What If We Rebuilt Healthcare Operations for AI?

Transcript

Healthcare has a technology problem, but maybe not the one most people think. Health plans have technology, lots of it. Claims platforms, CRM systems, enrollment systems, contact center technology, workflow tools, automation, analytics, bots, and increasingly, artificial intelligence. The problem isn’t necessarily that healthcare needs another piece of technology. The problem is that much of healthcare’s technology was designed for a world that no longer exists, a world where the system’s job was to organize information and a human’s job was to do almost everything else. This is Healthcare Power Ups, a podcast that looks underneath healthcare operations to understand the people, processes, technology, and economics that make the system work. Not the PowerPoint version, but the real version.

To kick things off, this episode asks a bigger question: What would healthcare operations look like if the industry stopped adding AI to workflows designed decades ago, and started designing the workflows for AI? To see why this matters, imagine an associate processing a healthcare claim. A claim falls out of automatic adjudication and lands in a queue. The associate’s job is to figure out why. So they open one system, then another. They look at the claim, pull up a standard operating procedure, and search for the right policy. Maybe they need authorization information, provider information, or to understand the member’s benefits. They make a determination and document it. If something still isn’t right, the claim may move to another team. That team has another workflow, another standard operating procedure, another queue, maybe another system, and another human being who needs to figure out what happened before the work arrived.

Now, none of this is necessarily broken. In fact, this is how healthcare operations have worked remarkably well for decades. But looking beneath the surface, the entire workflow assumes a human is doing the work. That isn’t an accident. Many of the core systems supporting healthcare operations today were designed ten, twenty, or even thirty years ago. Their fundamental design philosophy was straightforward: technology moves information, and people execute the tasks. Over time, healthcare became much better at this. Analytics were added, robotic process automation arrived, bots were built, and individual pieces of workflows were automated. But fundamentally, the industry was improving a system designed around humans. Then generative AI arrived, and suddenly, another possibility emerged. Instead of asking how technology can help this person perform the process faster, healthcare organizations can ask: If this process were designed today, which parts would require a person at all?

That’s a very different question. Consider a healthcare workflow containing two hundred individual tasks. Historically, humans might perform most of them. But some of those tasks are predictable, like retrieving information, comparing fields, summarizing a document, checking a rule, routing a case, generating correspondence, or updating a system. These are exactly the kinds of activities increasingly capable of being performed or supported by AI agents. But healthcare isn’t an Amazon order. Sometimes there is clinical nuance, significant financial impact, regulations that apply differently depending on the state, conflicting information, or moments where someone simply needs to make a judgment. So the future isn’t AI replacing the healthcare operation. It’s AI changing where humans belong inside the healthcare operation.

In this model, AI agents can perform clearly defined tasks, while humans remain in the loop where clinical judgment, compliance requirements, regulatory complexity, financial consequences, or ambiguity demand expertise. But something has to connect the two. That is where another concept becomes critical: orchestration. Healthcare’s other enormous challenge is fragmentation. Take appeals, for example. An appeal sounds like one process. A member or provider disagrees with a decision, someone reviews it, a determination gets made, and it’s done. Except that isn’t remotely what happens. Information has to be extracted, policies identified, medical documentation reviewed, a determination made, correspondence generated, and deadlines monitored. Different systems and different teams may own different parts of that journey. Even someone responsible for an appeal end to end may depend on multiple other teams or systems to actually complete it.

The Economic Paradox From Transaction Fees to Outcome Accountability

Each individual piece may work, but the lifecycle doesn’t necessarily work as one system. And that’s a recurring theme across healthcare. Claims has a process, appeals has a process, utilization management has a process, provider services has a process, and member services has a process. Each function becomes more efficient inside its own four walls. But the member doesn’t experience four walls, nor does the provider. They experience healthcare. That is the paradox of operational efficiency. A health plan can have highly optimized departments and still have a deeply inefficient customer journey. Silos can exist at the process level, the business unit level, or the application level. When those silos prevent information and action from flowing across the healthcare lifecycle, they constrain the organization’s ability to improve both administrative and medical costs. To see this in practice, consider a provider calling a health plan about a claim.

In the traditional operating model, the health plan knows how many calls it receives. A service provider might look at that volume and say: There are this many calls, this many people are required to answer them, and here are the service levels. Perhaps automation can make the operation somewhat more efficient, but the fundamental unit being managed is still the call. Now, change the question. Instead of asking how efficiently these calls can be answered, ask: Why are providers calling in the first place? That creates a completely different operating model. Maybe providers repeatedly call about claim status, which analytics can identify. Better digital capabilities might eliminate some of those calls entirely, a virtual agent may handle another portion, and an AI assistant can help human agents resolve complicated calls more quickly. Some conversations, an appeal for example, may absolutely belong with a person. Now the organization isn’t simply optimizing a call center. It’s orchestrating the provider experience by combining IVR redesign, virtual agents, digital capabilities, analytics, and human advocates.

