Oracle Rebuilds Its EHR Infrastructure to Support AI-Native Workflows

Oracle has launched a new electronic health record (EHR) platform that diverges entirely from the architecture of Cerner, which it acquired in 2022 for $28 billion. As reported by CNBC, the platform is browser-based, cloud-native, and designed to support AI-driven workflows, including voice queries, ambient documentation, and clinician-adaptive interfaces.

Key functions—such as Oracle’s Clinical AI Agent, which auto-generates clinical notes from recorded visits are already live at more than 70 sites. The system can answer freeform, patient-specific questions ("Has this patient had a lung CT?") and adapt to clinicians' ordering habits over time. But Oracle executives emphasized that this is not a user-interface redesign layered on old infrastructure:

"You’re not going to put new things on top of crumbling infrastructure,” said EVP Seema Verma. "What we’re introducing is something brand new."

That architectural choice reflects a larger problem, which is that most EHRs today struggle with brittle configuration systems—often hundreds of interlocking modules, templates, and manual governance steps that limit how deeply AI can be integrated. In Cerner, maintaining a single intake form could involve up to 12 distinct configuration tools and a week of specialist time. Hence Oracle’s decision to rebuild, as a direct response to that complexity.

The approach aligns closely with the "Elastic EHR" model proposed in JMIR AI (May 2025), which outlines a five-tier framework for embedding AI across the EHR stack in autonomous database tuning, internal config optimization, behavior-driven workflow suggestions, and real-time clinical guideline integration, all in addition to the interface level. 

See also: The Rise of AI Agents in Biotech, Where Are We Now?

The core argument is that AI copilots are only as effective as the systems they sit on. If the underlying configuration is out-of-date, inconsistent, or manually governed, surface-level tools can’t reliably deliver their intended value.

Topics: AI & Digital