Healthcare doesn’t have an AI problem; it has a staffing orchestration problem

Tonight, a hospital will run a unit one nurse short. In the same city, a qualified nurse who wanted that exact shift will sit at home because their credentials have been under review for nine days. Both things happen at once, in a country short of more than 250,000 nurses, and they will happen again tomorrow.

The technology to clear that nurse in a day exists; most health systems have already funded some version of it, but the technology never reaches the units. Guidehouse’s 2026 Healthcare AI Trends Report, which HIMSS put back in front of its members this week, explains why. HIMSS surveyed 50 healthcare decision-makers; Guidehouse analyzed the answers and found that organizations approve AI investment and then stop before anything is successfully deployed. The report calls this “execution paralysis” and blames internal disagreement, undefined use cases, and the lack of a single executive owning the rollout.

I have sat across the table from those organizations for years. The report describes a technology problem. From where I sit, it is a staffing orchestration problem, and the use cases it documents are mostly financial: revenue cycle, claims denials, expense management. That is where health systems have been willing to put narrow AI to work, and by Guidehouse’s account, it is working. 

Using AI to manage your workforce solves the same kind of problem. Credentialing and scheduling are currently manual processes that, in my experience, are the projects that stall.

I do not think most executives underestimate the technology. They underestimate what the waiting costs.

A manager building a schedule in a spreadsheet spends roughly three hours on it, then rebuilds it every time someone calls out. A manager who cannot see float pool availability, per diem availability, and open shifts on one screen makes phone calls and guesses. Credentialing runs on the same kind of delay: about two weeks at most facilities, and for those two weeks, the clinician cannot work, and the shift stays open.

I have watched what those gaps turn into. An unfilled OR block becomes canceled cases, and a surgeon who books the next ones somewhere else. An unfilled ICU shift becomes overtime, an agency premium, or a heavier assignment for the nurses who showed up. A skilled nursing facility that cannot staff a wing turns away an admission and eats the empty bed for the month.

The HIMSS finding I keep returning to is that commitment to AI varies by role and seniority. In practice, this often means the CNO is eager to adopt new technology, while the scheduling team remains tied to legacy workflows. Rather than viewing this as pushback, I recognize that the existing system keeps the unit afloat week to week. Asking a team to learn new software during a severe staffing shortage demands time they simply do not have. Frequently, the individual relying on the spreadsheet is the very backbone of the unit’s operations.

Overcoming this divide requires a practical execution strategy:

• Single executive ownership: Assign a clear executive sponsor to drive accountability and oversee the outcome.
• Firm transition date: Establish a definitive deadline to decommission legacy processes.
• Protected training time: Provide schedulers with dedicated hours to master the platform. Organizations that skip dedicated training end up running parallel systems for a year before incorrectly declaring the technology a failure.

There is one more reason these projects stall, and the report gets at it under organizational readiness. Executives hear “AI for workforce management” and picture ripping out the scheduling system they spent three years implementing. That fear is rational; a health system running scheduling in UKG, Oracle, or Workday has it tied to payroll, timekeeping, and contract rules. Because replacing core software requires multi-year efforts, we designed Talent Fusion to operate without demanding a complete system overhaul.

For facilities that rely on spreadsheets and group chats, it replaces those manual processes, serving as a central hub for schedule creation, gap coverage, and a centralized credentialing system.

Where existing scheduling tools must remain, it operates alongside HCM systems like Workday, UKG, and Oracle rather than replacing them. It ingests published schedules, labor requirements, and workforce data to automatically route open shifts through tiered labor sources, starting with internal staff and lower-cost options before escalating to premium agency rates. Direct integrations with HRIS, payroll, and timekeeping ensure trusted operational workflows stay intact, while a robust governance layer and rules engine enforce rates, approvals, and compliance with real-time visibility across all units.

I describe Talent Fusion to CNOs as a command center for the workforce. Most existing scheduling platforms generate a static schedule, leaving managers to manually handle shift call-outs, census fluctuations, and last-minute vacancies over the phone. Talent Fusion automates dynamic scheduling by predicting coverage gaps before they occur to optimize staffing, matching open shifts exclusively with clinicians who meet facility credentials and qualifications. 

To healthcare leaders considering delaying adoption until an upcoming scheduling software upgrade is finished, my advice is clear: act now. Enterprise system rollouts take years, but schedule vacancies need immediate resolution. Workforce orchestration can be deployed today and continue adding value through future system migrations.

If I were running a health system today, I’d start by looking closely at three things: how long credentialing takes, how many hours teams waste building and tweaking schedules, and how many open shifts get filled by agency staff. Many health systems don’t even track all three, but once you look at those numbers, the need for change becomes pretty obvious.

My message to fellow health system leaders is simple: treat staffing as core infrastructure right now, not as an emergency you react to later. Organizations that make that shift today will see better shift coverage and lower labor costs by 2027. Those who stay stuck in execution paralysis will keep paying premium agency rates and counting the cost of waiting.