‘Ichibot’ by Tel Aviv Sourasky University Medical Center – Ichilov


Customer Stories

Ichilov is moving from Cloud-First to Agent-First and building a secure AI framework for the hospital of the future.

Organization: Tel Aviv Sourasky University Medical Center – Ichilov
Partner: AllCloud, Anthropic, AWS
Industry: Healthcare / Hospital Operations
Scale: Israel’s largest ER hospital · 2M patient visits/year · 8000 employees · 220 systems, 170 interfaces unified

Tel Aviv Sourasky University Medical Center – Ichilov, Israel’s largest emergency-care hospital, partnered with AllCloud and AWS to build an agentic AI operating layer on top of its cloud infrastructure.
Ichilov was the first hospital in Israel and Europe to fully migrate to AWS, serving 8000 employees and 2 million patient visits per year across 220 systems.

The centerpiece of the deployment is The AI Framework, including Ichibot an AI ecosystem where can develop any agent with any tool integration over AWS, one of the agents was Ichibot a Internal knowledge ai agent that runs on Amazon Bedrock using Anthropic’s Claude models for clinical and operational reasoning. Ichibot searches ServiceNow, SharePoint, and OneDrive simultaneously, answers staff questions with mandatory source citations, and autonomously opens and routes IT tickets.

Results: 25–30% reduction in manual IT inquiries, sub-60-second query resolution, and a 40% employee engagement target for year one.

Claude’s role: Claude powers the reasoning engine behind each agent which will be included in the OS Framework, including Ichibot, via Amazon Bedrock, serves as the unified conversational interface layer.

Cloud migration solved Ichilov’s infrastructure problem – reliability, security, and speed. It did not solve the human problem of fragmented knowledge, overloaded service desks, and staff time pulled away from patients. The Agentic Hospital OS, built on AWS ecosystem over Claude, Amazon Bedrock models at its reasoning core and governed through a dedicated AI Trust Zone, is designed to close that second gap – starting with IT support, and built to extend, by design, into clinical and administrative operations across the entire hospital.

Tel Aviv Sourasky University Medical Center (Ichilov)  is Israel’s largest emergency-care hospital and was the first hospital in Israel and Europe to fully migrate its infrastructure to AWS, as part of Project Nimbus  –  a program run under the strictest public-sector compliance standards. That migration alone delivered strong infrastructure gains: 80% fewer failures, a stronger security posture, faster innovation cycles, and shorter time to value.

Moving the infrastructure to the cloud enabled the hospital’s 8000 employees to actually work day to day efficiently by building an agentic operating layer on top of the migrated environment.

Ichilov wanted a foundation for adding AI agents quickly as new needs come up across the hospital. That foundation had to be secure by default, governed centrally, connected to existing enterprise systems, and able to scale to the entire workforce.

Solution Implemented

Ichilov worked with AllCloud to build The AI Framework, an agentic AI operating layer on top of its AWS environment. It is built  securely through a  dedicated AWS environment. The framework works as an ecosystem where the hospital can develop new agents with any tool integration, all sharing the same identity, security, guardrails, and observability.

Ichibot was the first agent built on the framework and marks the first milestone in Ichilov’s agentic roadmap. It is an internal knowledge and service assistant that searches enterprise knowledge sources simultaneously and answers staff questions with mandatory source citations. It also handles all logistics operations from text to actions. 

The Claude’s Role

Claude’s role is not incidental – it sits at the reasoning core of the system in distinct places:

As the reasoning engine behind the AI Ecosystem, running on Amazon Bedrock: Claude is the model performing the “advanced clinical and operational reasoning” step – reading retrieved content from ServiceNow, SharePoint, and OneDrive, and producing a cited, grounded answer to the staff member’s question, rather than an unsourced generation.

Category

  • AI and reasoning
  • Knowledge and data
  • Integrations
  • Security

Technologies

  • Anthropic Claude, Amazon Bedrock, Bedrock Agents, Bedrock Guardrails,
  • Amazon OpenSearch (vector search), Amazon S3, Aurora PostgreSQL, Amazon Agentcore Knowledge base, Arize AI Phoenix
  • ServiceNow, SharePoint, OneDrive and more
  • Microsoft Entra ID SSO, AWS KMS, AWS Secrets Manager

Ichibot’s Phase One rollout (IT service desk) has already produced measurable outcomes:

  • 25–30% reduction in manual IT inquiries
  • Under 60 seconds to resolve a typical query
  • 40% employee engagement target set for year one
  • First hospital in Israel and Europe to fully migrate to AWS, now positioned as an industry reference point

“The systems synchronized fully, and the results were visible immediately after launch.” – CEO, Ichilov Medical Center

Hospitals evaluating similar transformations are reportedly already looking to Ichilov as a reference implementation.