Fig. 1

Evolution from ChatGPT-based prototype to agentic AI workflow. a Initial ChatGPT-based scenario designer. A structured, custom interface powered by ChatGPT, specifically trained with simulation content and scenario templates. Educators interact with the AI via a chat-based interface within the ChatGPT platform, guiding the model through a series of prompts to generate scenario elements. The LLM generates an editable Canvas that allows users’ refinements and finalization. Limitations are text-based outputs, limited knowledge retrieval, and lack the capacity for external system integration or automated visual aid generation. b Advanced agentic workflow in n8n. Featuring an AI agent that autonomously makes decisions (research context, create scenario) and takes actions without constant human oversight. Educators still interact with the system via a chat-like interface but with greatly expanded accessibility. This interaction can occur through various channels, including local or web-based pages and mainstream messaging systems. The knowledge is an unlimited vector database, utilizes diverse tools, and seamlessly integrates external APIs. The key action is the generation of formatted files containing the complete simulation scenario and supplementary materials, such as arterial blood gases (ABGs), lab exam results, and imaging findings, delivered directly to the user, that has the task to review the material and finalize the scenario