly. While average processing times remained relatively stable previously, outliers frequently required tenfold the standard time as operators manually flipped through disordered pages. By achieving consistent processing times, Coronis Health revolutionized downstream planning, enabling accurate staffing forecasts, reliable cost projections, and steadfast adherence to client service-level agreements. Furthermore, error rates plummeted: the percentage of records containing at least one undetected human error dropped from seven percent to three percent.
The Payer AI Arms Race and Compliance Boundaries
Operating within healthcare compliance frameworks introduces stringent legal liabilities. Because Coronis Health submits claims on behalf of healthcare providers, erroneous coding exposes providers directly to severe audit liabilities and potential legal action. Marcey maintains an unwavering stance regarding the boundary of automation:
"We’re a services company, and our clients hire us for our experience and the quality of work that we do. We are never going to be in a place where we can go to a client and say, ‘oh, sorry, the bot messed up.’ That’s not our thing. From a compliance perspective, we are going to stand behind the codes that we put on the record, the submissions we’ve done, and you’re not going to end up in court for over-billing because your AI bot decided that they could get an extra RVU [Relative Value Unit] out of this other code."
This cautious boundary is further reinforced by an escalating technological arms race. Major insurance payers, including entities like United Healthcare, are increasingly deploying proprietary artificial intelligence agents designed to automatically review incoming records and aggressively flag or deny claims. Providers and billing intermediaries are acutely aware of this dynamic, which Marcey characterizes as a digital Cold War. Coronis Health actively advises its physician clients on how to structure medical documentation to ensure smooth passage through payers’ automated evaluation systems.
Concurrently, a wave of tech-focused startups is attempting to pitch direct-to-physician AI tools promising automated billing without traditional services firms. However, industry veterans note that software alone underestimates the immense friction of navigating 154 distinct EMRs and over 25 practice management platforms without comprehensive clinical context.
Knowledge Graphs, Denial Prevention, and Change Management
To stay ahead of denials, Coronis Health developed a proprietary knowledge graph platform that systematically links diagnosis codes and procedure codes directly to specific sections of payer contracts, client-specific policies, and internal codebooks. When a claim enters the pipeline, the system queries the graph for relevant contract clauses, passing targeted, highly specific context to an AI agent rather than loading entire multi-page provider contracts into a token-hungry context window. This architecture prevents the performance degradation commonly associated with massive prompt volumes. A secondary model subsequently generates root-cause summaries for flagged items, allowing quality control teams to focus their expertise where it is most needed rather than reviewing records blindly.
Managing internal workforce expectations during rapid technological acceleration has proven equally demanding. Coronis Health boasts extraordinary employee retention—typified by long-tenured staff members in regions like India marking up to 27 years of service. While these veteran professionals ensure exceptional quality control, the rapid pace of automation development can induce professional anxiety. Rather than framing automation as a replacement tool, leadership emphasizes the elimination of mundane digital toil.
Adoption of internal tools also required strategic refinement. While a citizen developer program experienced lackluster uptake, an enterprise AI chatbot deployment branded as "Aria" within Microsoft Teams achieved widespread success due to its familiar conversational interface. Staff actively utilize Aria to submit automation ideas from across the enterprise. Looking forward, the organization aims to connect this idea-capture mechanism directly to newly released coding agents to accelerate development cycles.
The volatile economics of token-based artificial intelligence have similarly influenced infrastructure decisions. Because unpredictable token pricing can instantly erode profit margins on contingency-based contracts, Coronis Health made the strategic decision to invest heavily in on-premises hardware and colocation facilities. By stabilizing fixed infrastructure costs, the firm insulates itself from the fluctuating expenses of cloud-based cognitive processing, while continuing to utilize lightweight, traditional machine learning models for tasks where token expenditure is unwarranted.
Broader Implications for Enterprise Automation
The trajectory of Coronis Health offers a compelling blueprint for automation implementation within heavily regulated service industries. The emphasis on tightening standard deviations rather than merely accelerating raw speed highlights a mature understanding of operational value. In high-stakes sectors where regulatory compliance, audit defensibility, and client trust are paramount, the concept of accountable intelligence provides a vital framework. Organizations that successfully navigate this landscape will be those that integrate cutting-edge cognitive tools while maintaining strict human ownership over final accountability.
