Summary: The University of Miami's Miller School of Medicine ran a 100-day agentic AI challenge focused on pathology, pairing physicians with technologists to build tools that automate parts of the diagnostic workflow.
Rather than a typical hospital IT rollout, the program treated pathologists as co-designers, testing where AI agents can reliably assist and where human oversight remains essential.
The effort reflects a broader shift in healthcare AI: away from black-box automation and toward supervised, task-specific systems. For Miami, it signals growing institutional appetite to build AI expertise locally rather than simply buying it off the shelf.
University of Miami Has Put Physicians at the Center of AI Development
Artificial intelligence is becoming part of modern healthcare, but many hospitals are still deciding how to introduce it safely into clinical practice.
The University of Miami's Miller School of Medicine recently explored one approach through a 100-day innovation challenge that paired pathologists with engineers to build AI tools for pathology workflows.
Rather than attempting to automate medical decision-making, the project focused on using AI to support clinicians by reducing repetitive work and improving efficiency while preserving physician oversight.
Building AI Around Clinical Workflows
The initiative centred on agentic AI—systems capable of carrying out multi-step tasks under human supervision.
Within pathology, those tasks can include reviewing digital slides, identifying areas that require closer attention, organising case priorities and assisting with report preparation before a pathologist performs the final review.
By involving physicians throughout the design process, the project aimed to build tools that fit naturally into existing clinical workflows rather than asking clinicians to adapt to unfamiliar technology.
That physician-led approach reflects a growing recognition that successful healthcare AI depends as much on clinical trust as technical performance.
Rapid Development Instead of Long IT Projects
The 100-day format encouraged teams to develop working prototypes quickly rather than spending years planning large technology rollouts.
Healthcare organisations have traditionally adopted new technology through lengthy procurement and implementation cycles. Rapid development programmes allow clinicians to test ideas earlier, identify practical challenges and refine solutions before wider deployment.
The focus remained on narrow, well-defined tasks where AI can improve efficiency rather than attempting to replace medical expertise.
This mirrors the direction many healthcare organisations are taking, using AI to augment clinicians rather than automate complex diagnostic decisions.
A Growing Trend in Academic Medicine
Medical centres across the United States are increasingly embedding AI development inside hospitals instead of relying entirely on commercial software vendors.
Working directly with physicians allows institutions to create tools that reflect their own workflows, patient populations and operational needs.
Pathology has emerged as a particularly promising area because digital pathology produces large volumes of image data while still requiring expert clinical interpretation.
That combination makes it well suited to AI-assisted workflows while maintaining the high level of human oversight required in patient care.
What This Means for Miami
The University of Miami's initiative highlights how South Florida's healthcare sector is becoming an active participant in AI development rather than simply adopting technology created elsewhere.
For Miami's growing healthtech ecosystem, collaboration between clinicians, engineers and researchers creates opportunities to develop and validate AI tools within real hospital environments.
It also helps strengthen the region's talent pipeline by exposing physicians and technical professionals to practical AI implementation in healthcare.
As Miami continues to expand its AI ecosystem, projects like this demonstrate that some of the city's most significant innovation may come from highly regulated industries where improving workflows—not replacing professionals—is the primary goal.