How AI can reset healthcare’s administrative cost structure

Rethinking the administrative operating model: From task automation to step-function improvement 

For health systems, the full value of agentic AI comes not only from automating routine tasks but also from changing how work moves across the operating model. That shift is significant. But health systems do not need to redesign an entire administrative function at once to begin capturing value. They can start by identifying where AI can simplify the workflow and accelerate decisions. Then they can focus human review where it adds the most value. 

In AI-driven redesign, health systems use AI as a tool to drive operating model change that improves outcomes, rather than simply using AI to automate existing tasks.  

Step-function improvement comes from redesigning the operating model with AI as a core tool. A future-state view helps leaders look beyond today’s workflows, shifting from “How can AI automate this task?” to “What role can AI/technology play in making the workflow more efficient? What steps or tasks can we eliminate? Where is human review essential?”  

For example, end-to-end revenue cycle processes today require multiple handoffs, from managed care to scheduling to coding to back-end revenue cycle follow-up to payment integrity. With agentic AI, health systems can fundamentally rethink how they accomplish this work. 

For instance, AI can help with authorization submission, claims status checking, inpatient and outpatient coding, submitting appeals with appropriate supporting documentation, incorporating payer guidelines, and identifying improvement suggestions for clinical documentation. Staff can focus instead on high-risk authorizations and complex coding scenarios. 

At Bronson Healthcare, early efforts at AI-driven redesign in the revenue cycle produced swift results, including: a 10-fold ROI, accelerated cashflows, and a better patient experience.  
Read the case study

Within any given administrative function, the pace and magnitude of change will vary by process and sub-process, depending on readiness, magnitude of opportunity, operational complexity, and workforce adoption.  

A longer-term vision can guide the health system’s investment and operating model decisions, including how staff roles will evolve as AI changes the nature of administrative work. As the health system reduces routine burden, staff can shift toward higher-value activities that require judgment, problem-solving, stakeholder engagement, and exception management. 

Examples from other industries demonstrate that the greatest productivity gains rarely come from making legacy processes incrementally faster but from redesigning for a better outcome with technology as a core component of the solution.  

Banking provides a useful example. The industry did not improve productivity simply by digitizing teller transactions. Over time, it redesigned the operating model around customer access, transaction convenience, and lower-cost service delivery. 

What banking teaches about future-state operating model redesign  

When the banking industry sought to reduce the cost of client transactions, instead of simply asking how to make teller-mediated transactions faster, it rethought the entire transaction model. 

Banks started with the outcomes customers were trying to achieve, such as depositing a check, withdrawing cash, transferring funds, paying a bill, or getting financial guidance. They then separated routine transactions that could be completed through automated or self-service channels from interactions that required human judgment or relationship support. 

What emerged was a fundamentally different operating model that changed where work happened and how human staff were deployed. Routine transactions moved to lower-cost, more scalable channels such as ATMs, online banking, and mobile deposits. Human capacity shifted toward advisory support, relationship management, and human actions required under regulatory or policy requirements. 

How AI is driving better outcomes in practice: Supply chain  

We can look at supply chain to better understand how health systems can use AI to move beyond incremental task automation, even if a fully AI-driven function remains a longer-term vision.  

The starting point is the outcome the function is designed to support. In this case, that’s reliable access to the right supplies at the right cost.

An AI-driven model would not simply automate individual tasks. It would redesign supply chain operations around an agentic capability that can interpret demand, coordinate routine purchasing and substitutions, and escalate decisions that require clinical, operational, or financial judgment.  

The value opportunity is to reduce the cost of the supply chain function itself while also improving contract compliance, rebate capture, supply availability, and overall spend performance.  

For example, a medium-sized integrated delivery network may receive annual supply rebates worth $10 million to $20 million, but it can easily miss 5% to 10% of rebates. An AI agent that continuously audits rebate activity and prompts follow-up could recover an additional $500,000 to $2 million annually. 

Even more significant, a typical health system with $500 million in annual supply spend often has 10% or more in benchmarked pricing opportunity across vendors. If AI-driven analytics, research, and monitoring helped capture even half of that opportunity, the organization could reduce supply spend by approximately $25 million (5% of supply spending).¹  

As this example illustrates, the greatest value comes from improving the performance of the supply chain function while reducing enterprise-wide supply spend. By taking on more routine monitoring, reconciliation, and follow-up, AI can also help supply chain teams shift capacity from manual tasks and avoidable rework. Instead, they can focus on higher-value work, such as strategic sourcing, supplier management, clinical collaboration, and exception resolution.  

Example: How to use AI to reduce costs and improve outcomes in the health system supply chain 

The supply chain example illustrates the broader opportunity across health system administrative functions. AI-driven redesign can improve the outcomes these functions exist to deliver—whether that means lower supply expense, faster cash, better workforce productivity, stronger compliance, or better patient and employee experiences. 

Build, buy, or partner: How to source for the future-state supply chain

As supply chain leaders consider where to build, buy, or partner, the health system should treat the sourcing decision as part of the operating model strategy, not just a technology selection.  

