The artificial intelligence (AI) powering your hospital's diagnostic tools, your insurer's risk models, and your employer's wellness platform is only as good as the data feeding it.

And that data, by nearly every available measure, is a mess. Per Vorro , about 60% of healthcare organizations report that inaccurate data directly affects their clinical decision-making.

Fragmented across thousands of disconnected systems, locked in outdated formats, missing entire chapters of patients' health histories, it is the foundation on which a multi-trillion dollar AI transformation is being confidently built.

That is what makes the quiet appointment of Dr. Jaime Bland as Chief Data Strategist at the Make America Healthy Again (MAHA) Institute worth a much closer look than it has received.

As founder and CEO of Aquila, Dr. Bland brings a rare operational lens to the MAHA Institute role. Before launching Aquila, she served as President and CEO of CyncHealth, where she spent years inside the infrastructure problem, working across state agencies, health systems, and federal partners to move data that was never designed to move.

She has experienced exactly where the AI promise breaks down, not in the sophistication of the models but in the catastrophic fragmentation and incompleteness of the data those models are consuming and policy that is antiquated and not ready for modern approaches to technology and governance.

Her appointment signals that MAHA Institute, an organization centered on longstanding issues in the healthcare space. MAHA institutes focus on chronic disease, prevention and public-private partnerships that are influencing better health outcomes, understanding that in grand detail is something most health care organizations would rather you didn’t: the limiting factor in healthcare data is not the compute power or algorithmic elegance.

It has always been the data layer beneath it.

This is the priority for MAHA Institute to address the inconsistencies in policy that has allowed for bureaucratic governance to flourish and data inaccessible to people, providers and policy makers.

In chatting with Shannon Kennedy, Healthcare Chair and Senior Fellow at The Digital Economist, she told me that "Taxpayers paid roughly $35 billion through Meaningful Use to put an Electronic Health Record (EHR) in every hospital, and what they got back were monolithic systems that don’t play nice with any innovation they can't control. But something important is shifting: for the first time, patients can use AI to decode the medical terminology and pricing that healthcare has hidden behind for decades. When patients meet the institutional innovators finally taking on the data and governance layer, that's where the real unlock happens."

The AI Investment Is Built on a Cracked Data Foundation

Consider what AI in healthcare is actually working with. Patient records scattered across hospitals, urgent care clinics, telehealth platforms, and specialty practices that rarely communicate with one another. Longitudinal health histories, that often have a third of the information inaccurate or a copy and paste from previous encounters, the continuous, time-sequenced data a predictive model genuinely needs, are the exception rather than the rule.

Millions of records remain tied to paper charts and physical media that no algorithm can touch. Himss and ONC data shows fewer than half of U.S. hospitals regularly engage in interoperable exchange across the core domains of sending, finding, receiving, and integrating patient data. The AI sitting on top of that infrastructure is leveraging data with gaps and inaccuracies. Without a good foundation of data, AI, at its best, is making educated guesses from incomplete evidence and calling it intelligence. The stakes are visible in real deployments. Universal Health Services (UHS), one of the largest health systems in the nation, launched AI agents with Hippocratic AI to handle post-discharge patient follow-up calls across its facilities, and is now acquiring Talkspace for $835 million specifically to build a behavioral health AI companion trained on clinical data. UHS CEO Marc Miller acknowledged the system is still in the "early innings of this AI game," a candid admission that even at scale, the data foundation required to make these tools perform is still being built.

For business leaders, the financial stakes are not abstract. National health spending has reached $5.3 trillion, representing 18% of U.S. GDP. Employers, insurers, and health systems are pouring capital into AI tools premised on the assumption that the underlying data is reliable enough to act on. In most cases, it is not, and the gap between what AI vendors promise and what fragmented data infrastructure can actually deliver is where that capital quietly disappears.

The private sector's biggest players are beginning to reach the same conclusion. Oracle Health, one of the largest healthcare technology companies in the world, became a CMS Aligned Network just this week, with Seema Verma, its executive vice president and former CMS administrator, stating plainly: "AI works well when it has good data underneath it. If we don't have complete data, we will not be able to leverage AI.

For Oracle, interoperability and an AI strategy go hand-in-hand." Oracle has also secured TEFCA Qualified Health Information Network designation, building toward a system where patient data access is as straightforward as checking a bank balance. That a company of Oracle's scale is making interoperability its central AI argument is not a coincidence.

Arcadia , the healthcare data and outcomes platform backed by Nordic Capital, is making the same foundational bet, recently expanding its leadership team specifically to address the intersection of data, AI, and policy. Arcadia currently connects more than 2,600 sources of data and manages over 170 million patient records. The companies willing to solve the data layer first, at every level from startup to enterprise, are positioning themselves to define what healthcare AI actually becomes.

Governance And Data Are the Competitive Advantage Nobody Is Talking About

What Dr. Bland is advocating for at the MAHA Institute, a framework centered on transparent and accessible governance, interoperability that is not limited by the EHR capabilities, and longitudinal records that follow individuals across care settings, is the precondition for AI that actually contributes to improved health outcomes. Longitudinal data transforms what AI can do. A model trained on episodic, visit-by-visit records identifies patterns within encounters. A model trained on complete health histories across years and care settings identifies trajectories, predicting where a patient is heading and where intervention changes the outcome before a crisis drives the cost.

As Dr. Bland told me: “The fragmentation problem is not unsolveable, the challenge is we need a cohesive approach to what the real issues are; the technology has been capable for some time, the root cause of the challenges lie in the varied implementations and fragmented systems that have really no obligation to send your data in digital form anywhere. The ONC has policy efforts underway to solve part of it but the authority only extends so far, we are missing a significant amount of data and focusing more on quality data outputs that are correct and representative of the care is where the policy needs to shift, right now the documentation follows the money not the person.”

Oracle’s Verma echoes this directly: the Electronic Healthcare Record (EHR) companies themselves have historically been part of the problem, blocking broader data exchange while the industry made incremental progress on common standards. Real data flows between organizations remain inconsistent, limiting the usefulness of every AI tool that depends on complete information. What is shifting now, with MAHA Institute, Aquila Health, Oracle, and Arcadia all converging on the same diagnosis, is that the infrastructure argument is finally winning at the policy and enterprise level simultaneously.

That shift from reactive to predictive is where genuine ROI lives for every organization currently spending on healthcare AI.

And it requires someone willing to do the infrastructure work that does not generate press releases or keynote moments, building governance frameworks, establishing data standards, expected outcomes, negotiating interoperability across jurisdictions, and ensuring individuals have real digital access to their own complete health records, is the foundational work and it's not built into the incentives that are rolling out.

The Department of Health and Human Services (HHS) doesn’t have to take the lead here, employers and Governors could take the lead , define how data is going to work for people where they seek healthcare, would go a long way to curbing the rising cost of care.

Business leaders who grasp this dynamic have a meaningful advantage. The organizations that will extract real value from healthcare AI in the next five years are investing now in the governance and data quality layer, because that is what separates AI that performs from AI that merely impresses in a demo.

The frontier has moved with data and AI, and it is not where most executives are looking.