AI Readiness Gap: Most IT Leaders See the Need, Few Have Built for It

Most IT leaders now agree that AI belongs in the workplace. Far fewer have changed how their organizations actually work to make it pay off. That gap is the core finding of the sixth volume of Lenovo's Work Reborn report, and it's a more useful signal than the usual enthusiasm numbers.

The headline figures are stark. Among 1,200 IT leaders surveyed across the UK, US, Canada, Australia and eight other countries, 80% say they understand the potential AI holds for their workplace. Only 10% have led an AI transformation in their organization. Lenovo calls the space between those two numbers an "AI readiness divide," where ambition runs well ahead of preparation.

Adoption is wide, redesign is rare

Here's the thing: AI is already spreading across business functions, but the survey suggests most companies are layering it onto existing processes rather than rethinking them. Just 10% of organizations have approached implementation by redesigning how work gets produced. That's the difference between giving a team a faster tool and changing the workflow the tool sits inside.

Governance looks similarly thin. Only 10% of businesses report governance that covers AI tools, AI agents and data access together. As agents take on multi-step tasks and touch more systems, those three areas are where access control and accountability get tested first. A company that hasn't defined who or what can read which data has little basis for trusting the output.

Measurement is the weakest link of all. Just 10% of respondents say they have an ROI measurement in place to track AI's impact on results. The rest appear to be running what Lenovo describes as a "fire and forget" approach: deploy the tool, then hope the benefits show up somewhere. Without a baseline, nobody can say whether the spend worked, and that makes the next budget conversation harder than it needs to be.

What the readiness index measures

Lenovo scores organizations on its AI Workplace Readiness Index, which looks at five dimensions: architecture, operations, trust, workforce and value. A company has to perform across all five to be considered ahead, and that's a sensible design choice. A strong model deployment running on brittle infrastructure, with no trained staff and no way to measure results, scores well on one dimension and poorly on the rest.

The spread between groups is notable. Organizations with AI fully in use across their core business functions average 4.1 out of 5. Organizations still evaluating AI average 2.7. That 1.4-point gap suggests readiness is not a prerequisite that companies clear once and forget. It's built through the work of running AI in production, which is also why the laggards are at risk of falling further behind. Teams that have already deployed are accumulating the governance, measurement and workflow knowledge that the evaluators lack.

Measuring value at the workflow level

Rakshit Ghura, Vice President and General Manager at Lenovo Digital Workplace Solutions, put the question plainly: "Every company we're talking to is asking the same thing: what value are we actually getting from AI? That's where workflow-level measurement matters. Are cycle times down? Are repeat incidents falling? Is the outcome better?" He frames that kind of measurement as the way to "show where AI is creating value." I think that framing is the right one. Aggregate productivity claims are hard to verify and easy to dispute. Specific operational metrics, such as how long a request takes to close or how often the same incident returns, are things a team can baseline before a rollout and compare afterward. They also force a conversation about which workflows AI is supposed to change, which is often the step companies skip.

Ghura also argues that the workplace itself is part of the problem. He describes "a workplace designed for the pre-AI era" as something that stops AI from being used effectively. His proposed fix is to "redesign how people and AI work together, with the technology, support and governance around them built for how work is changing." That's a broader ask than buying licenses. It touches tooling, training, support models and oversight at once.

What to watch next

The survey is a vendor-produced snapshot, and its readiness index is Lenovo's own construct, so the 4.1 and 2.7 scores are best read as directional rather than as industry benchmarks. Still, the underlying pattern is one many teams will recognize: pilots spread quickly, while redesigned workflows, governance and measurement lag behind.

For technology leaders, the practical test is simple. Before expanding an AI deployment, can you name the workflow it changes, the metric that will show whether it worked, and the rules that govern which agents can access which data? If the answers are vague, the bottleneck probably isn't the model. It's the organization around it, and that's where the divide will be decided over the next few budget cycles.