What $750 million in healthcare automation actually looks like

By Jack Corbell, a senior AI automation practitioner — with direct experience delivering enterprise automation programs across major US and UK healthcare providers

Numbers like $750 million have a way of sounding abstract. Let me make it concrete.

That figure represents the measurable value VOPS has delivered to a single US healthcare provider over a multi-year automation program. It is not a projected or modelled number. It is the sum of outcomes that are live, auditable, and continuing to grow as the program scales into new areas.

I want to use this piece to explain how programs of that scale actually happen; the answer is considerably more interesting, and considerably more instructive.

Healthcare as a proving ground

Healthcare is one of the hardest environments to automate well. The regulatory requirements are among the most stringent of any sector. The consequences of errors are not financial write-offs — they are patient safety incidents. The people you're asking to work alongside AI agents are clinicians whose primary concern, rightly, is patient care, not operational efficiency.

And yet healthcare is also one of the sectors where intelligent automation delivers the most transformative value, precisely because the administrative burden is so severe and the capacity constraints are so acute. Across the US health system, a significant proportion of clinician time is consumed by documentation, authorization processes, and administrative coordination that has nothing to do with clinical care. That is both a problem and an opportunity.

The referral management program

One of the anchor programs we built was in home health referral management. This is the process by which patients transitioning out of hospital settings are referred into home health, hospice, and palliative care services.

Before automation, this process was largely manual. Referral data arrived through multiple channels in inconsistent formats. It required significant human effort to capture, validate, and route correctly. With 250,000+ referrals per year flowing through the system, the manual processing load was enormous — and the error rate, while not catastrophic, was high enough to create downstream problems in care coordination.

The automation we built handles the full end-to-end process: ingesting referral data from multiple sources, applying validation rules, routing to the appropriate service line, flagging exceptions for human review, and updating the downstream care management systems. It runs continuously, processes without fatigue, and applies consistent logic every single time.

The outcome: the equivalent of 38 full-time staff freed from referral administration, delivering over $25 million in value across five years. More importantly, the solution now spans multiple agencies across the provider network and continues to scale. The architecture was designed for growth from the outset.

The medication adherence program

A second major program addressed prescription adherence, the process of ensuring patients are filling their prescriptions on time and following through on treatment plans across multiple care pathways.

Non-adherence is one of the most significant and underappreciated problems in healthcare operations. Patients who don't fill prescriptions generate avoidable readmissions, deteriorating chronic conditions, and downstream care costs that dwarf the cost of the medication itself. The challenge is identifying at-risk patients early enough and reaching them through the right channel at the right time.

The automation we deployed manages outreach across the patient population — identifying patients approaching refill deadlines, initiating contact through the appropriate channel, tracking responses, escalating to a care coordinator when needed, and updating the patient record. It runs at a scale and consistency that is simply not achievable through manual outreach.

The result: $29 million in value over five years, the equivalent of 28 FTEs in saved effort, and measurable improvement in prescription adherence rates that directly affect patient outcomes.

The pharmacy reauthorization program

A third program tackled reauthorization workflows for chronic condition medications such as lipid management, hypertension, and diabetes. These are high-volume, high-stakes processes where delays in reauthorization result in patients running out of medication.

Before automation, the reauthorization process involved significant manual review: clinicians and pharmacy staff checking eligibility, validating clinical criteria, managing documentation, and processing approvals. The backlog was persistent. The solution automated the end-to-end workflow — reading clinical data, applying reauthorization criteria, generating documentation, and routing for final clinical sign-off where required.

Outcome: $25 million in value, 39 FTEs in equivalent capacity, and the elimination of the chronic backlog that had been a persistent operational problem.

What makes healthcare automation work

Across all of these programs, a few principles have been consistent.

First, clinical involvement from the outset. Every automation we have built in a healthcare setting has been designed in close collaboration with the clinicians and care coordinators who use it. Their process knowledge is irreplaceable. Their trust is essential. Automation that is built without them tends to be worked around; automation built with them tends to be championed.

Second, human oversight built into the design. These are not black-box systems that make decisions autonomously. They are systems with clearly defined decision boundaries; what the agent can do, what it flags for human review, and what escalates to a clinician. That governance is not a constraint on the automation's value. It is the reason it has been trusted to operate at scale.

Third, a managed service model. The program described above are not one-off implementations. They are live, managed, continuously monitored deployments that evolve as the clinical environment changes. The value is not in the initial deployment, it is in the sustained operation and continuous improvement over time.

Healthcare is difficult. The stakes are high. But the opportunity is real, and for organizations willing to approach it with the right rigor and the right partnership, the outcomes are extraordinary. $750 million in value from a single provider is not an anomaly. It is what a well-designed, well-managed, long-term program looks like.