AI is stealing the spotlight in boardrooms everywhere. It’s the buzzword in earnings calls and a “feature” in almost all tools employees use. But ask most companies about their ROI from AI, and you’ll often get silence, shrugs, or even worse, vague metrics with no real business impact. According to Gartner, over 40% of agentic AI projects are predicted to be cancelled by 2027 due to soaring costs, ambiguous business value, or lacking risk controls.
Why is this happening?
It’s not because AI doesn’t work, it does. The issue lies in how companies identify the solutions they need. Or more importantly, who is making those decisions. Most organizations are relying on the wrong people to guide the most critical technology decisions of the decade. And they’re paying the price for it.
The AI Decision Making Disconnect: Why Current Strategies Fall Short
Many organizations are grappling with underwhelming returns on their AI investments, and the reasons are becoming increasingly clear. Despite good intentions, current AI implementation strategies are often undermined by two fundamental flaws: misguided decision making structures and flawed validation processes. This leads to a significant gap between AI’s potential and its real world impact.
When the Wrong People Make the Call
Current AI committees, while appearing robust on paper with senior leadership representation (CIOs, COOs, etc.), are frequently detached from the day-to-day operational realities AI is meant to enhance. Most impactful AI use cases today such as drafting documents, summarizing meetings, or automating customer service replies are operational. These tasks are performed by associates, coordinators, and interns who possess an intimate understanding of existing inefficiencies. Yet, these crucial frontline employees are rarely involved in strategic discussions around AI implementation.
Excluding those closest to the work results in an AI strategy that is fundamentally out of touch. It results in AI that does something but not necessarily what’s needed. Tools are deployed without proper workflow integration, processes are automated in ways that generate more overhead, and the very employees meant to benefit perceive AI as an added layer of complexity rather than a solution. This isn’t transformative; it’s performative, and it actively hinders business agility and growth.
The Mirage of Validation: Beyond Marketing Hype
The second significant challenge lies in how AI solutions are validated and procured. A substantial portion of purchasing decisions are driven by perceived market momentum, fuelled by extensive marketing, paid search campaigns, and conference visibility. This often means that what’s deemed the “best” solution is merely the loudest, not necessarily the most effective.
Venture capital influence also plays a role. While VC due diligence is rigorous in selecting investments, their network driven recommendations to CIOs, though well intended, don’t always guarantee optimal fit. Highly promoted AI tools can sometimes be “hollow,” with their validation stemming from marketing spend rather than demonstrable customer outcomes or genuine functionality. This creates a bizarre loop where perceived ubiquity is mistaken for superior performance, overlooking truly effective tools that are quietly delivering value through deep integrations and high user satisfaction.
Rebuilding for Real ROI: A Fundamental Shift
To unlock the true ROI of AI, companies must fundamentally rethink who makes AI decisions and how those decisions are made.
First, AI committees must be restructured. Front line managers and team leads should constitute at least 50% of these groups, armed with genuine decision-making authority, not just advisory roles. If an AI committee cannot articulate the nuances of a specific operational process like client onboarding, enterprise search, or pitch deck generation, it is a clear signal that the committee is misaligned with the actual needs of the business.
Second, CIOs must adopt a more rigorous interrogation of their choices. The question should not merely be, “Does this tool have a good demo?” but rather, “Is this the best tool for our specific use case, and how do we definitively know this?” Relying on trusted recommendations or CEO panel appearances is insufficient. Validation must stem from direct, rigorous testing and demonstrated performance in real world scenarios. This demands a disciplined approach, challenging consensus and discerning earned credibility and manufactured visibility.
From Hype to Tangible Impact: Embedding AI Where it Matters
Ultimately, AI’s promise was never about buzz; it was about delivering better outcomes: faster execution, reduced friction, higher quality, and fewer errors. This transformation only materializes when the right AI is seamlessly integrated into the core of operational processes in content creation, internal communication, document workflows, and operational decision making.
Too often, companies are merely “bolting on” AI at the periphery instead of embedding it where the actual work begins. Until leaders recognize AI as a foundational component of their operational engine, they will continue to miss its transformative potential. Empowering the individuals who perform the work to shape how AI enhances it, and choosing solutions based on genuine fit and function rather than familiarity or fanfare, will shift AI from a mere organizational acquisition to a powerful driver of daily business progress.
