To Speed Up AI Success, Strategic Leaders Champion a Slower Pace

Almost as quickly as it began, the rush for rapid AI productivity gains is ending. Now, forward-looking leaders are replacing that early hope with a more deliberate strategy. Among the many who attempt, however, only a few will succeed. The most effective of them all will be those that build finely tuned, highly orchestrated systems.

For a practitioner's take, we spoke with Will Clevenger, Senior Director of Strategy at Avanade. An Agentic and Generative AI specialist, Clevenger has a long history of strategic work with Fortune 500 organizations. He's also a participant of the 2025 Perplexity AI Business Fellowship Program and the creator of WILLIS, an internal AI-powered digital twin. From his perspective, the industry's obsession with speed is fundamentally misguided.

"Are people willing to slow down to go past all their competitors? Everybody feels it needs to be a technology solution, when it's probably the people and process side first," Clevenger says. For him, the first mistake was the market's "over-rotation" on massive, general-purpose AI.

• A specialist's prescription: Real ROI usually emerges from smaller, finely-tuned models rooted in a business's unique processes, Clevenger explains. "The market rushed to the generalist LLM. It’s like having a throat issue and going to a general practitioner instead of an ENT. Domain is what’s key. A small, tuned model rooted in your specific processes will always be more valuable."

But the pivot to process almost always hits an immediate wall: the human element. Without hands-on experience, many leaders default to simply "bolting on" new technology to old workflows, Clevenger explains.

• Popping the kernel: Often, the disconnect stems from a misunderstanding of the technology's purpose, he continues. "The moment I figured out how to build an agent myself was when my head popped open and I could rethink everything. Leaders who haven't had that personal 'kernel pop' moment are the ones just bolting AI onto existing processes because they haven't had the hands-on experience."

• Magician's trick: Meanwhile, viewing AI as a simple assistant overlooks its potential as a tool for deep process re-engineering, Clevenger says. "If I were up against a skeptical leader, I would approach it like a magician dealing with a skeptic. You don't argue, you demonstrate. I'd ask for their time for just one session to show them what's possible, because you often have to break through a fixed mindset."

As a result, this cultural change often has a direct impact on how companies select projects and consider ROI. For many enterprises, the new modus operandi is to chase massive, single-threaded projects with the biggest potential return. However, with the underlying technology in constant flux, Clevenger sees this strategy as flawed. "Enterprises want to treat AI like a stamping machine: you plug it in, and it runs the same way every time. But with this technology, by the time you sit back down at your desk, something has already changed."

• ROI reframe: "Stop chasing the single biggest ROI. The smarter move is to sequence smaller projects with shorter paybacks that create reusable components. Do three lower-ROI projects this quarter, and you might accomplish that big project next year twice as fast because half of it is already solved," Clevenger says.

• Ready, set, second: Then the task becomes orchestrating these efforts across business units to achieve larger strategic goals. "You have to balance the platform's innovation with your own to stay in front. In this market, you can still be a second mover—but you must be an 'out-of-the-blocks' second mover, ready to execute instantly."

Executing this deliberate strategy without rigidity requires an operating system that can adapt to constant change. For Clevenger, that means reviving proven business disciplines for a new technological reality.

• An Agile answer: Here, he points to Agile as a good example of a proper tactical framework. "Agile is re-emerging as a critical practice, not for DevOps, but because it provides the framework to manage this constant level of change. You need a built-in mechanism to re-evaluate strategy frequently, because a new model, connector, or protocol could drop at any time and completely change your path."

• Old skill, reapplied: In turn, that revival will also require the cognitive skill of systems thinking. "There's a reason the World Economic Forum puts systems thinking on its list of core skills for 2030. It's an old-school business process management skill we buried years ago, and now, to succeed in AI, everybody needs it."

As the market moves toward what Clevenger predicts will be a world of multi-agent systems by 2026, "The question is: who wins? Hyperscalers, specialists, or workflow players? It is not about the technology. It's about the leaders who are willing to slow down, learn, and rethink how people and processes fit together. And that's the real accelerant."

Ultimately, the winners will be the companies with a methodical approach, Clevenger concludes. For him, that includes those who are already outmaneuvering the organizations that adopted AI the fastest. But the change in mindset also warrants a reframing of the entire challenge as a holistic workforce strategy. "If I were a CHRO, I would be all over the idea of digital labor," he says. "I would be thinking about how the organization balances digital labor with human labor. Because that's what we're talking about. If it's a physical manifestation, it's a cobot. And if it's not, it's an agent or a genie or a copilot."