AI is getting smarter at remarkable speed. Employees can draft, analyze, search and automate routine work faster than they could a year ago. Recent research on the state of AI shows organizations continuing to expand adoption across functions and use cases.
Yet in many organizations, the business still works the same way. The same work moves through the same systems, waits in the same queues and stops at the same approvals. AI speeds up a task, and the benefit disappears into a process designed for another era.
I see this from two vantage points. CDW works across thousands of customer environments, so we have a broad view of where AI creates value and where companies get stuck. I also help lead CDW’s own AI strategy and transformation. We have made AI assistants available to more than 14,000 coworkers, built more than 12,000 agents and started redesigning core workflows.
That experience has convinced me the CIO’s challenge has changed. It is no longer simply enabling access to AI. It is helping the business move quickly without creating fragmented technology, uncontrolled cost and risks that IT will ultimately have to manage.
Where the productivity gains get lost
Here is the pattern I keep seeing. A company gives employees AI. People use it to summarize documents, prepare analysis and draft recommendations. Employees report saving time. But business performance barely moves.
The reason is simple. Every task sits inside a larger workflow. An employee may finish one step faster, but the output still lands in the same queue and waits for the same sign-off. Another team may still re-enter it in a different system. The task got faster. The constraint just moved.
We saw the same thing inside CDW. Usage climbed quickly, but usage alone couldn’t tell us whether the business was performing better.
The task is not the real unit of business performance. The workflow is. In many organizations, the next phase of AI value creation will come from improving how work moves across the enterprise, not simply automating individual tasks.
Don’t automate the workflow you already have
Once leaders see the workflow as the constraint, they’re tempted to apply AI to every step. That instinct is often a mistake.
Many steps in a process were built around the limits of older technology or around organizational boundaries that no longer make sense. Others survive because no one has questioned them. Put AI into each of those steps and you get a faster version of a process that shouldn’t exist in its current form.
The better approach is to start with the business challenge, not the technology. Define the outcome that matters and work backward. Which steps still add value? Which can disappear? Then decide who, or what, should do the remaining work: conventional automation, an agent or a person who brings judgment and accountability.
Our experience building agents inside CDW has changed how I think about them. I used to see agents as more capable assistants. Now I am more interested in how they can move work forward. That is what creates the opportunity to remove steps, not just speed them up.
That experience has also made us pragmatic. Not every problem needs an agent. Sometimes a simple rule works better and costs less.
Don’t automate the workflow you have. Design a better one.
Employees create readiness. Business leaders create authority.
A redesigned workflow still has to run through people, and each group has a different job.
Employees experiment and surface opportunities. Hands-on use is how people learn what AI can and can’t do and where it can help. But employees can only change the tasks they control, not the handoffs before and after them.
Business leaders choose which workflows matter and own the outcomes. They have the authority to bring functions together. But a workflow designed from the top will struggle if the people running it have never used AI or don’t trust it.
Inside CDW, we need both: broad access so coworkers build fluency, and leadership focus on a few core workflows where redesign can change performance.
Neither scales, though, if every team has to solve for tools, data, security and cost on its own.
The CIO mandate: Build the enterprise AI foundation before workarounds take hold
The CIO doesn’t need to own every use case or redesign every workflow. The mandate is to build the enterprise AI foundation quickly enough, and make it useful enough, that business teams build on it rather than around it. It’s also increasingly reflected in the questions shaping CIO agendas as leaders balance AI opportunity with governance, operating models and business value.
That foundation combines approved AI tools, trusted data and business context, connections into enterprise applications and workflows, security and governance, and visibility into performance, cost and value. Many of these principles align with established guidance for managing AI risk at scale. It gives employees a safe way to experiment and business teams a repeatable way to scale what works.
Timing matters. Business teams won’t wait for IT. If the foundation isn’t there, they’ll find their own tools and build their own connections. AI adoption will continue, just out of sight. The CIO then inherits the result: fragmented data, duplicated spending, inconsistent governance, cyber risk and little visibility into what AI is doing across the company.
Usefulness matters just as much. A foundation that is too restrictive or hard to use drives people around it. One that works becomes the default, and each new agent or redesigned workflow can reuse the data, connections and controls already in place. That is how we are approaching it inside CDW.
Build the highway for AI transformation, or spend the next several years managing the risks created on the side roads.
Measure what changed
AI will keep getting more capable. The competitive difference will come from what companies are willing to change around it. The measure of that change isn’t how much AI a company has deployed. It’s whether the business performs better.
That is the standard we are applying inside CDW. It is also the standard I would encourage every CIO to use. Do not stop at licenses, usage or hours saved. Ask whether AI has changed an outcome the business cares about.
Does the company now work better?