Our Take
Health systems are buying AI solutions without buying the organizational change required to use them; the problem is not software, it is leadership structure and workflow redesign.
Why it matters
Healthcare leaders under margin pressure are tempted to treat AI as a plug-and-play efficiency fix. This analysis suggests that assumption is backwards and will waste budget.
Do this week
CIO or Chief Medical Officer: audit your current AI deployments this week by asking which ones changed a clinical workflow or reduced a measurable cost, versus which ones shipped and sat idle.
AI adoption without AI outcomes
Health systems have begun deploying artificial intelligence tools across clinical and operational functions, but McKinsey's assessment finds a persistent gap: deployment is not translating into measurable value. The underlying diagnosis is specific: organizations are acquiring AI capabilities without rethinking the workflows, staffing models, and decision authority required to make those capabilities matter.
The McKinsey analysis frames this as a CEO imperative, not a technology imperative. It argues that capturing real value from AI requires "CEO-led transformation and a fundamental rethink of how care and operations run." The implication is that CIOs or vendor partnerships alone cannot close the gap.
Workflow precedes software
Health systems operate under tight labor and capital constraints. The appeal of AI is straightforward: reduce administrative burden, free clinician time, lower cost-per-procedure. But implementing that appeal requires answering upstream questions: Which workflows does this AI tool actually change? Who owns the changed workflow? What training or certification does that require? What happens to the staff member previously doing that work?
A radiology AI model, for example, only reduces cost-per-read if it changes how radiologists are staffed, scheduled, and compensated. Installing the software without changing compensation structure means the tool absorbs cost without freeing capacity. The McKinsey framing suggests many health systems have installed the software.
The second-order risk is budget fatigue. If the first wave of AI deployments ships without measurable return, board pressure to fund the next wave drops, even if the second opportunity is stronger. McKinsey is flagging a category of waste before it hardens into skepticism.
Measure before you buy
If you are a health system evaluating AI tools, start with the outcome, not the tool. What operational metric or clinical outcome are you trying to move? What workflow change would move it? Who owns that workflow today, and what would it take to change their incentive? Build the vendor evaluation around those questions, not around model capability or feature richness.
If you have already deployed AI tools in clinical or administrative functions and do not have a clear owner or a measured outcome, audit those deployments now. The McKinsey position implies that the absence of CEO sponsorship and workflow redesign is predictive of failure. Reallocate that budget toward projects where leadership has already committed to the organizational work, not just the software.