Every week, another healthcare organization announces a major AI initiative. Ambient documentation. Automated prior authorizations. AI-assisted care coordination. The investments are real, the intentions are genuine, and yet, many of the results often still disappoint. A global 2025 McKinsey & Company survey found that most organizations struggle to scale AI beyond the initial pilot phase, and KLAS Research has repeatedly documented a persistent gap between AI adoption and measurable clinical or operational impact.
The commonly cited problem is that organizations are deploying AI into broken processes. The popular prescription is to fix the process first, then layer in the technology. That framing is understandable, but it’s also insufficient.
AI doesn’t just demand better processes. It requires a complete redesign of traditional healthcare operations for a new level of precision, traceability, and governance. A new operating model that the vast majority of healthcare organizations simply aren’t built around today.
Organizational intelligence is key, but often nowhere near AI-ready
It is true that in the age of AI, every patient call, every scheduling decision, every intake handoff can become observable, measurable, and improvable. But that visibility only creates value if the decision logic underneath it is sound.
Part of why it rarely is: for most of healthcare’s operational history, process and data quality improvement initiatives were episodic. Organizations would sample a handful of prior authorization cases, identify patterns, recommend changes, and return months later to check progress, while the majority of workflows and data remained invisible and undocumented. Not because anyone was hiding them, but because humans cannot watch every single interaction at once.
While process mining tools have begun to change this over the last decade, organizations have largely continued to manage by approximation. The result is that institutional knowledge never got properly documented and structured. Instead, it got absorbed by people.
Clearly defined processes are rarely waiting to be automated. Often, institutional and tacit knowledge lives in the heads of staff: schedulers who’ve worked the same desk for ten or more years, care coordinators who know which physicians respond to a fax versus a phone call, and front-office personnel who’ve developed informal escalation paths because the official ones quietly broke down.
This is the organizational intelligence that keeps healthcare running, and it almost never lives in a document or database. And when those schedulers retire, and those care coordinators move on, that intelligence walks out with them, quietly eroding the operational foundation that any AI deployment will depend on.
The foundational work of AI-readiness is making that knowledge legible and actionable across the organization as a whole — for humans and machines. What are the actual decision rules? Who owns each step? What signals indicate a workflow is breaking down before a patient feels it? What happens if it still does?
If an organization cannot answer these questions with precision every single time, neither can technology. When organizational intelligence is properly documented and decision logic is clearly defined, organizations lay the foundation for AI to deliver the results it promises.
Without governance, AI becomes a liability
Another critical dimension that healthcare leaders consistently underestimate is governance – more specifically, the structures required to maintain organizational intelligence and sustain AI performance over time.
Successful implementations require codifying knowledge from frontline staff, leaders, and care coordinators into structured and optimized workflows that AI can support – but it doesn’t stop there. Leaders must establish a clearly defined governance framework to ensure the underlying data and process infrastructure updates in real time, so that AI is not reliant on institutional memory that has already drifted from reality.
An AI agent operating on scheduling rules that were quietly revised one week ago is not a technology problem — it is a governance failure. Fail here, and AI becomes a liability regardless of how sophisticated the LLM model is.
This framework must also account for a new kind of workforce onboarding: one where the new employee is an AI agent. As healthcare organizations begin to deploy AI alongside human staff, governance structures must define how those agents are trained in an organizational context, how their performance is monitored, and how they are updated as workflows evolve. The time to restructure onboarding for an AI workforce is now, so it can work safely and effectively.
Transformational change and sustainable ROI demand organizational groundwork first
The organizations that will capture the real ROI from AI are not the ones that merely deploy the most sophisticated models and voice interfaces. They are the ones that do the harder work first: surfacing the organizational intelligence that has always lived in people rather than systems, documenting decision logic with rigor, and building the governance structures to continuously update processes and keep data usable. Done well, AI strategy creates a more resilient and self-aware organization that understands how it actually operates and can improve continuously rather than episodically.
AI is a powerful enabler but to create transformational and sustainable change, organizations must be able to manage and maintain the process, data, and orchestration layer underneath.
Photo: Just_Super, Getty Images
Timm Schneider is the Co-Founder and COO of Third Way Health, an AI-enabled healthcare services company transforming front-office operations for medical practices, payers, and MSOs. Third Way Health ensures best-in-class service quality through its AI-enabled end-to-end operating platform, a meticulous implementation process, and top-tier personnel. Before launching Third Way Health, Timm spent over a decade in management consulting at EY, where he supported global Fortune 500 companies in large-scale operating model and digital transformation initiatives, including the design and implementation of Global Business Services models such as Shared Service Centers and Centers of Excellence. Through that work, he saw firsthand that technology alone doesn’t deliver results — it requires robust process, data, and governance structures to realize its full potential. Outside of work, Timm loves traveling the world with his wife and two daughters and is an avid runner.
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.
