AI Agents in Healthcare: Incremental Gains or System-Level Transformation?
Healthcare leaders are hearing a familiar promise: AI will improve efficiency, reduce burden, and accelerate decision-making. However, across health systems, there’s a growing disconnect between expectation and reality. Many organizations are investing in AI and seeing only modest returns.
The reason is simple: most are applying AI to systems that were never designed for it.
As highlighted in a recent article in Harvard Business Review, early factories made a similar mistake with electricity, replacing steam engines without redesigning workflows, resulting in only marginal gains. Healthcare is now at that same inflection point with AI.
Two Paths Emerging in Healthcare AI
A clear divide is forming between organizations that are inserting AI into existing systems and those rethinking how their systems operate altogether.
The Incremental Model: AI as an Add-On
This is the most common approach today. AI is introduced to improve specific tasks without changing the underlying structure of how information drives the organization.
Typical use cases include:
● Clinical documentation and summarization
● Image analysis tools in Radiology and Cardiology
● Workflow automation (e.g. scheduling, prior authorization)
● Analytics dashboards layered onto existing data systems
While valuable, this approach tends to produce limited results. Efficiency improves at the margins, but core challenges, fragmentation, clinician burden, and delayed insights, remain.
The Rewired Model: AI-Native Operations
A smaller group of organizations is taking a fundamentally different approach. Instead of inserting AI into existing workflows, they are redesigning workflows from the ground up around AI capabilities.
This model focuses on:
● Structuring and unifying data so it can be directly used by AI systems
● Enabling systems to communicate with each other through AI-specific APIs rather than relying entrirely on human-driven information exchange
● Allowing AI agents to coordinate workflows across tasks and departments
The difference isn’t incremental, but rather exponential. In some cases, administrative processes that once took weeks or months can be completed in minutes when systems are fully integrated and AI-accessible. Clinical processes will be more difficult to transform, but stand to see similar gains in efficiency.
Why Healthcare Struggles to Realize AI Value
The challenge is no longer a lack of AI tools. It’s that healthcare infrastructure is deeply human-centric and based on personal trust. Systems were designed for people to read, interpret, and navigate, not for machines to process and act on. This creates three major barriers:
1. Fragmented Data
Clinical, operational, and financial data are stored across disconnected systems. Much of it exists in formats like PDFs, scanned documents, or narrative notes optimized for human workflows.
2. Workflow Friction
Clinicians and staff are required to move between systems, click through interfaces, and manually reconcile information. AI currently has to mimic these same inefficient processes rather than bypass them.
3. Limited Visibility
Insights are typically retrospective. By the time data is aggregated and analyzed, significant opportunities for intervention and improvement may have often passed, and people’s attention has focused elsewhere.
Why This Challenge Is Amplified in Cardiology
Cardiology is uniquely affected because of the volume and complexity of data involved in patient care.
A single patient journey may include:
● Imaging data (echo, CT, MRI)
● Procedural and hemodynamic data
● Registry and quality reporting requirements
● Financial data
● Longitudinal outcomes tracking
Despite this richness, these data streams are rarely connected in real time. As a result, opportunities to improve outcomes, efficiency, and resource utilization are often missed.
ASCEND’s Vision: Moving Beyond Aggregation
This is where the distinction between incremental and rewired approaches becomes critical.
Traditional healthcare analytics focuses on:
● Aggregating data
● Building dashboards
● Supporting retrospective reporting
ASCEND’s approach is different. It focuses on creating a foundation for real-time, AI-enabled decision-making by:
● Unifying access to cardiovascular data across systems
● Providing AI agents with the underlying knowledge framework that allows them to work intelligently across disparate data and workflows.
● Enabling cross-domain insights that connect clinical, operational, and financial performance
● Building AI agents that help providers and administrators in ways they understand and to which they can adapt, rather than building a black box.
The goal isn’t just automation of the existing siloed workflows, it’s transitioning to a system where AI actively enables the healthcare enterprise to deliver better care..
A Practical Comparison
Capability | Incremental AI Model | Rewired Model |
Data | Fragmented, human-readable | Accessible to AI agents within a context-aware framework |
Workflow | Human-driven, AI-assisted | Agent-enabled, human-supervised |
Insights | Retrospective | Real-time and predictive |
Productivity | Incremental gains | Transformational impact |
Clinician Role | Task execution | Judgement, decision-making and oversight |
The Path Forward for Health Systems
Organizations that succeed with AI will not be those that adopt the most tools, they will be the ones that rethink how their systems operate within the burgeoning AI ecosystem.
That shift will involve:
● Connecting healthcare systems with AI-ready APIs
● Deploying agents that work affectively across this data, within grounded knowledge framework, and within workflow optimized for human performance.
● Embedding governance and verification into AI-driven processes
These changes don’t require a complete overhaul overnight, but they do require a shift in mindset, from optimizing tasks to redesigning systems.
Final Thought
AI is reshaping our world, and we are just entering the middle of the beginning stages.. The successful healthcare organizations of the future will fundamentally change how care is delivered through systematic transformation.
Comments