According to Liz Eversoll, CEO of Career Highways, “AI doesn’t just expose structural weaknesses. It exposes the absence of skills and intelligence. Most enterprises don’t have a governed view of what skills exist in their workforce, how those skills map to job architecture, or how AI will shift demand curves. Without that foundation, leaders can’t design a rational response plan. The companies that integrate AI successfully are not just deploying tools. They are modernizing workforce infrastructure, governed job architecture, dynamic skills taxonomy, and career pathways that absorb automation rather than react to it.”
“If you can’t quantify skill impact, you default to blunt instruments—headcount reduction or hiring pauses—rather than targeted reskilling, redeployment, or workflow redesign,” says Eversoll.
“Before AI exposed where authority is unclear and incentives misaligned, most enterprises were already hiding those problems behind dashboards,” says Aaron Gibson, CEO of Hurree. “We’ve spent years working with companies, from mid-market teams to organizations like Comcast and Lloyds Bank, helping them unify data across dozens of tools. What we consistently see is that the dysfunction AI now surfaces was always legible in the data. Nobody was looking.”
“The distinction between summarizing an email and rebuilding a workflow is the difference between generating a prettier dashboard and actually changing what gets measured,” adds Gibson. “Most AI adoption is cosmetic analytics: faster reporting on the same misaligned KPIs. The decision architecture underneath doesn’t move. You’re optimizing the speedometer while the steering wheel is disconnected.”
In this Forbes commentary, Jason Snyder argues that enterprise AI is not underperforming because of weak models, but because organizations are mistaking visible adoption for structural transformation. While AI licenses are expanding, dashboards are glowing, and usage metrics are rising, measurable ROI and productivity gains remain modest. The disconnect, he contends, lies in how companies deploy AI.
Most enterprises have adopted AI at an assistive layer—improving drafting, summarization, and research—without redesigning the decision systems that determine cost structure, margin, and competitive advantage. True compounding value requires integrating AI into the “decision layer” of the organization: reshaping workflows, clarifying authority, modernizing incentives, and redesigning recurring decision processes. Without this structural shift, AI accelerates activity but does not transform economics.
The article draws parallels to prior technology waves such as ERP, CRM, and cloud adoption, where early gains were driven by installation and reporting improvements, but long-term value only materialized when operating models were fundamentally redesigned. AI, Snyder argues, is following a similar trajectory—diffusing quickly but compounding slowly.
A central theme is what Snyder calls “coordination theater”: the appearance of enterprise-wide alignment and progress through dashboards, presentations, and reported usage, while underlying workflows and decision rights remain unchanged. This creates an illusion of transformation without true integration.
The article also highlights workforce implications, quoting Career Highways CEO Liz Eversoll, who notes that many enterprises lack a governed view of workforce skills, job architecture, and evolving demand curves. Without a clear skills infrastructure, organizations default to blunt instruments—such as headcount reductions—rather than targeted reskilling, redeployment, and workflow redesign.
Ultimately, Snyder argues that the constraint on AI is no longer technological or economic, but managerial and structural. Companies that treat AI as a margin lever may see short-term efficiency gains. Those that redesign how decisions are made—modernizing workforce architecture, data governance, and accountability structures—will be the ones that achieve sustained, compounding advantage.