Solving chronic care fragmentation may be agentic AI’s biggest opportunity
A layer of automated coordination and informed execution can help organizations rethink how they approach the chronic care delivery process.
For many patients living with complex conditions such as diabetes, the everyday experience of chronic care management is one of deep and frustrating fragmentation.
Patients managing diabetes move between providers, suppliers, payers, educators and support teams while navigating prior authorizations, supply reorders, documentation requirements and coverage changes — often with minimal guidance or support. Every handoff introduces the potential for delays, interruptions in care and the need to repeat the same information over and over again.
For organizations supporting these patients, the operational burdens of moving patients through the system are also profound. Chronic care management generates high volumes of critical data stored in multiple disconnected systems, from customer relationship management systems and knowledge bases to order management platforms and clinical records.
The result is predictable — documentation gaps, processing delays, handoff errors and inconsistent patient experiences, particularly during demand surges that strain staffing capacity.
Building a seamless experience
The imperative to create a more seamless experience is leading healthcare organizations to explore agentic AI, a new type of technology designed not only to generate information, but also to actually execute operational tasks across enterprise environments.
Agentic AI systems differ from traditional conversational AI because they are designed to take action inside workflows. Rather than simply answering questions, these systems interact with the patient and can retrieve information from multiple business systems and determine what task needs to happen next. They can invoke APIs, surface missing documentation and route work appropriately across operational processes.
This offers a layer of automated coordination and informed execution that can help organizations rethink the way they approach the entire chronic care delivery process.
Introducing an orchestration layer
The real challenge right now for chronic care isn’t how to store data or even how to access it. It’s how to orchestrate the necessary activities using that data to create a process that offers patients what they need exactly when they need it.
This is where agentic AI can play a role in supporting tasks that historically have required human agents. It’s humans who have had to navigate multiple disconnected systems while coordinating fulfillment of services or medical devices, such as continuous glucose monitors and insulin pumps.
Some early efforts have shown that agentic AI can be effective and serve as an important tool in reducing operational fragmentation, which can improve continuity of care for patients within healthcare systems.
As an example, CCS has led an effort to develop such a platform, collaborating with Deloitte and Adaptive ML to implement and test AI agents capable of executing operational tasks directly across enterprise systems. The effort leans heavily on "function calling,” which enables large language models to reliably interact with external platforms such as data management systems, knowledge bases and authentication tools.
This is critical in healthcare environments, because a malformed parameter or failed API call does not simply generate an inaccurate answer – it can stop a workflow entirely, forcing escalation to a human agent and introducing additional delays into already complex patient interactions.
To address this challenge, these organizations applied reinforcement learning-based fine-tuning to a compact open-source model deployed on AWS infrastructure. The resulting system achieved more than a 90 percent reduction in latency, compared with the baseline of a proprietary model.
Enabling near-real time responses can improve operations and facilitate a more seamless patient journey. Success here – especially as supported by millisecond response times – can determine whether operational AI systems are viable in real-world workflows.
The missing link between disparate systems
The lessons from this proof of concept informed the broader rollout of an enterprise-wide agentic AI platform developed by CCS, which can support multiple operational workflows for organizations.
Agentic AI platforms, such as this one, hold the potential to automatically process most intake documents, enabling the identification of documentation gaps that can delay patient onboarding and approvals of reorders. Platforms have the potential to independently handle approximately a portion of incoming patient calls while helping reduce call handling times, providing the potential to automatically process large numbers of intake documents.
But the benefits aren’t just about low-latency automation or faster patient call times. They’re also about continuity and building more robust connections between disparate systems. Many healthcare AI deployments still operate as isolated tools layered onto disjointed workflows. Agentic AI systems introduce the possibility of maintaining operational context as work moves across systems and between human and digital agents.
Platforms can be designed around a division of responsibilities between AI agents and human support teams. AI systems can manage high-volume, repetitive and rules-based work, while human agents can focus their efforts on complex cases, exceptions and patient interactions requiring judgment or empathy.
With this approach, patient cases that require escalation can be surfaced by systems with additional context from previous activities so support staff have a head start on solving problems.
That distinction is particularly important in chronic disease management, where patients may interact with organizations repeatedly over long periods of time and where operational disruptions can directly affect adherence and continuity of care.
A more unified approach
Agentic AI deployments highlight a broader shift taking place across healthcare AI strategy. Increasingly, organizations are moving beyond pilots and chatbot-style interfaces to evaluate whether AI can reliably coordinate operational workflows across environments. But scaling operational AI inside chronic care environments will require governance frameworks that mature just as quickly as the technology itself.
Today, healthcare organizations are defining how AI systems are validated and where human review remains necessary so that there is confidence in AI-powered actions. In chronic care management, where continuity of care is inseparable from outcomes, supporting and delivering this continuity of care is the opportunity in front of us to ensure meaningful long-term value across the chronic care journey.
Richard Mackey is the chief technology officer at CCS, where he leads the technology organization and digital strategy to support collaborative care programs and home‑delivered supplies for people with chronic conditions, especially diabetes.