Healthcare has been running AI pilots for three years. Most of them worked. Almost none of them scaled.
That’s not a coincidence, and it’s not a technology problem. A pilot is designed to succeed under favorable conditions: dedicated staff, narrow scope, a vendor highly motivated to deliver, and leadership that approved the project and needs it to justify the decision. None of those conditions survive contact with the enterprise. The pilot didn’t reveal whether the technology could scale. It revealed whether the technology could perform in a demonstration.
The harder question — whether the organization itself is willing to change to support it — never gets formally asked. By the time the pilot ends, the window for asking it has usually closed. The results go into a deck. The deck gets presented once. The next pilot begins.
Chris Boyer and Reed Smith examine what perpetual piloting actually costs health systems, and why the gap between “we piloted this” and “we run this” is where most healthcare transformation quietly ends:
- Why the conditions that make a pilot succeed are precisely what makes it an unreliable predictor of enterprise viability
- How piloting became a way to signal innovation without requiring organizational change
- Why the decision to pilot and the decision to scale are different decisions — and why most organizations only make the first one
- What the one healthcare AI use case that has actually graduated pilots has in common with successful enterprise deployments in other industries
- What it looks like to design a pilot you actually intend to graduate, starting before the pilot runs
A new Kyndryl study found 76% of healthcare organizations report having more AI pilot programs than they can scale. Not more than they’ve scaled. More than they can scale. That number is not a technology indictment. It’s an organizational one.
The question for every health system running pilots right now isn’t whether the technology worked. It did. The question is whether the organization is prepared to decide.
Mentions from the Show:
- TP427: The Case for Failing Faster to Address Disruption: touchpoint.health
- Kyndryl Research, March 2026: https://www.healthcareittoday.com/2026/03/08/bonus-features-march-8-2026-69-of-physicians-struggle-to-access-recent-records-from-outside-providers-76-of-healthcare-orgs-have-more-ai-pilot-programs-than-they-can-scale-plus/
- Define Ventures C-Suite AI Survey via Healthcare Finance News, 2024: https://www.healthcarefinancenews.com/news/payers-providers-increasing-investments-ai
- McKinsey, Reimagining Healthcare Service Operations in the Age of AI, 2024: https://www.mckinsey.com/industries/healthcare/our-insights/reimagining-healthcare-industry-service-operations-in-the-age-of-ai
- Bain & Company / KLAS Research, Healthcare IT Investment: AI Moves from Pilot to Production, October 2025: https://www.bain.com/insights/healthcare-it-investment-ai-moves-from-pilot-to-production/
- Reed Smith on LinkedIn: https://www.linkedin.com/in/reedtsmith/
- Chris Boyer on LinkedIn: https://www.linkedin.com/in/chrisboyer/
- Chris Boyer website: http://www.christopherboyer.com/
- Chris Boyer on BlueSky: https://bsky.app/profile/chrisboyer.bsky.social
- Reed Smith on BlueSky: https://bsky.app/profile/reedsmith.bsky.social
