Inside the 100-Day Agentic AI Challenge Transforming Pathology at the Miller School
Faculty, trainees and staff at the University of Miami Miller School of Medicine are building real-world AI agents that automate clinical and administrative workflows, marking a shift from conversational AI to practical health care innovation.
Building an AI agent may seem the province of hard-core software developers. But a group of innovative medical professionals at the University of Miami Miller School of Medicine is building and deploying AI agents into their everyday workflows.
At a recent Innovation Mixer and Agentic AI Summit hosted by the Department of Pathology and Laboratory Medicine, faculty, staff and trainees gathered to share the practical results of the group’s 100-Day Agentic AI Challenge.
This structured pilot program was designed to demystify artificial intelligence, giving physicians, researchers and administrative staff the tools to build their own workflow digital assistants. The initiative was spearheaded by Yanyun Wu, M.D., Ph.D., M.B.A., clinical professor of pathology and laboratory medicine and vice chair of business development and innovation for the department, in collaboration with University of Miami AI experts David Wayne, Ph.D., UM’s chief artificial intelligence officer, Inti Bryon, manager of digital product management at UM, and a Microsoft team led by Cloud Solution Architect Arielle Theodore.
From AI Curiosity to Clinical Application
The summit highlighted the evolution in how health care professionals use digital tools. Dr. Wu explained that, while many are now comfortable with conversational AI, the department wanted to explore efficient task automation through the building of AI agents.
“The goal of the challenge was simply to build a practical bridge,” Dr. Wu said, “taking the mystery out of the technology and allowing people to safely automate repetitive, time-consuming tasks.”
Agentic AI is a fundamental breakthrough, given its ability to apply cognitive flexibility to task automation. Agentic systems perceive an overarching objective, evaluate the digital tools at their disposal, cross-reference messy or unstructured data and self-correct when they encounter errors. Within pathology’s dense influx of multi-page diagnostic reports, complex regulatory criteria and meticulous compliance checklists, the shift significantly reduces the mental friction and transaction costs of medicine.
By turning reactive documentation and administrative tracking into a proactive digital assembly line, of sorts, agentic systems allow care teams to shift focus away from repetitive tasks and back to critical diagnostic decisions and direct patient support.
Inside the 100-Day AI Challenge
The challenge divided its 100-day timeline into a manageable, step-by-step progression: Learn, Apply and Showcase. Spanning multiple teaching sessions, collaborative group discussions and open coaching, the program assumed participants had no software development or coding experience.
As the cohort advanced to the “Apply” phase, they began working with Microsoft Copilot Studio and Power Automate. These low-code platforms securely plug into common workplace systems across the Microsoft suite. All systems function within the University of Miami’s private enterprise tenant. Every tool worked inside a secure environment.
At the summit, participants showcased the functional, custom-built AI agents they developed. Their goal was to ease administrative burden, streamline operations and work in a new and better way. Presentations included:
• The ACGME FAQ Chatbot: GME coordinator Faith Siem presented an agent designed to help navigate complex ACGME guidelines, providing instant, accurate answers to training, compliance and administrative questions.
• The Pathology Specimen Request Guide: Sophie Egea, Ph.D., the department’s director of clinical research operations, demonstrated a digital assistant that guides researchers and clinicians through the requirements, protocols and paperwork needed to request and utilize pathology specimens for scientific study.
• The CPT Coding Assistant: This AI Agent, developed by department resident Felipe Ruiz Casas, M.D., helps automate the task of matching complex clinical narratives to CPT billing codes. By cross-referencing institutional guidelines with report text, it reduces a task that traditionally requires 20 minutes a day down to a matter of seconds.
• The Radical Prostatectomy Report Validator: Presented by Diego Montoya Cerrillo, M.D., an assistant professor of anatomic pathology, this agent serves as an automated quality check for surgical pathology reports. It reads complex, post-operative reports and instantly cross-checks documentation against strict staging criteria, replacing a slow, manual audit with an immediate, reliable secondary review.
• The Breast Pathology Patient Assistant: This agent focuses on compassionate patient communication. Presented by Carmen Gomez-Fernandez, M.D., clinical professor of pathology and laboratory medicine and the department’s vice chair of education and trainee mentoring and director of undergraduate pathology medical education, it translates technical pathology jargon into accessible explanations at a supportive sixth-grade reading level, helping to alleviate patient anxiety.
• The Path Extract Agent: Catalina Amador-Ortiz, M.D., associate professor of pathology and laboratory medicine, showcased this workflow tool that automates structured data extraction from thousands of pages of natural language search reports, greatly reducing time spent on data entry and accelerating research capabilities.
• The Platelet Inventory 24-Hour Expiry Tracker: Designed to actively monitor the brief shelf-lives of critical blood products, this agent, presented by fellow Menatalla Nadim, M.D., flags units nearing expiration to optimize utilization, protecting lifesaving resources and reducing waste.
• The Staff Schedule Assistant: Dr. Wu built this agent to handle complex shift coverage by cross-referencing service needs, role assignments and staff availability, automating shift rosters.
• Scan SOPs-CAP Gap Analysis: Yi Zhou, M.D, Ph.D., clinical associate professor of pathology and laboratory medicine, created a regulatory compliance tool that automatically cross-references internal standard operating procedures against evolving College of American Pathologists accreditation checklists, simplifying laboratory audits.
• Eye Q: This analytics copilot, designed by Giselle Ricur, M.D., M.B.A., executive director of virtual care at Bascom Palmer Eye Institute, analyzes post-encounter patient comments across three domains and automatically generates professional, leadership-ready reports detailing trends, critical issues and barriers to virtual care.
• Opportunity and Research Intelligence Optimization Network (ORION): Rafael Frankenberg, director of research support for the department, presented this institutional intelligence platform designed to continuously map faculty expertise and identify, evaluate and activate high-value private sector partnerships. ORION aligns research capabilities with industry demand for a more diversified funding portfolio, stronger industry collaboration and a more proactive, strategically positioned research enterprise.
A Cultural Shift Toward Practical AI in Medicine
To understand how these new tools might spread throughout the broader institution, Dr. Wu referenced Rogers’ Diffusion of Innovations theory, which explains how new concepts gradually gain acceptance within an organization.
“The 100-Day Challenge was an effort to engage our innovators and early adopters,” Dr. Wu said. “Their willingness to try something new is essential because they validate the tools and help their colleagues feel comfortable. The practical success stories we saw today provide the real-world proof points needed to gently encourage the wider university community across the chasm into active AI engagement.”
“Innovators spark profound change before the world even realizes it’s possible,”said Merce Jorda, M.D., professor and J.R. Coulter Chair of pathology and laboratory medicine at the Miller School. “By cultivating a safe sanctuary to boldly experiment and fail forward, we have ignited a practical movement. We must continue journeying upward together, fueled by this shared, unstoppable passion.”
More from the Miller School of Medicine
New research from the University of Miami shows AI-driven operative reporting reduced documentation time by 70%.
AI analysis of bone marrow biopsies uncovers immune signals that may guide personalized multiple myeloma treatment decisions.
Dr. Thomas Albini performed the first surgery in Florida to place a microscopic implant designed to treat MacTel.
At the 2026 Innovations in Medical Education Conference, educators examined how AI is transforming medical training.
Tags: AI, artificial intelligence, Department of Pathology and Laboratory Medicine, Dr. Carmen Gomez-Fernandez, Dr. Catalina Amador, Dr. Diego Montoya, Dr. Giselle Ricur, Dr. Merce Jorda, Dr. YanYan Wu, Dr. Yi Zhou, pathology, technology