{"slug": "how-we-built-eir-a-secure-healthcare-agent-fleet-on-google-cloud", "title": "How We Built EIR: A Secure Healthcare Agent Fleet on Google Cloud", "summary": "A team of developers built EIR, a secure multi-agent healthcare platform on Google Cloud, for the All Things Agentic Hackathon. EIR coordinates specialized agents for patient access, recovery, scheduling, and hospital operations, using Google Agent Runtime, ADK, Gemini 3.5 Flash, and FHIR R4, with features like persistent memory, identity-based security, and human approval for critical actions. The team emphasized that useful agents require more than an LLM, including memory, authorization, and observability.", "body_md": "For the past month, our team has been building **EIR: Healthcare Agent Fleet**, a secure multi-agent platform for patient access, recovery, scheduling, and hospital operations.\n\nWe created this article specifically for the purpose of entering the **All Things Agentic Hackathon**.\n\nOur goal was to move beyond a single healthcare chatbot and build a real fleet of agents that can collaborate over time, remember context, interact with hospital systems, and stop when human approval is required.\n\nHospitals already use many software systems, but workflows are still fragmented.\n\nA patient may need to schedule an appointment, recover after discharge, answer a follow-up, or reach a clinician. At the same time, hospital teams manage reviews, inventory, scheduling, and supply operations.\n\nEIR coordinates these workflows through specialized agents rather than asking one model to do everything.\n\nThe Patient Access Agent runs on **Google Agent Runtime** using **Google ADK and Gemini 3.5 Flash**.\n\nWe integrated:\n\nMemory Bank lets EIR remember safe preferences between sessions. For example, a patient can say they prefer Main Clinic and afternoon appointments, then return in a new session and have EIR use that preference when searching for availability.\n\nAgent Identity and Agent Gateway give the agent its own least-privilege identity and govern outbound calls. Model Armor adds another security layer, while backend RBAC remains the final authorization boundary.\n\nEIR includes separate experiences for patients, clinicians, and operations teams.\n\nPatients can:\n\nClinicians can:\n\nOperations teams can:\n\nScheduling is backed by **FHIR R4 on Google Cloud Healthcare API**.\n\nLong-running workflows use **Firestore, Pub/Sub, Cloud Scheduler, and Cloud Run**.\n\nRecovery Episodes can last for days or weeks.\n\nEIR schedules follow-ups, processes structured patient responses, checks adherence and risk signals, and escalates cases when a clinician should take over.\n\nWe deliberately avoid autonomous diagnosis.\n\nCritical actions use deterministic policies and human approval.\n\nThe same pattern also applies to hospital supply workflows. Procurement agents may draft purchase orders, but a person must approve them before execution.\n\nWe also integrated **Vertex AI Veo** to generate short recovery education videos from already-approved care instructions.\n\nThe generative model does not invent diagnoses, medications, doses, or new clinical guidance.\n\nThe biggest lesson was that useful agents need much more than an LLM.\n\nPersistent memory, identity, authorization, observability, asynchronous execution, and human approval all matter.\n\nBuilding EIR forced us to treat agents as real software actors instead of isolated chat interfaces.\n\nLive demo:\n\n[https://eir-ui-658898892127.us-central1.run.app/](https://eir-ui-658898892127.us-central1.run.app/)\n\nSource:\n\n[https://github.com/ybedoyab/eir](https://github.com/ybedoyab/eir)\n\nBuilt by Yulian Bedoya, Juan Quintero, Marycielo Berrio, and Alejandro Feria Gonzalez for the **All Things Agentic Hackathon**.", "url": "https://wpnews.pro/news/how-we-built-eir-a-secure-healthcare-agent-fleet-on-google-cloud", "canonical_source": "https://dev.to/ybedoyab/how-we-built-eir-a-secure-healthcare-agent-fleet-on-google-cloud-27h1", "published_at": "2026-08-31 22:51:18+00:00", "updated_at": "2026-08-31 23:23:05.642043+00:00", "lang": "en", "topics": ["ai-agents", "ai-products", "ai-infrastructure", "developer-tools", "generative-ai"], "entities": ["Google Cloud", "Google Agent Runtime", "Google ADK", "Gemini 3.5 Flash", "Vertex AI Veo", "FHIR R4", "Yulian Bedoya", "Juan Quintero"], "alternates": {"html": "https://wpnews.pro/news/how-we-built-eir-a-secure-healthcare-agent-fleet-on-google-cloud", "markdown": "https://wpnews.pro/news/how-we-built-eir-a-secure-healthcare-agent-fleet-on-google-cloud.md", "text": "https://wpnews.pro/news/how-we-built-eir-a-secure-healthcare-agent-fleet-on-google-cloud.txt", "jsonld": "https://wpnews.pro/news/how-we-built-eir-a-secure-healthcare-agent-fleet-on-google-cloud.jsonld"}}