Fifty-seven percent of enterprises now have AI embedded in core business processes. That is up from 35% a year ago. By any deployment metric, the rollout is working.

By any outcome metric, it is not.

Kyndryl’s second annual People Readiness Report, published July 17, surveyed 1,100 senior business and technology leaders across eight countries. The headline finding: only 32% of organizations with broad AI deployment have achieved even one of their top two AI objectives. Eleven percent have hit both. The rest are running AI that does not produce measurable results for the business.

And the workforce side is moving in the wrong direction. Only 23% of business leaders believe their people are fully prepared for AI. That is down six points from 2025. Not flat. Down. The more organizations deploy, the less confident their leaders are that the humans in the system can keep up.

The confidence collapse at the team level

The Achievers Workforce Institute published parallel data in its 2026 State of Recognition Report. Just 19% of workers feel confident using AI tools. Only 18% feel supported in adapting to them. That means more than 80% of the typical enterprise workforce has neither the confidence nor the clarity to integrate AI into daily work.

This is not a training problem in the way most leaders frame it. Saying “we need more workshops” misdiagnoses what broke. The role was never redesigned. The workflow was never rebuilt. The tool arrived and the person was told to figure it out.

Meanwhile, worldwide AI spending is forecast to hit $2.52 trillion in 2026, a 44% year-over-year increase according to Gartner. Capital flows into infrastructure. Human enablement lags behind it. The spending is accelerating. The readiness is not.

What the 9% are doing differently

Kyndryl’s report identifies a group it calls Pacesetters, roughly 9% of respondents, that are actually hitting their AI targets. The performance gap is not subtle. Pacesetters are 1.5 times more likely to achieve AI-related revenue growth and 1.6 times more likely to report improved innovation.

What separates them has nothing to do with budget or which model they picked. It is four practices that the other 91% skip or half-finish:

They redesign roles around AI before deploying it. Not after. Not “eventually.” Before the tool hits the team, the role changes to accommodate it.

They run structured change management alongside technical deployment — a dedicated workstream with owners, timelines, and accountability.

They establish governance guardrails early. Decision-rights policies that define what AI can and cannot do. Monitoring registries that make the system auditable. Pacesetters are roughly twice as likely to have fully implemented AI governance compared to peers.

They invest in workforce readiness as a program, not a line item. Only one-third of organizations in the study have fully implemented training focused on working alongside AI. Pacesetters treat that training as load-bearing infrastructure.

The trust deficit is an operational risk

Eighty-one percent of organizations expect AI agents to make impactful business decisions within the next year. Only 25% currently have complete trust in autonomous AI systems. That gap between planned deployment and actual trust is where failures live.

Kyndryl’s data shows a direct link between governance maturity and workforce trust. Organizations with stronger governance frameworks report higher employee confidence and significantly better outcomes. But only 27% are using registries and monitoring across all AI systems. Most enterprises are deploying systems they cannot fully audit to teams that do not feel equipped to use them.

What this looks like in six months

If your organization deployed AI broadly in the last year but cannot point to achieved objectives, the diagnosis is specific. The tools work. The models are capable. The roles never changed. The team was never brought along. And every month you wait, the distance between what’s running and what people can actually use gets harder to close.

The Pacesetters did not wait for confidence to appear. They built it by redesigning the work first. The other 91% bought the tools and hoped the people would catch up. That bet is losing. And the longer it runs, the harder the gap is to close.

Six months from now the Pacesetters will have compounded their lead. Their teams will have six more months of daily reps — workarounds discovered, edge cases internalized, judgment built through use. You cannot replicate that with a Q4 training sprint. The gap between organizations that redesigned and organizations that just deployed is already structural. It widens every week someone’s role description says nothing about the AI sitting in their workflow.