{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"The Root Cause - Business of Medicine Podcast","title":"Episode 20: Guardrails for the AI Frontier — Innovating Without Betting Your License with Travis Garland","html":"","width":"100%","height":180,"duration":4112,"description":"In the most tech-forward episode yet, brother doctors Erik and Davin Lundquist sit down with Travis Garland — co-founder and COO of My Compliance Citadel and a two-decade healthcare-operations veteran (United Healthcare, UF Health, and earlier work that earned an award for redesigning Florida's foster-care system) — to map the collision of AI, innovation, and compliance in medicine.\nTravis breaks down a distinction most clinicians miss: a GPT prompt simply answers, while an AI agent makes decisions and acts autonomously — sometimes spawning other agents. He explains why prompts are only ~50–60% reliable (\"a pinky swear with a robot\"), how one rogue agent erased a company's entire production database, and why the deploying practice, not the developer, is who regulators hold accountable. My Compliance Citadel acts as a fortress around that risk: a runtime layer that intercepts agent actions and renders one of three verdicts — pass, pause, or prevent — logging everything to a tamper-proof audit chain. Erik stress-tests it against two real scenarios: pasting patient labs into open ChatGPT, and using AI to comb an EMR for high-risk patients.\nFor clinicians, students, and small-practice owners excited about AI but wary of HIPAA and PHI exposure, it's a grounded primer on innovating boldly without betting the license.","thumbnail_url":"https://img.transistorcdn.com/RDUSeE44PXw6C0vroYb_a-tROTMjDYFL7tE1bj-xPjc/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS9kZDU2/ZDYwYTUyOTJkZDE3/ZTYwZjk4NzgzNzBj/OWQ4Ni5wbmc.webp","thumbnail_width":300,"thumbnail_height":300}