Healthcare conference season has a way of surfacing similar optimistic narratives year after year. However, last month’s conversations at Becker’s felt different. More candid, more urgent. The leaders in the room weren’t talking about possibilities. They were talking about what happens if you wait.
For academic medical centers especially, the pressure is compounding. Margins are thin. Workforce gaps are widening. AI is moving from pilot project to core infrastructure. Most importantly, patients have stopped tolerating systems that weren’t built with them in mind. Here’s what stood out as the five priorities no AMC can afford to treat as aspirational right now.
Stop Treating AI Like a Project
Across every session, the message was the same: health systems still running AI as a portfolio of disconnected pilots are going to keep starting over. Vendors change. Projects end. Institutional knowledge doesn’t accumulate. The next wave of capability arrives before the last one was absorbed.
The shift that’s working is treating AI as capital infrastructure. Budget it that way, govern it that way, and build internal capability instead of outsourcing your entire data strategy to vendors who may not be around in three years. The organizations doing this now are the ones who will actually be able to leverage what comes next.
“AI’s near-term value isn’t replacing clinical judgment. It’s removing the low-value work that gets in the way of it.”
Redefine What 'Access' Actually Means
Margins are averaging around 1.5% nationally. There’s a projected shortage of 90,000 physicians by 2036. Anyone still trying to solve access by adding headcount or square footage is fighting the wrong battle.
The AMCs gaining ground here have reframed the problem entirely: access isn’t about capacity, it’s about routing, coordination, and removing friction. That means self-scheduling that actually works, digital intake that doesn’t require a phone call, AI triage that gets patients to the right level of care before a human scheduler ever gets involved. One system showed 25% of specialty referrals were being misrouted, and AI corrected that to near zero within two weeks. Misrouting at that scale is an access problem, not a technology one.
Integration Beats Collaboration Every Time
Academic medicine has historically operated as three parallel institutions: clinical care, education, and research, each with its own culture, leadership, and metrics. That structure made sense once. It doesn’t anymore.
What consistently separates the leaders in this space is not smarter strategy. It’s structural integration. Clinical, IT, data, operations, and finance working from shared workflows and shared accountability, not occasional cross-functional meetings. One health system described moving to multi-monthly integrated leadership meetings until it became cultural. The result was faster implementation, better adoption, and change that didn’t evaporate when the project lead moved on.
Your People Strategy Is Your AI Strategy
One of the more honest moments from the conference came when a leader shared their experience. Their system deployed ambient AI, saw physicians free up significant time, and immediately used that capacity to push more patient volume. Physicians left. Not because the technology failed, but because the organization broke trust the moment it delivered.
The leaders getting this right understand that sustainable transformation has a sequencing problem. You build cultural trust first. You give teams space to get comfortable with new tools. The performance gains come, but they have to be allowed to come naturally, not mandated before the ground is stable. Employee experience is the foundation of patient experience, not a parallel track.
Raise the Floor Before You Chase the Ceiling
The most repeated warning across multiple sessions: isolated excellence is not transformation. A department with great outcomes, or a campus where AI is working well, doesn’t mean your enterprise is on solid ground.
The playbook from systems that have scaled was as follows: standardize first, prove it works, then expand. Fix the referral pathways. Standardize the intake process. Build governance that lives in the system, not in the person who championed it. Because when that person leaves, and they often do, the work needs to survive them.
When something fails, ask the harder question before walking away: was it the concept, or the execution? Most unsuccessful care model innovations fail because the right stakeholders weren’t in the room, or the workflow didn’t match operational reality. That’s a recoverable problem, but only if you diagnose it honestly.
The institutions that define the next decade of academic medicine won’t be the ones with the best strategy. They’ll be the ones who hired the leaders capable of executing it, two or three years before they needed them.
Informed by key insights from the 2026 Becker’s Healthcare Conference.
English 
Español de México