The administrative burden of healthcare is not a documentation problem, it is a workflow problem.
Over the past year, healthcare organizations have widely adopted generative AI for a variety of documentation-related activities, such as writing appeal letters, producing patient-friendly summaries, and even assisting with administrative writing. While these tools have improved the way information is created, they do little to address the underlying problem: a fragmented, manual, and procedurally complex administrative system that consumes time, delays care, and exhausts clinical and operational staff. From prior authorizations and decision-making to reimbursement and tracking, healthcare workflows remain laden with steps that require coordination, tracking, and escalation. This is a job that goes far beyond writing a text.
An administrative crisis that the industry can no longer ignore
When administrative workflows remain fragmented and manual, the burden does not remain abstract, but is compounded across the entire workforce. The consequences of workflow-driven inefficiency are now visible in staff shortages, increased workloads, and measurable physician burnout. What begins as a failure to automate end-to-end processes ultimately manifests in wasted time on prior authorizations, delays in patient access, and unsustainable operational strain across healthcare systems.
This gap is growing as labor shortages intensify. The National Center for Health Workforce Analysis It is projected that between 2023 and 2038, there will be a major shortage of licensed practical and vocational nurses (LPNs). By 2028, supply will only cover around 83% of demand, while that figure will sharply decline to just 70% by 2038. Not to mention that administrative procedures in clinical practice are actively slowing down its efficiency, as around 80% of doctors in the recent survey of the American Academy of Allergy, Asthma and Immunology report that they always experience delays in accessing prescription medications and biologics. Furthermore, in the same report, more than 80% of them believed that APs interfered with chronic treatment, and more than 96% thought that APs somewhat or significantly negatively affected clinical outcomes. In general, all of them express that the burden of PA in their practices is high (26%) or extremely high (71%).
Behind these numbers are patient access teams, case managers and revenue cycle staff with increasing workloads. They manage eligibility checks, benefit verification, documentation alignment, authorization tracking, and denial management. If AI can only make the written part of this job more efficient, the overall load barely changes.
Enter Agentic Systems: From Support to Automation
Agent systems are the game-changers that healthcare desperately needs. Unlike generative tools that generate text, these AI-powered agents handle end-to-end workflows with intelligence and autonomy. They pull data from disparate sources, apply payer-specific rules, verify details, submit data through portals, monitor progress, escalate issues, and log everything for audits, all without human intervention.
Think of it as going from a typewriter to a full-fledged executive assistant. A generative tool could draft an appeal letter; An agent system identifies the reason for the denial, collects supporting documents (such as lab results and medical records), packages the appeal, routes it appropriately, automatically follows up, and updates the EHR. One creates information; the other drives results.
Interoperability as a new competitive advantage
Agent systems offer something that generative AI cannot: the ability to navigate the disconnected technologies of healthcare. They can extract data from EHRs, payment portals, laboratory systems, and internal databases and then act on them to complete entire workflows.
This is important because interoperability is no longer optional. TEFCA, CMS rules, and state mandates now require not only data sharing, but also an actionable and traceable flow of information, something that traditional RPA or text generation tools cannot offer.
By bringing together systems that don’t communicate with each other, agent systems finally solve the “last mile” problem of healthcare, eliminating the manual re-entry, portal hopping, and record matching that slows everything down.
The result is faster decisions, fewer denials, and a smoother patient experience. As pressures increase, the ability to orchestrate workflows across fragmented systems will not just be a matter of compliance; It will be a fundamental strategic advantage.
An ethical imperative that prioritizes the patient
Beyond efficiency, agent systems defend ethics and equity. In an era of skepticism about AI, its transparent and auditable nature builds trust—every decision can be traced back to the data, mitigating the biases that plague opaque generative models. For underserved populations, this means faster access—no more disproportionate delays for low-income patients who must navigate complex payment rules. This is about justice, about ensuring that AI not only accelerates the status quo but dismantles barriers to care.
Operational benefits for health systems
Agent systems alleviate administrative pressures where they are most costly and disruptive. They take less time to perform repetitive activities, free up time for physicians to work on complex cases and communicate with patients, and ensure the accuracy and integrity of payer submissions. They also help organizations scale without having to add staff proportionately.
Health systems are facing rising costs, increased patient volume, and insufficient workforce. Tools that simply produce content cannot solve these structural problems; Only systems that are designed to perform multiple procedures simultaneously can do so.
A vision for system-wide collaborative adoption
The shift towards agent systems is already here. Organizations that move now will gain measurable benefits in operational efficiency, approval rates and staff retention. The first victories already show what this looks like in practice. .
In Catalonia, the public health system implemented an agent assistant called ALMA to bring evidence-based clinical guidance to physicians’ daily workflows. The results were surprising: 65% of users integrated it into their routine work, with a satisfaction rate of 98%. The program was extended to all primary care and is now positioned to expand to additional services.
Beyond administrative workflows, agent systems are also transforming clinical trial execution. Tempus deployed his TIME program that is an AI-powered network that orchestrates test matching, site activation, and patient enrollment in distributed care environments. The system analyzes clinical data to identify potentially eligible patients and then coordinates multiple agents: patient pre-screening algorithms display matches, clinical nurses review eligibility, and site activation workflows are activated in parallel with patient care. The TIME network has already generated measurable impact at scale: TriHealth Cancer Institute reported a 64% annual increase in patients enrolled in clinical trials, and Tempus TIME drove 95% of that growth. This agentic orchestration has addressed one of the most complex coordination challenges in healthcare: getting the right patients into the right trials at the right time.
The opportunity is clear. The question is not whether to adopt agent automation, but how soon you can start. Start here:
- Identify your high-volume, error-prone workflows (prior authorization, benefits verification, denial management)
- Assess where manual work is creating bottlenecks
- Look for vendors with proven interoperability, real-world deployment data, and transparent human protocols
- Pilot with defined scope and measurable metrics
If healthcare embraces this next phase with coordination and intention, we can build an administrative ecosystem that ultimately keeps pace with clinical innovation, restoring operational capacity, reducing avoidable delays, and ultimately strengthening the patient experience. This is the opportunity before us: to move from isolated efficiencies to a system-level transformation, driven by automation that can actually get the job done.
About Benjamin Easton
Benjamin Easton He is the co-founder and chief technology officer of Develop health. Builds software infrastructure that improves communication between healthcare providers and insurers, with a focus on reducing physician burden and improving patient access to medications. He was recognized by Forbes on their 30 under 30 list (2022) and has a long history of creating health-focused products and startups.
