We’re launching Agentic ED to provide a high-fidelity simulation layer for hospital operations.

The Context: DispoRx is an application designed to reduce Emergency Department wait times by orchestrating real-time patient flow and coordination.

The Agentic Layer: To ensure the system holds up in high-stakes environments, we’ve built LLM-based agents that log in and operate within the platform. These agents take on two primary roles: Attending and Consulting physician.

Allowing administrators to:

- Test workflow changes in a sandbox before deploying to production.

- Train staff on the platform using realistic, agent-driven simulations.

- Validate system performance under extreme load or surge conditions.

- Identify bottlenecks and optimization opportunities by observing how "agent-users" navigate the current workflow.

Standard testing tells you if your code works; Agentic ED tells you if your process works. We would love to hear your thoughts on using agentic users for QA and system validation!

Premature send - Also is there any logging or validation around why agentic decisions are being made?

This is a super interesting platform play. Question do you guys use synthetic data or how do you protect pii?

dk970