The following is a guest article by Sagnik Bhattacharya, CEO at Rhapsody
Healthcare is entering an agentic era, one where software doesn’t just generate insight, but takes action.
Unlike earlier forms of AI, agentic systems are designed to plan, decide, and execute multi-step workflows across systems with minimal human intervention. What was once assistive is becoming operational. AI is no longer just sitting alongside clinicians and staff, it is beginning to act on their behalf.
That shift is profound, but it’s also widely misunderstood.
Much of the current conversation around AI in healthcare still centers on models: accuracy, performance, and the race to adopt the latest generation of large language models. But the limiting factor in this next phase will not be intelligence. It will be whether healthcare systems can support action.
Because agents don’t work in isolation. They depend on the ability to access data, move across systems, harness tools, and trigger workflows in real time. Without that, they don’t scale. They stall.
This is where the gap begins to show.
Healthcare data remains deeply fragmented across EHRs, imaging, claims, and digital health tools. These systems were never designed to operate as a unified environment. They were built independently, often with different standards, latency expectations, and governance models.
Fragmentation is the True Barrier to Scale
Agentic AI assumes that fragmentation has already been solved. It hasn’t.
The result is a growing mismatch between what AI systems are capable of and what the underlying infrastructure can support. An agent may be able to determine the next best action in a care pathway, but if it cannot reliably access the necessary data, or execute that action across systems, the value stops at insight. It never becomes impact.
In other industries, this problem is already emerging. Early deployments of agentic AI have struggled not because of model limitations, but because of incomplete data, weak integration, and insufficient governance. In healthcare, where accuracy, timeliness, and accountability are non-negotiable, those gaps carry far greater risk.
Real-time data exchange becomes critical in this environment. Agents operate continuously, not episodically. They require access to current, contextual information, not batch updates or delayed feeds. Latency, inconsistency, or gaps in data are no longer inefficiencies, they are failure points.
Governance Must Be Built into the System
As agents take on more responsibility, governance can’t live in policy documents. It has to be embedded directly into systems, defining what actions are allowed, ensuring traceability, and maintaining human oversight where it matters.
The shift is from managing software to managing behavior.
The organizations that succeed in this transition will not be the ones that adopt AI fastest. They will be the ones that invest in the infrastructure required to make it usable, connecting systems, standardizing data, and enabling workflows to operate across environments in real time.
Because in the agentic era, intelligence is only one part of the equation. Systems have to be able to act on it, and most aren’t built to.
Sagnik Bhattacharya is the Chief Executive Officer of Rhapsody, an industry veteran with experience spanning Epic, PatientPing, and HealthEdge, where he held leadership roles across population health, care coordination, and payer solutions. He serves on the Board of Directors for Carequality and is passionate about connecting clinical leaders and health system stakeholders to simplify data exchange and drive better patient outcomes. Rhapsody provides infrastructure for AI-ready interoperability, delivering healthcare integration, identity, and clinical terminology solutions to more than 1,900 customers worldwide. Learn more at rhapsody.health.
No comments:
Post a Comment