AI Won’t Transform Health Systems, CEOs Will — 6 Principles to Drive the Transformation

Early in my career, at a large nonprofit academic medical center, we were losing thousands of patients a year because our doctors were refusing hospital transfers. I listened to tapes of physicians lecturing outside doctors about why they should handle cases themselves. With the backing of the CEO, we changed the policy overnight, which more than solved the problem. Not because the plan was brilliant — because leadership made it a top priority.

In every leadership role I’ve held since I’ve seen the same thing. The team doesn’t follow the plan on the whiteboard. They follow whatever the leader is paying attention to, and whether that leader is willing to make hard calls and absorb the consequences.

The transformation that AI is enabling demands the same kind of leadership. If AI is not one of the CEO’s top two or three priorities, the organization will produce pilots and slide decks while better-mobilized peers pull ahead and the institution’s ability to fulfill its mission erodes. In a $5.3 trillion industry where administrative overhead consumes a third of spending, the value to be created over the next decade is enormous — and it will require the CEO to determine whether that value is created on the organization’s terms or extracted by others. Given the structural pressures bearing down — workforce shortages, administrative bloat, and reimbursement compression — the opportunity cost of inaction is higher than it has ever been. The trajectory compounds the stakes: national health spending is heading toward nearly twenty percent of GDP by 2033, a national-security concern given the federal deficit and the other societal needs competing for the same dollars. Healthcare benefits have become one of the fastest-growing expense lines for employers, eroding corporate competitiveness and suppressing take-home wages — and we are paying more for care that is not delivering better outcomes.

AI is the only cross-cutting lever I have seen that bears on cost, quality, workforce, and access simultaneously. The pace of AI advancement is compounding in months, not years, and the gap between organizations that are mobilizing and those still evaluating is widening fast.

The organizations that capture that value will be the ones where the CEO and board treat AI as a governance-level priority — where the CEO’s own evaluation is tied to measurable AI outcomes, and accountability cascades through the executive team via KPIs, compensation, and promotion. Mobilizing requires not a single decision but a set of imperatives, each building on the one before it, and in many cases requiring the right partners to execute at the necessary speed and scale, pursued with curiosity, calculated risk-taking, and urgency:

1. Own the transformation personally. This is the CEO’s responsibility — not the CIO, not a consultant, not a committee. The CEO needs to set direction, build enough fluency to make capital allocation decisions, adjudicate resource conflicts, and is accountable to the board for quantitative results. That accountability cascades through every member of the ELT.

2. Govern with speed, not consensus. This is the imperative most at odds with how health systems traditionally run. A small cross-functional group is needed — clinician, finance, operations, legal, technologist — chaired by the CEO with direct board visibility. If it turns into a senate where approvals take longer than building the pilot, the structure has failed.

3. Build a transformation unit, not an innovation center. Health systems are conglomerations of hundreds of businesses built for stability, not nimbleness. What’s needed is an operating unit that drives enterprise-wide transformation — mapping cost structure with precision, identifying functionally verifiable work where automation delivers the fastest returns, evaluating the vendor landscape, and applying the build-buy-partner decision for each domain. Staff it with operators who know unit staffed by operators who know where the money goes and have personally felt the friction in the current state. Their job is to map the cost structure, find the work where automation will pay back the fastest, and decide who is going to do it. More often than not, that work should be done by a partner rather than being an internal build. A company running any strategic, but non-core, service for dozens of health systems can spend more on engineering than any one of those systems would on its own, and will see far more data in the process. Look for partners who have already made that investment and will hold themselves to outcomes and costs which will come down over time. Keep internal development for problems where nothing adequate exists.

4. Start non-clinical and build trust through outcomes. Revenue cycle, supply chain, workforce management, IT, call centers — these are functions where output is easily verified and measured, and where partnership models can drive performance and cost deflation quickly. For each, apply the framework: what should we build, buy, or partner on — and where can external firms innovate faster and bring engineering talent and data scale that organizations cannot replicate on their own? Success builds confidence before the organization confronts higher-stakes clinical questions.

5. Extend into clinical domains carefully, with evidence gates. Same framework, but add clinical validation, regulatory pathway mapping, and workforce transition planning. Prioritize domains where AI already performs at or above human baselines — diagnostic imaging, ambient documentation, clinical decision support, predictive risk stratification — and build from there. Clinical AI advances the safety mission, but the implications for patients and caregivers demand rigor and transparency with your clinical workforce.

6. Measure everything, then scale what works. Track cost-to-serve trajectory, quality, time-to-deployment, and workforce impact across every domain. Tie these directly to executive compensation — KPIs, bonuses, promotion, and resource allocation flow to leaders who deliver and away from those who don’t. Every successful pilot needs a predetermined pathway to enterprise-wide deployment. The CEO evangelizes what’s working, demands business unit leaders perform against AI-enabled benchmarks, and is accountable to the board for results.

The opportunity — and the risk of inaction

The structural advantages health systems hold — the patient relationship, trust, regulatory standing, workforce infrastructure, and deep community roots — are real but depreciating. As AI capability grows and its cost declines, the barriers that historically protected incumbents are becoming solvable problems for well-capitalized technology companies that do not share our mission obligations. Organizations that don’t act will watch the value get extracted by others rather than created on their own terms.

The build-buy-partner framework that runs through these imperatives reflects a core conviction: health organizations do not need to do this alone, and they should not try. Partnership is not weakness. It is a recognition that this transition’s scale and speed exceed what any single institution can execute alone, and that the right structures create value for both parties while keeping the organization in control. The danger is the opposite posture — an “elbows-up” conviction that a system can do it all itself and become, overnight, a software-development organization building enterprise-grade agents for mission-critical work. That execution risk should not be underwritten by the health system; it should be expected of, and contracted to, expert partners who underwrite both the execution and the value.

The next few years will determine which organizations lead this transformation and which are reshaped by it. Those that don’t lean in will find the transformation happening to them, on someone else’s terms. Those that do — with the right partners, the right governance, and a CEO who makes this consequential for everyone around them — will redefine what a health system can be.

Photo: Sergey Khakimullin, Getty Images

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.