Artificial intelligence has entered the enterprise with enormous promise—and even greater confusion. Leaders know it will reshape their businesses, but they often don’t know how to sequence the journey or where ownership should sit. As a result, two unproductive extremes have emerged. I refer to them as the barbell problem: an organizational barbell with heavy weight on both ends and nothing of substance connecting them in the middle.
On one end of the barbell, companies lean into opportunistic experiments. They empower citizen developers, encourage personalized productivity tools, and let frontline teams adopt generative or agentic AI however and wherever they see fit. This bottom-up creativity is well intentioned, but it inevitably leads to fragmentation. Every team builds its own agents, its own automations, and its own micro-solutions. The result is a patchwork of disconnected tools that neither scales nor transforms the enterprise.
On the other end of the barbell sits an equally unproductive belief: that AI is primarily an IT initiative. Companies hand the problem to the technology organization and ask them to “come up with ways to automate things” or improve efficiency through advanced tooling. But this view reduces artificial intelligence to the next chapter of automation—another systems modernization project, another opportunity to squeeze out incremental savings, and another burden placed on already overstretched CIO teams.
Both ends of the barbell fail for a simple reason: AI is no longer a technology deployment. It is an operating-model redesign.
And operating-model redesign cannot be delegated to the extremes.
The Illusion of Bottom-Up AI
The enthusiasm around democratized AI is understandable. Tools like generative AI and agentic systems lower the barrier to entry and allow anyone to build workflows, assistants, or decisioning agents. But when adoption begins with random experimentation, organizations fall into a familiar trap: local optimization with no enterprise coherence.
Frontline tools rarely align with corporate data standards, governance structures, process maps, or customer expectations. And because they are not connected to a broader architectural vision, they remain isolated pockets of automation. Useful, yes—but strategically irrelevant.
Companies cannot innovate their way into transformation from the bottom up. The barbell’s “citizen developer” end may create noise, but it does not create compounding enterprise value.
The Illusion of IT-Driven AI
The opposite extreme—the belief that IT should own the AI agenda—is equally flawed. When AI is treated as an IT responsibility, it is immediately framed as a tooling problem. IT will look for platforms, evaluate vendors, add capabilities, and attempt to bolt new technologies onto legacy processes. That approach inevitably leads to narrow automation use cases, long implementation cycles, and very little change in the actual structure of work.
AI is not an IT project. It is a business reinvention project that requires IT. When ownership begins in the technology organization rather than the business, companies almost always under-reach, under-invest, and misunderstand the scale of change ahead.
The barbell’s “IT responsibility” end creates stability but not ambition. It ensures control but not transformation.
The Center Weight: CEO and Board-Level Strategy-Driven AI
The real weight must sit in the middle of the barbell: CEO, board-level and senior-leadership ownership of the AI mandate.
Artificial intelligence—especially agentic AI capable of end-to-end orchestration—has the potential to change how companies operate, sell, serve, build, and scale. It can fundamentally reconfigure cost structures, reshape customer experiences, rewrite operating models, and create entirely new categories of offerings.
You cannot achieve that by letting employees experiment in isolation. You cannot achieve it by asking IT to automate a few workflows. You achieve it only when AI becomes a true enterprise investment strategy, owned by the top of the house.
CEOs, Boards and senior management need to form cross-functional subcommittees—comprised of business leaders, not technologists—to define the strategic intent of AI for the organization. That includes:
- A vision for how AI will alter the company’s go-to-market model
- A redesign of customer experience and support models
- New approaches to operations and supply-chain orchestration
- Reinvention of employee experience and the future of work
- Structural changes to cost, margin, and capital allocation
This group—not IT, and not citizen developers—must determine the enterprise architecture, investment plan, and organizational design necessary to support AI at scale. IT will be a critical partner, but not the originating force. Citizen developers will be valuable participants, but not the architects of the enterprise roadmap.
When the “center weight” takes ownership, the organization can finally unify its AI investments. Governance becomes consistent. Data strategy becomes intentional. Agentic systems can operate across the value chain instead of within departmental islands. And the company can pursue new AI-enabled business models—models that may have been unimaginable just a few years ago.
From Experiments to Transformation
AI’s greatest promise will not come from scattered bottom-up creativity or from centralized automation mandates. It will come from leaders who recognize that artificial intelligence is a once-in-a-generation opportunity to reimagine the entire business.
The companies that win will be those that shift the weight from the ends of the barbell to the strategic center—where direction, investment, governance, and vision reside. AI must start as a business strategy, not a technology project. It must be orchestrated, not improvised. And it must be led from the top, not delegated downward.
The organizations that understand this now will define the next decade of competitive advantage. The ones that cling to the barbell extremes will simply modernize around the edges while their competitors reinvent the game.
