Google.org Funds Two-Year AI Health Assistant for Clalit's 5 Million Patients

Google.org has awarded Clalit Health Services a two-year grant to build an AI-powered conversational platform that will serve the organization's 5 million patients — Israel's entire largest healthcare network, by far the country's biggest.

The "Digital Front Door" assistant will handle patient intake, symptom triage, appointment preparation, booking and follow-up care. Clalit frames a clear operational payoff: by collecting information before appointments, the system frees medical teams to spend more time on care itself rather than on administrative intake.

Clalit Director General Prof. Eytan Wirtheim said the organization's diverse patient population is central to its work, and that Google.org's support combined with Google Cloud technology will let Clalit deliver care "in patients' own languages and in ways that fit their needs."

Why voice, and why dialects?

The design decisions in this project target a specific failure point. Israel's communities speak many languages and dialects, but conventional digital healthcare services depend on formal written language. That makes online tasks effectively unusable for people who primarily speak local dialects. Clalit reports that independent online appointment booking rates remain low across its patient base.

The platform's answer is speech-first. It will run large language models, including Google Gemini, with speech recognition tuned for local dialects and cultural differences. Patients will complete intake, symptom triage and appointment booking by speaking the language they use in everyday life — no written forms, no formal register.

What about patients without smartphones?

The architecture deliberately reaches below the smartphone tier. Clalit plans to connect an interactive voice-based AI service directly to landlines and standard mobile phones. That covers patients who lack internet access or whose handsets cannot run standard web applications or even receive SMS-based services.

The goals are concrete: expand participation in digital services and increase the share of appointments confirmed before visits, which reduces no-shows and scheduling waste.

Boaz Maoz, Managing Director of Google Cloud Israel, said conversational AI tailored to local dialects and multiple communication channels can help remove barriers to healthcare access. Google Cloud will back the project with its infrastructure and help Clalit develop a model that other healthcare systems can adapt.

How is patient data protected?

The platform runs on Google Cloud's secure infrastructure under a Zero Trust security framework with strict data privacy controls, according to Clalit. Every patient interaction with the assistant will be:

• encrypted in transit,
• stripped of personally identifiable information,
• anonymized before further processing.

For a health system handling 5 million patient records, those controls are the difference between a deployable service and a regulatory liability.

Will the model spread beyond Israel?

After an initial pilot, Clalit plans to expand the platform nationwide. The more consequential move is what follows: Clalit intends to release its voice AI architecture and its cultural adaptation frameworks as an open-source model for public healthcare systems. Wirtheim confirmed that Clalit plans to make the model available to healthcare organizations worldwide through that open-source architecture.

If delivered as planned, the resources would let health organizations in other countries adapt similar tools for their own linguistic and cultural contexts without starting from zero — turning a national deployment into reusable public infrastructure.

The takeaway: a two-year grant, Gemini-based dialect-aware voice AI, and an open-source release plan make this one of the most transferable public-health AI deployments now in motion.