AI Strategy February 2026 By Gabe Rivero

CX AI in 2025: What It Means for the Rest of Us

McKinsey's State of AI 2025 report puts a number on something most CX leaders already feel: nearly every organization claims to be using AI, but only a small fraction is actually extracting meaningful value from it. The gap between adoption and impact is not a technology problem. It is a discipline problem.

Here is what the data actually shows, and what it means for CX leaders trying to make AI work in the real world.

The gap is real — and it is structural

The report shows that roughly 88 percent of organizations claim AI integration, but close to two-thirds remain stalled in pilot projects that never convert to production. In CX specifically, where the operational surface area is large and change is disruptive, this pattern is especially pronounced.

The organizations that are pulling away are not necessarily the ones with the biggest budgets or the most sophisticated models. They are the ones that did three things before they deployed anything:

Fixed broken workflows before layering AI on top of them
Built governance structures that made AI outputs trustworthy
Defined clear problems instead of chasing capability

AI applied to a broken process produces faster broken outcomes. The sequence matters more than the technology.

Why CX teams stay stuck in pilot mode

Pilot mode is comfortable. It generates activity, creates the appearance of innovation, and avoids the hard organizational work of deciding who is accountable for outcomes when something goes wrong.

Most CX AI pilots stall for one of three reasons. The first is that the problem being solved was too vague to begin with — "improve customer experience" is not a problem definition, it is a direction. The second is that the data required to make AI reliable was not in shape before the tool was deployed. The third is that no one owned the result. Success metrics focused on tool adoption, not customer or operational outcomes.

What smaller organizations can do faster

Here is where the McKinsey data actually works in favor of mid-market and smaller CX teams. Large enterprises move slowly. They have more complexity, more stakeholders, and more legacy infrastructure to navigate. Smaller organizations can make decisions faster, run tighter experiments, and change course without a procurement cycle.

The advantages are real — but only if they are used deliberately:

Pick one problem that is specific enough to measure. Not "reduce handle time" — "reduce handle time on billing inquiries by 20 percent in 90 days."
Get your data in order first. If the knowledge base is outdated or inconsistent, AI will amplify the inconsistency.
Define accountability before you deploy. Who owns the result? Who can override the AI? What triggers a review?
Scale only what works. Do not expand a pilot that is generating activity but no measurable improvement.

The honest takeaway

The organizations winning with AI in CX are not doing anything exotic. They are applying discipline that was already best practice before AI existed: clear problem definition, clean data, explicit accountability, and measurement tied to outcomes rather than activity.

AI amplifies what is already there. If the foundation is solid, the amplification is positive. If it is not, you get faster noise.

You do not need a large budget to do this well. You need structure, transparency, and a plan that makes AI part of everyday work rather than a separate initiative sitting on top of it.