As a result, the cost of individual interactions can fall. But more importantly, the partner stops being accountable merely for staffing calls and starts becoming accountable for the outcome. That subtle change has enormous implications. Healthcare outsourcing has historically been built around transactions: How many calls? How many claims? How many appeals? How many full time employees? But consider what happens when the partner responsible for processing a transaction is also responsible for eliminating transactions that never needed to happen. The incentive changes. If a company is paid for every call it answers, fewer calls are actually bad for business. But if the company is accountable for the entire outcome, there is every incentive to eliminate the call that never needed to occur. That is where AI begins to change more than productivity. It changes the economics of healthcare operations. Instead of buying capacity, health plans can increasingly buy outcomes. Rather than simply saying we will process your claims…

…the conversation becomes: We’ll improve the speed and accuracy of the claims lifecycle while reducing unnecessary rework. Instead of promising to answer provider calls, it becomes: We’ll reduce the cost of provider engagement while improving the provider experience. Instead of paying exclusively for transactions or staffing, health plans may increasingly consider models based on broader operating outcomes. The partner becomes responsible not only for performing work, but also for finding ways to prevent unnecessary work. With that shift in mind, return to the claim. A claim falls out of automatic adjudication. Traditionally, an associate finds the standard operating procedure, searches for information, makes a decision, and routes the exception, which might eventually land in rework. Now imagine that workflow differently. The claim arrives, AI agents gather relevant information, and they execute clearly defined tasks. The workflow knows which steps can happen autonomously…

…it knows where a check is required, and it knows where healthcare expertise is needed. When judgment matters, it brings a human into the workflow with the information necessary to make that decision. Then the process continues, not as a collection of disconnected queues, but as a unified claims lifecycle. That is the fundamental shift. The workflow itself is redesigned: What should an AI agent do? What intelligence does it need? Where should automation occur? Where should a healthcare domain expert intervene? And what outcome is the entire lifecycle designed to produce? But there is something easy to miss in all of this. The AI may not be the hardest part. There will be powerful AI platforms, sophisticated models, agents, and tools. But a model doesn’t inherently know how healthcare operations should work, nor why claims rework happens in the first place.

It doesn’t inherently know why an enrollment falls out, why provider calls keep increasing, which decision carries clinical risk, which regulatory requirement applies, where an error upstream creates cost downstream, or which part of an apparently inefficient process exists for a very good reason. To redesign healthcare operations, technology is essential, but so is healthcare knowledge. The challenge isn’t simply determining what AI can do; it’s understanding what AI should do and where people still matter most. This is the idea behind Sagility’s Synchrony approach. Synchrony isn’t simply about adding technology to an existing process. It brings together technology, people, process, analytics, and healthcare expertise to orchestrate work across the lifecycle. The technology and orchestration layers matter, but healthcare domain knowledge determines what should actually be orchestrated. This is where three parts of Sagility’s transformation strategy connect. CoreIQ provides intelligence…

…using healthcare and operational data, analytics, and root cause insights to understand what is happening across the operation and where value can be created. SmarTec enables action, with AI agents and assistive technologies performing or supporting tasks within healthcare workflows. And Synchrony provides orchestration and outcome accountability, bringing intelligence, action, people, and process together across an end to end healthcare lifecycle. Another simple way to think about it is that CoreIQ helps understand WHAT needs attention, Synchrony determines HOW the work should come together, and SmarTec helps DO the work. Intelligence, Orchestration, Action. They are not three disconnected technology products, but three layers addressing different parts of the same operating problem. That distinction becomes especially important when organizations talk about AI transformation. There is a temptation right now to look at every healthcare process and ask where AI can be added. But that may be the wrong question. A better question is:

If AI had existed when this process was designed, would the process have been designed this way at all? Would claims rework be a separate function? Would people spend so much time searching multiple systems for information? Would providers call about claim status as often as they do? Would appeals move through the same sequence of handoffs? Would healthcare operations still be organized around the same functional silos? Maybe some of them would, but probably not all of them. And that’s the bigger transformation: not automating the old operating model, but reimagining it. There will still be people. In fact, the people remaining in these workflows may require more healthcare expertise, not less. There will be new skills, new management models, new ways of measuring performance, and new ways of defining accountability. Because the ultimate measure of transformation won’t be how much AI was deployed, nor how many agents were built, and definitely not…

…how many people were removed from the process. The measure is much simpler. Did the member have a better experience? Did the provider get an answer faster? Was the claim processed more accurately? Did unnecessary rework decline? Did the health plan reduce unnecessary cost? Did the outcome improve? That is the idea underneath Synchrony: AI led orchestration of healthcare operating workflows at the lifecycle level, with accountability for the outcome. There is another reason this model matters. Healthcare changes constantly. Regulations, CMS requirements, clinical policies, products, and member expectations all change. Traditional technology architectures can be extraordinarily difficult to modify. While a manual process can change relatively quickly, the underlying technology often cannot. An orchestration layer can potentially change that equation. Instead of rebuilding an enormous core system every time the operating environment changes, the workflow around those systems can become more adaptable. That doesn’t eliminate regulatory complexity, legacy technology, or the need for human expertise, but it can make the operating model more responsive to change. Which brings the story back to where it started. Healthcare has a technology problem, but the answer isn’t simply more technology. Healthcare is entering a world where technology can increasingly perform work that existing operating models were never designed to give it. For decades, systems were built around people, and then automation was added to help those people work faster. Now, there is an opportunity to do something different: start with the outcome, work backward through the lifecycle, decide what technology and AI should do, decide where human expertise matters most, and then orchestrate those pieces together. The future of healthcare operations isn’t human or AI; it’s understanding what each does best and rebuilding the operation around that answer. And that is this Healthcare Power Up: one piece of healthcare operations pulled apart, put back together, and made a little easier to understand. Because the most important question about AI in healthcare operations may not be where AI can be added. The better question may be: If healthcare could redesign the operation today, knowing what AI can do, what would it build differently? Once that question is asked, the conversation stops being about automating the process that already exists and becomes a conversation about creating a better one. This is Healthcare Power Ups, where the complexity of healthcare is taken apart to better understand what actually happens next.