Case Study
Experity scales urgent care patient engagement with Care Agent
Care Agent is Experity's HIPAA-compliant, multi-agentic workflow patient assistant. Deployed across 2,000+ clinics, it automates discharge, lab delivery, intake, and document access over web and SMS while keeping PHI secure and staff focused on care.
“Care Agent doesn't just communicate; it lets you build lasting relationships. Staff is doing the work that improves experience and throughput rather than repeating administrative tasks.”
The Problem
Urgent care clinics face a constant wave of repetitive, high-volume tasks: patient intake, discharge instructions, lab result delivery, and document sharing. Those workflows overload front-desk and clinical staff, drive up costs, lengthen phone wait times, and hurt patient satisfaction.
Experity needed a solution that could handle this at national scale without adding another system for staff to babysit, and that could do it in a HIPAA-compliant way.
Solution at a Glance
Care Agent is a dual-channel patient engagement assistant: SMS/RCS for registration and notifications, secure web chat for all PHI-bearing interactions: triage, discharge notes, lab results, payment collection, and care reminders. It integrates deeply with Experity's EMR and practice management systems so that context flows from first touch through post-visit follow-up.
- HIPAA-compliant by design — Identity verification, audit logging, and data handling are core to the system, not add-ons.
- Skills architecture — Clinics turn on the skills they need (e.g., digital discharge, lab results) in a marketplace model; the underlying orchestration and session state are defined and constrained by a deterministic architecture..
Architecture
Care Agent runs on AWS: Amazon Bedrock for model execution and agent orchestration, AWS Lambda for serverless workflow coordination, and Amazon S3 for secure document storage and session persistence.
The architecture defines the session model, agent routing, and schema-validated handoffs, and it connects to Experity's EMR APIs, OAuth2, and real-time and SMS gateways. The stack stays serverless, scalable, and within Experity's HIPAA-eligible boundary.
Outcomes
Care Agent is live across more than 2,000 urgent care clinics, serving approximately one million unique patients and over 3.5 million chatbot sessions. The system now handles roughly 68% of patient discharge interactions, cutting administrative load, improving response times, and raising satisfaction, all while operating in a HIPAA-compliant environment.
In under six months, the deployment drove over $10M in AWS compute consumption from an initial $125K investment. In practice, that has meant thousands of hours of staff time reclaimed. By automating routine follow-ups, document delivery, and lab result communication, Care Agent has saved the equivalent of tens of thousands of additional patient visits in capacity.
Patient engagement metrics have strengthened as well: Experity's existing patient engagement solutions already drive an average of 5.7 additional visits per day per clinic; with Care Agent expanding into a full platform of skills across the journey, Experity expects even greater gains in efficiency, revenue potential, and patient experience.
Lessons Learned
- Modular design at scale — Keeping agent workflows modular (skills, handoffs, clear boundaries) made it possible to operate and evolve the system without disrupting live patient care.
- Compliance as core — Identity verification, audit logging, and data handling had to be built-in from day one. Treating HIPAA as a core product requirement, not a later checkbox, was essential for rollout across enterprise clinics.
- Deterministic when it matters — For structured tasks (intake forms, payments, document submission), deterministic UI components outperformed pure conversational flows. A3's hybrid approach (deterministic UI + probabilistic agents) matched what urgent care workflows actually need.
Why A3
GenUI built A3 - the Agentic App Architecture - based on the lessons we learned helping Experity build Care Agent from the ground up. A3 is built for full-stack, production-ready agentic systems: chat-based agent logic, deterministic and probabilistic workflows, API integrations, session memory, and orchestration in one stack, similar to how WordPress delivers a complete site out of the box, but for multi-agent conversational AI.
A3 establishes a solid foundation for agentic applications in healthcare at scale:
- Session-first multi-agent routing — Pre-visit, in-visit, and post-visit "skills" (discharge, lab results, intake, payments) behave as specialized agents that hand off within a single patient conversation. No flowchart authoring; routing is dynamic and context-driven.
- Deterministic + probabilistic in one experience — Structured actions (forms, payments, file uploads) render as deterministic UI components inline in the thread; conversational AI handles intent, triage, and natural language. That hybrid keeps critical EMR and compliance workflows reliable while preserving a chat-native feel.
- Deploy and evolve without re-architecting — A3 was designed to work with multiple providers and execution environments and to evolve as the AI ecosystem changes. Experity can adopt new models and capabilities without rebuilding the core.
See how A3 can work for you
Care Agent is a product of Experity. A3 is an open-source framework for multi-agent chat applications by GenUI.
Want to go faster with A3? Talk with GenUI.