An AI-powered wilderness rescue simulator.
https://summit-sim.bhamm-lab.com/
Summit-Sim uses human-in-the-loop review to generate curriculum-informed, interactive backcountry emergencies for dynamic Wilderness First Responder (WFR) training.
Wilderness First Responder (WFR) training relies heavily on static paper scenarios or expensive live-action roleplay. Students rarely get enough dynamic, unpredictable repetitions to truly test their decision-making under pressure. When the unexpected happens in the backcountry, textbook memorization isn't enough—responders need dynamic critical thinking.
Summit-Sim provides infinite, medically accurate WFR scenarios through a dynamic AI game loop. Instead of multiple-choice questions, responders use natural language to evaluate scenes, check vitals, and apply treatments. The system evaluates every action against a hidden medical "truth," evolving the scenario exactly as a real patient's condition would change in the wild.
- Infinite Scenario Authoring: Generate highly specific emergencies based on environment, group size, and difficulty level.
- Dynamic State Machine: The patient's condition evolves realistically based on the responder's timeline and medical interventions (or lack thereof).
- Human-In-The-Loop (HITL) Validation: Instructors review and approve scenarios before publication, with all feedback logged to MLflow for continuous improvement.
- Objective Debriefing: Post-simulation evaluation scores the responder's actions against established WFR protocols.
Summit-Sim is built with a sophisticated, production-ready AI stack focused on latency, strict output schemas, and agentic orchestration.
- Orchestration (LangGraph): Manages the complex state flows of the simulation. Utilizes
StateGraphfor the core loops,interrupt()for instructor HITL injections, and checkpointing for state persistence. - Agent Framework (PydanticAI): Ensures absolute medical safety and system stability by strictly enforcing all LLM inputs and outputs via Pydantic
BaseModelschemas. - Frontend UI (Chainlit): A fully asynchronous, reactive Python UI tailored for conversational AI, providing seamless UX for both scenario authoring and active gameplay.
- Observability (MLflow): Comprehensive LLM span tracing, variable logging, and feedback tracking to monitor agent reasoning and model performance.
The application is cleanly divided into two interconnected graphs joined by a shared Scenario ID:
- The Authoring Graph: Takes environmental parameters and dynamically generates the baseline blueprint (Setting, Patient vitals, Hidden Medical Truth, and Turn 0). Instructors review via HITL interrupt before scenarios go live.
- The Simulation Graph: A continuous game loop where the AI evaluates open-ended student actions against the hidden truth, dynamically updating the active scene state and generating the next narrative frame.
Three specialized PydanticAI agents power the system:
- Generator: Creates wilderness rescue scenarios from minimal configuration
- Action Responder: Evaluates student free-text actions and provides cumulative scoring (0-100%)
- Debrief: Generates post-simulation performance analysis against WFR protocols
Planned: MLflow automatic validation judges (Safety, Realism, Pedagogy) for medical accuracy assessment.
Built for the Weber State AI Hackathon 🐾
- https://www.scribd.com/document/38484292/Wilderness-First-Responder-Course
- https://wildsafe.org/wp-content/uploads/2021/11/CWS-%E2%80%93-WFR-Partcipant-Workbook-2021.pdf
- https://www.nols.edu/category/wilderness-medicine/case-studies/
- https://www.nols.edu/category/wilderness-medicine/case-studies/page/2/
- https://www.letsgoexploring.com/public/wilderness-first-responder-infosheet-side-A-2022.pdf
- https://www.wildmedcenter.com/uploads/5/9/8/2/5982510/standard_wfr_syllabus.pdf
- https://sierrarescue.com/wp-content/uploads/2013/12/WILDERNESS-FIRST-RESPONDER-General-Info.pdf
- https://sierrarescue.com/coursepdf/WFR.pdf
- https://www.nols.edu/wp-content/uploads/2025/08/23732-Student-Logistics.pdf
- https://www.deepsprings.edu/next/wp-content/uploads/2024/07/Student-WFR-Training.pdf