Agentic AI demo for EHR intelligence

A live demo showed LangGraph agents orchestrating multiple large language models with Neo4j and SambaNova to analyze electronic health records in near real time. (youtube.com) Agentic AI runs multiple autonomous "agents" that plan, call external tools, and maintain state—like a coordinated team of specialists answering multi-step questions. (langchain.com) SambaNova’s walkthrough video, published about two months ago, runs three different LLMs on a single SambaStack node and shows agents generating and executing Cypher queries against a Neo4j graph of EHRs. (youtube.com) Data Science Dojo promoted the demo on social channels and lists a hands-on workshop and a ten-week Agentic AI bootcamp that cover LangGraph, Neo4j, and SambaNova integrations. (academy.datasciencedojo.com) SambaNova’s materials explicitly highlight on‑premise deployments for sensitive healthcare data, and Neo4j’s documentation recommends knowledge graphs to ground AI for accuracy and explainability. (sambanova.ai) Technically, LangGraph provides stateful orchestration for long‑running multi‑agent workflows; Neo4j represents patients, encounters, and medications as nodes and relationships, and agents use Cypher plus node embeddings to find similar patients. (langchain.com) Vendors point to performance and lower external API exposure on SambaStack, while independent tutorials and GitHub repositories note integration complexity, the need for governance, and careful access controls when working with real EHRs. (sambanova.ai) SambaNova published the demo video and integration code, Data Science Dojo is teaching the stack in workshops, and several example repositories are available for engineers who want to reproduce the EHR agentic workflow. (youtube.com)