Point solutions can create meaningful near-term value, particularly when they address high-volume pain points such as invoice exceptions, contract compliance, demand forecasting, or substitution management. But health systems should also evaluate sourcing decisions against the future-state operating model and whether these solutions help the organization move toward a more coordinated, AI-supported supply chain model. 

For example, a health system evaluating a contract compliance tool should consider whether that tool can eventually connect to item master, purchasing, inventory, utilization, supplier, and finance data to support more proactive decision-making. Similarly, the health system should assess a substitution management tool for its recommendation logic as well as for how it fits into clinical approval workflows, supply chain governance, and site-level operations.

The best sourcing decisions don’t add another layer of fragmented tools. Rather, they create value today while strengthening the foundation for a more coordinated supply chain model over time.

Health system considerations for whether to build, buy, or partner on a supply chain AI capability 

How AI is driving better outcomes in practice: Revenue cycle

Revenue cycle provides a clear example of how different the operating model can become when leaders redesign around the desired outcome rather than the current workflow.

Today, most revenue cycle functions already use some form of rules-based automation. But many still depend on manual review, payer-specific knowledge, and staff intervention to manage exceptions, denials, and follow-up. An AI-native revenue cycle would move closer to a real-time autonomous (or “touchless”) operating model, in which AI continuously evaluates what needs to happen next and coordinates action across the revenue cycle.

In this model, the revenue cycle begins before the patient arrives. AI can identify payer requirements, authorization risk, benefit implications, documentation needs, and patient financial responsibility earlier in the process.  

During the encounter, AI can prompt clinicians at the point of care with questions that drive a cleaner, faster billing process and help improve clinical care and financial counseling as the patient interacts with the provider.  

Before claim submission, AI can evaluate documentation, coding, medical necessity, authorization alignment, contract terms, payer rules, and denial risk as part of a single claim-readiness process. After submission, AI can monitor claim status, identify underpayment risk, initiate routine follow-up, and feed payer and denial intelligence back into upstream workflows.

The revenue cycle becomes a seamless part of care delivery, efficiently and effectively supporting appropriate payment while enhancing clinical quality and patient experience. The work shifts from retrospective correction to proactive prevention. 

Example: How to use AI to improve costs and outcomes in the revenue cycle 

The future state is not simply an AI-enabled version of today’s revenue cycle. In an AI-enabled model, AI may assist individual tasks, such as drafting an appeal, suggesting a code, or summarizing a payer response. In an AI-native model, AI helps orchestrate the revenue cycle around the outcomes it exists to deliver. This results in cleaner claims, faster cash, fewer denials, lower administrative cost, improved patient financial clarity, and stronger enterprise performance.

The case for an administrative cost reset

Bringing this same mindset to other administrative functions will be the work ahead for healthcare organizations. The aim is a more fundamental reset of the administrative cost structure, not through blunt cuts but by redesigning work in ways that improve both costs and outcomes. 

That ambition may sound audacious, but it should not feel unrealistic. Like health systems, airlines, hotels, and utilities companies also operate in regulated, labor-intensive, 24/7, asset-heavy environments. Yet their administrative and support cost structures are often much lower—in many cases, roughly half that of healthcare. 

Healthcare administrative costs are much higher than other complex regulated industries² 

Healthcare is different from these other industries in important ways, but the comparison highlights the magnitude of the structural gap. For a typical mid-size integrated health system with a $2 billion annual cost base, administrative activities may represent 25% to 30% of total costs, creating a $500 million to $600 million cost pool.³ 

If the health system reduced administrative and support costs from 25% to 30% of the cost base toward levels typical of other operationally complex industries, the potential annual savings would be roughly $160 million to $440 million. 

That gap should prompt bigger questions: What would it take to reduce the administrative cost base by 20%, 30%, or even 50% over time? Which work could health systems eliminate? Which activities truly require human judgment, and which could shift to more scalable AI-supported models?

Transformation of this scale is difficult and will not happen overnight. It will require disciplined prioritization and sequencing across functions. But keeping a step-function goal in view can help leaders move beyond incremental improvements and focus on structural change that can unlock meaningful improvement in cost, performance, and enterprise outcomes.  

For a health system operating near break-even, reducing administrative costs by roughly 10% to 16% would be enough to increase operating margin by approximately 3% to 4%—a material improvement in enterprise performance. 

Additional contributors: Cindy Lee, Chief Strategy Officer, and Anneliese Gerland, Partner and Senior Vice President, Strategy

Sources

¹ Chartis experience 

² Non-healthcare industry estimates are directional and based on public company/industry financial statement proxies, including SG&A, G&A, administrative and general expense, customer support, sales/distribution, and other support-function costs. Because accounting categories vary by industry, comparisons should be interpreted as directional rather than like-for-like cost accounting benchmarks. 

³ Chartis analysis based on Syntellis benchmark data and proprietary information. 

Our services