Health system executives are under pressure to move fast on AI. They are implementing tools for ambient documentation, agentic AI, clinical decision support, revenue cycle automation and more.
Seema Verma, executive vice president and general manager of Oracle Health and Life Sciences, argued at the Oracle Health and Life Sciences Summit on Sept. 23 that the industry is asking the wrong question. The issue isn’t whether to adopt AI but what healthcare is putting it on top of.
“This moment isn’t about making yesterday’s systems a little smarter,” Ms. Verma said. “It’s about building a fundamentally different foundation for healthcare designed for an AI future.”
That argument starts with data. Before a health system can build any AI strategy, it has to solve a more basic challenge of identifying the data it has, where it lives and how it flows across the organization.
“You can’t really have an AI strategy unless you have a data strategy,” she said. “The AI is only going to be as good as the data that it has available to be able to reason across.”
For decades, healthcare organizations have acquired technology in layers: EHRs on top of legacy billing systems, AI tools on top of EHRs, analytics platforms pulling from all of it, often without a unified data architecture underneath. The result is an industry of talented people working with technology that creates more manual effort than it eliminates, scatters information across disconnected systems and leaves clinicians and administrators with an incomplete picture of what is happening around them.
“Healthcare doesn’t just need AI,” she said. “It needs AI built specifically for healthcare, for the way healthcare works, the decisions it demands.”
Oracle has a solution. The company developed and deployed enterprise clinical AI agents to support physicians and nurses in a variety of workflows. Documentation burden reduction is the most visible: AI agents are generating clinical notes in real time, capturing what is said during a patient encounter without requiring the physician to build the record afterward.
“For clinicians, Oracle Health AI agents are already reducing documentation burden, increasing accurate coding, and helping clinicians spend more time with patients,” Ms. Verma said. “We’re expanding those capabilities with new AI agents supporting nursing, discharge care coordination, prior authorization and many other workflows.”
Oracle is applying the same model to the business of healthcare. At many hospitals, revenue cycle operations still require significant manual work across billing, contract management and care coordination. Oracle is targeting those areas for redesign rather than step-by-step automation.
“The goal is not to automate every step of a broken process. It is to redesign the processes across the entire health ecosystem,” she said. “This isn’t faster bureaucracy; it’s less bureaucracy.”
Delivering on that redesign requires a different approach to data connectivity. Oracle has been building out its interoperability infrastructure, including participation in CMS-aligned networks that make patient data available more seamlessly across institutions.
Payers are part of that interoperability strategy. There is persistent friction in how claims data moves between providers and payers, typically transferred in bulk, in forms that require significant manual effort to act on. Oracle is focused on streamlining these processes and reducing friction between payers and providers while its broader data strategy is aimed at changing the nature of what gets transferred.
“We want to help them be able to retrieve the data that they need in real time, not just by dumping a bunch of data off at the doorstep, but bringing it in organized format that gives them the summaries that they need so that they can do their work quickly and we can reduce administrative costs for the whole system,” Ms. Verma said.
The data strategy extends into life sciences. Oracle has launched the Oracle Life Sciences Data Intelligence Solution, which can aggregate more than 130 million anonymized medical records alongside external claims, genomics, pathogen and prescription data. Researchers and health systems can contribute their own data to the platform and use AI tools to accelerate analysis. A companion team, Oracle Research Services, works to convert that data into clinical evidence and insights.
A particular focus is clinical trial access. Only a fraction of eligible patients currently participate in trials and the matching problem, getting the right patient to the right study, is one Oracle is targeting as one of the central barriers through AI-driven enrollment tools.
“You’ve got both sides that are eager to work together, but yet there hasn’t been this convener,” she said. “You should be able to do a clinical trial right inside your EHR.”
For smaller institutions that have never run a clinical trial, the goal is activation. Oracle is working with pharma partners to bring community and rural hospitals into studies they don’t have access to now. The aspiration for patients tracks the same logic.
“That means not waiting for patients to find the clinical trial, but the clinical trial finds the patient,” Ms. Verma said.
At the Department of Veterans Affairs, Oracle is executing what Ms. Verma described as the largest EHR deployment in the world.
“We are building a modern foundation for care at national scale for those that have served our country, and this year alone will complete 13 sites, proving what technology, leadership and partnership can achieve,” Ms. Verma said.
Oracle has also redesigned its patient portal to provide a better and more engaging patient experience. The new version gives patients direct access to their health records and intelligence that helps them understand and use the information more effectively. They can ask questions about their health record and schedule appointments regardless of which system their provider uses.
On the model side, Oracle is not locked to any single AI partner. The company hosts multiple foundational AI models in its cloud infrastructure and leverages whichever performs best for a given application, including through its partnership with OpenAI. Keeping AI operations inside Oracle’s cloud also strengthens cybersecurity by reducing the exposure that comes with AI tools operating across external environments.
That model flexibility is paired with a rigorous internal governance process. Oracle requires all health AI capabilities to undergo a thorough review process before release that includes assessments for overall risk, design and data strategy, model evaluation, and privacy and security. The evaluations are supplemented with ongoing post-deployment monitoring and a standardized health model card to provide transparency to customers. This holistic process has been recognized with ISO 42001 certification — the gold standard for AI management and governance.
The focus on responsible and trustworthy AI development and maintenance processes is underpinned by 6 principles Oracle has published around health AI development. Chief amongst those is human-centered design, but that goes further than the phrase “human in the loop.” “It’s ‘human in command,’ where they’re really making the final decisions,” she said. “We look at AI as more of a tool; it’s something to support, not replace, physicians or providers or nurses.”
The healthcare workforce is under acute pressure with an aging population driving demand that the current staffing structure was not built to absorb, and time lost to documentation and administrative tasks is time unavailable for care.
“If they’re spending less time in front of the computer, that means that’s more time that they have for patient care and to see more patients,” Ms. Verma said.
Speed of delivery has become part of Oracle’s product proposition. Ms. Verma said the company is applying AI to the implementation process itself to shorten what she called “time to value,” or the gap between contract and realized benefit, because health systems under financial pressure cannot wait years to see results.
“We can’t build our products fast enough because the industry is so desperate for these changes,” she said.
The measure of whether any of it works, is not a performance metric on a dashboard. It is whether patients can feel the difference.
“Patients will not applaud the architecture. They won’t celebrate interoperability. They won’t care which cloud processed the data or which AI agent generated the insight,” she said. “But they will remember that the diagnosis came sooner, that their doctor already knew their story, that the right treatment reached them at the right time, that expertise could find them wherever they lived, that they spent less time navigating the system and more time living their lives.”
The post Oracle’s Seema Verma: Building ‘healthcare designed for an AI future’ appeared first on Becker's Hospital Review | Healthcare News & Analysis.