Insights by Dr. Chethan Sathya, Vice President of Strategic Initiatives at Northwell Health
If you are a regular listener to “The Big Unlock” podcast, you will notice a bit of a pattern in our conversations about “AI in healthcare.” Many episodes are heavy on ambition, light on execution. They celebrate breakthroughs, but they skip the messy middle: the place where promising tools either become part of daily care or quietly fade out after a pilot.
That is what makes this recent episode where hosts Rohit Mahajan and Ritu M Uberoy, both Managing Partners at BigRio and Damo, sat down with Dr. Chethan Sathya, Vice President of Strategic Initiatives at Northwell Health, worth a listen!
The episode is not built around hype, demos, or speculative futurism. It is built around the operating truth that healthcare is not short on ideas. It is short on integration, adoption, and scalable implementation. Dr. Sathya’s ubique “center of gravity” is implementation. He explicitly frames himself as an implementation scientist who focuses on making ideas usable in real-world clinical environments, and he’s clear that this is where healthcare innovation succeeds or fails.
The conversation stays anchored to what real clinicians will tolerate, what real systems can absorb, and what leaders must do to move beyond experimentation into repeatable operational value.
The Real Scalability Problem Isn’t the Model
The primary message that Dr. Sathya had was this, scalability is not primarily a model-performance question. It’s an implementation question. He explained, “Many organizations can find smart tools. Many can run pilots. Many can produce dashboards that look promising. But healthcare is a high-friction environment. The clinical day is full. The operational machine is complex.” He then added “one more thing is, it’s not neutral—it’s costly.
That’s why Dr. Sathya emphasizes a core scaling truth, “If it’s not built for clinicians… it’s not going to work. If AI isn’t designed for clinicians and doesn’t fit their workflow, it will fail to integrate and won’t scale.”
He went on to describe how “integration” and “scalability” are less about preference and more about physics. A tool that requires extra steps, extra context switching, or extra training becomes another load on already overloaded teams.
Integration Beats “Innovation Theater”
One of the most refreshing parts of the conversation is how plainly Dr. Sathya talks about integration. He describes a common failure pattern that many healthcare leaders will recognize immediately: a new tool arrives that requires clinicians to take on another app, another login, another workflow, another tab, another mental shift.
Even if it’s “good AI,” it’s still friction. As he said to Ritu, “If I have to download another app… it’s not integrated into my workflow.”
Dr. Sathya then explained that from his perspective and experience, the barrier is not just inconvenience. The barrier is that healthcare workflows are already dense, and clinicians are already managing multiple systems, constraints, and interruptions. An extra step doesn’t feel like “an extra step.” It feels like something that competes with patient care.
Ambient Documentation is a Proof Point for Scalable AI
As the conversation continued, Dr. Sathya offered ambient documentation as a good concrete example. Why? Because ambient documentation represents a category of AI that is scaling for a very clear reason, it solves a daily pain point and fits the natural flow of care.
He notes that ambient documentation can replace scribes for many clinicians. That matters operationally because scribes are often viewed as a practical relief valve for documentation load. If a tool can reduce that burden, the value is immediately understandable.
“Ambient listening is a great example of what I have been talking about. It works, it’s easy to use, it’s succeeding right now because it’s integrated into a lot of our workflows and that’s why it’s replacing scribes for a lot of clinicians.”
Springboarding from this example, he went on to describe what scalable AI looks like in healthcare:
- It removes a task clinicians already want removed.
- It works in the background.
- It fits the normal visit experience.
- It produces value without requiring clinicians to become tool operators.
There’s also a lesson here for strategy. Dr. Sathya explained how many AI efforts target “advanced” clinical tasks first. But the fastest scaling opportunities are often the basic burdens—documentation, routing, scheduling, triage, and administrative work that eats time every day.
Scalability Requires Cooperation, Not Just “Buy-in”
Dr. Sathya also said that “cooperation” has to be a cornerstone of effective implementation and scalability. He explained to Rohit that even when a tool is effective, scaling it across a large health system requires alignment across multiple groups:
- clinicians and clinical leadership
- operations and workflow owners
- IT, security, and infrastructure teams
- compliance and governance
- training and change management
- sometimes revenue cycle and finance
A lack of cooperation produces predictable outcomes:
- “local wins” that never spread
- inconsistent practices
- tool sprawl
- duplication of effort
- fragile adoption that fades when champions move on
In contrast, cooperation enables standardization: the ability to take what works in one place and make it repeatable across many sites.
This is why “AI scalability” is not only a technology initiative. It is an operating model initiative.
“If you want AI to scale, you need cooperation that is structural, not personal,” Dr. Sathya said. “Without cross-functional cooperation, the default outcome is fragmentation: pockets of use, inconsistent practices, and tool sprawl. With cooperation, AI becomes an operational asset rather than an IT experiment.”
The Next Wave: Agentic AI Will Raise the Stakes
As the interview drew to a close, Dr. Sathya predicts that more autonomous, “AI agents” AI will begin to take off, and he links that to workforce and operational implications.
“Agentic AI is going to take off, and I think that will significantly enhance our jobs. But it will also lead to some workforce disruptions that we will have to expect and be prepared for. How we train people and adapt to this next phase is going to be what this year is all about.”
This is a key point for scale-minded leaders. The more autonomous the system becomes, the more important it becomes to define – what the system is allowed to do, where it must ask for approval, how it escalates exceptions, who monitors quality over time, and how accountability is assigned.
Agentic AI will not scale safely through enthusiasm alone. It will scale where governance is mature, workflows are clear, and cooperation is strong.
The Takeaway – Scaling What Actually Works
Dr. Sathya’s message is refreshingly practical: healthcare doesn’t need more AI demos. It needs more integrated tools and more cooperative operating models. The organizations that win won’t be the ones with the most pilots. They’ll be the ones that can standardize, support, and spread what works—because their workflows, teams, and governance are built for scale. Dr. Sathya’s strong focus on practical innovation leaves you with a practical checklist based on his unique implementation lens.
- AI doesn’t scale as an add-on. If it requires extra apps, extra steps, or extra friction, adoption fades fast.
- “Built for clinicians” is the scaling requirement. Usability and workflow fit are not optional if you want sustained adoption.
- Time saved drives adoption. In the real world, clinicians adopt what reduces burden and is easy to use.
- Ambient documentation is a model example of scalable AI. It removes daily work and fits the natural visit flow.
- AI can help solve evidence overload. It can reduce the burden of staying current by surfacing and digesting clinical information at scale.
- Agentic AI will raise the bar for governance and workforce readiness. More autonomy means more need for clear boundaries, accountability, and cross-team alignment.