Google Cloud Opens Gemini for Healthcare Preview in the U.S.

By Sophie Lindqvist

Google Cloud opened a U.S. preview of Gemini for healthcare on November 10, 2026, for providers and payers. The industry-specific system extends Google’s universal Gemini agent with healthcare skills and data connectors. It aims to assemble clinical evidence and coordinate administrative work across records, claims, and coverage tools already in use.

That promise targets a documented burden rather than an abstract AI benefit. Google cited a 2024 Harris Poll for Google study in which clinicians reported nearly 28 hours of administrative work per week, compared with 34 hours for medical office staff and 36 hours for claims staff. The online U.S. survey ran from August 26 to September 9, 2024, among 821 healthcare providers, 209 payors, and 2,079 adults.

The survey supplies pre-launch workload context rather than a performance result for Gemini. Its figures establish the case for reducing assembly work, but they do not show how much time the healthcare preview saves. The product still has to prove that its citations, safeguards, and connectors can shorten real workflows without displacing clinical responsibility.

Google is putting an agent close to sensitive medical and coverage data while keeping people responsible for consequential decisions. The preview is available to Gemini Enterprise Standard and Gemini Enterprise Plus customers. It extends the universal Gemini agent rather than arriving as a standalone healthcare application. Google introduced that underlying agent on October 8, 2026, when healthcare was still labeled “coming soon.”

What can Gemini do in clinical workflows?

Google says healthcare skills package instructions and relevant context from domain experts so Gemini can execute defined, multi-step tasks. Health systems can also create custom versions using their own policies and standard operating procedures. That customization makes the agent’s operating rules a local governance choice, not just a vendor setting.

For providers, Google says the system can:

• Combine electronic health record (EHR) data with outside records to build a cited patient history across care settings.
• Separate active problems from resolved ones.
• List allergies and identify medication discrepancies.
• Check for intervening events, current medications, and outstanding clearances, laboratory work, or consents before a scheduled visit or procedure.

The cited history provides the mechanism behind the system’s clinical utility. It gives reviewers a traceable account that joins internal records with outside information, while the pre-visit check focuses on changes that could affect a visit or procedure. Google’s documentation still leaves chart entries, orders, and final sign-off with clinicians.

In short, Gemini assembles and checks context; it does not finalize clinical decisions.

Can it manage coverage and appeals?

For authorizations and appeals, Google says Gemini assembles chart evidence against payer policy and identifies documentation gaps before submission. After a denial, it drafts an appeal with citations only when the clinical record and plan policy support one. That condition limits the agent to evidence-backed arguments rather than a ready-to-send appeal without support.

For payers, Gemini checks requests against medical policy and organizes traceable findings for clinical reviewers. Its member-service workflows can estimate out-of-pocket costs, track deductibles, and verify network status using plan documents and eligibility data. In each case, Google is positioning the agent to organize evidence for people who retain approval authority.

This is administrative coordination, not autonomous coverage policy. The agent prepares the evidence trail, and reviewers still decide whether a request should proceed.

Which systems can it connect to?

Google’s initial Epic and Oracle Health integrations are read-only. Epic connects through open.epic APIs, while Oracle Health uses Ignite APIs for FHIR, a clinical-data interoperability standard. Neither integration writes back to the underlying patient chart. Product documentation specifies that chart entries, orders, and final sign-off remain clinician-led.

Beyond Epic and Oracle Health, Google lists these connectors:

• Keragon connectivity to more than 400 healthcare applications.
• Current Centers for Medicare & Medicaid Services coverage determinations.
• The National Provider Identifier registry.
• The RxNorm drug vocabulary.
• ICD-10-CM diagnosis codes.

Availability is uneven across that catalog. Optum Network Eligibility and Coordination of Benefits remain in limited release and testing, according to Google. NextGen Healthcare connectivity remains in development and is slated for release in the coming months. Healthcare connectors support global and U.S. locations during the preview.

The practical consequence is a staged system rather than a universal link to every clinical platform. A broad catalog can still produce gaps when individual data connections remain limited or unfinished.

Who is trying Gemini?

Google named Blue Cross Blue Shield of Massachusetts, Hartford HealthCare, HCA Healthcare, Highmark Health, Northwell Health, Optum, and Stanford School of Medicine as launch customers. The American Society of Clinical Oncology is listed as a contributor. HCA Healthcare said it is exploring how Gemini could synthesize patient information across multiple EHR systems and care settings.

Stanford School of Medicine is testing a research-oriented use case. Deputy chief information officer Todd Ferris said: “Our clinical data is only as useful as researchers’ ability to ask questions of it,” He described researchers querying an Observational Medical Outcomes Partnership database in plain language without first requiring SQL or data-model expertise.

That workflow could make structured clinical data more accessible to people who do not write database queries. It also gives Google a test of question handling beyond care administration, alongside the product’s administrative and coverage uses.

The customer list spans insurers, providers, and research, while HCA Healthcare’s stated role remains exploratory.

How are data and models governed?

Google says the platform is built to support HIPAA compliance. Patient records, clinical data, prompts, and outputs remain within the customer’s governed Google Cloud environment. Google says that data is not used to train or fine-tune Google’s foundation models.

The controls carry existing access rules into connected workflows. Connectors inherit existing identity and role-based permissions, while citations and audit logging let reviewers inspect outputs and activity. Google also lists customer-managed encryption keys and VPC Service Controls among the security controls.

Customers can choose models from Google, Anthropic, and OpenAI under one governance framework. According to Google, its routing sends high-volume administrative tasks to faster, lower-cost models while reserving higher-reasoning models for complex analysis. That split puts cost and latency choices behind the same router rather than requiring each customer to build a separate model workflow.

The preview offers pricing through per-user seat subscriptions or pay-as-you-go. Google has positioned model selection, permissions, auditing, and billing within the same enterprise product. The aim is to make an agent useful across existing systems without giving it unrestricted access or final decision-making authority.

The governance story rests on three linked controls: customer-governed data, inherited permissions, and inspectable outputs.

Can it deliver on the administrative promise?

The launch puts agentic AI into the assembly work around care and coverage while leaving chart entries, orders, clinical review, and final sign-off with people. That boundary reflects the read-only integrations and Google’s support for HIPAA-compliant workflows. It also keeps the product’s claim narrower than fully autonomous healthcare.

The preview’s practical test is whether cited, permission-aware workflows save time without weakening oversight. That test becomes harder where connector coverage remains incomplete, including Optum’s limited-release eligibility and benefits tools and NextGen Healthcare’s unfinished connection. For launch customers, the next proof points are working connections, verifiable citations, and measured time savings in administrative workflows.