Copyright International Business Times

AI discussions often conjure images of sprawling data centers and high-profile partnerships among major corporations. That perspective is understandable. After all, scale and investment hold the potential for sweeping industry transformation, and it's tempting to think of AI primarily as a tool of the largest players. However, that picture is incomplete. Within large organizations, structural complexity can slow the journey from promising experiments to lasting impact. Raymond Sheen, President of Product and Process Innovation, Inc. (PPI), a consultancy known for blending operational insight with tailored advisory teams, highlights these internal frictions. He offers a complementary perspective that meaningful AI progress may also come from smaller, more agile teams. Looking across the landscape, a recurring constraint for many large firms is about usable, connected data. According to MIT's GenAI Divide 2025 report, 95% of enterprise AI initiatives deliver zero measurable return. This figure illustrates the widening gap between ambition and operational readiness. That statistic invites a careful look at why so many pilots fail to mature into impact. Sheen shares, "In our work at PPI, we keep seeing the same pattern. When operational records aren't organized in a way AI can learn from, the technology gets stuck at the surface instead of helping leaders make deeper decisions." Sheen emphasizes that this is not to suggest that large organizations cannot benefit from AI. Rather, the pathway is different. For many enterprises, the first step may be data readiness. Sheen notes that AI tends to learn from historical context: decisions, process outcomes, and the signals that connect them. "We need to complete the data set and have a full foundation; if there are any missing parts to the foundation, the building won't be stable," he explains. PPI's approach has been to treat AI as a business redesign problem first and a technology deployment second. This means mapping value streams, identifying where contextual data is created, and helping clients build the simple referent datasets that allow models to generalize in useful ways. "AI without an organized record is like teaching from memory alone," Sheen stresses. "You can repeat a lesson, but you cannot build a curriculum." By contrast, as Sheen has observed, small businesses and solo operators often have the unexpected advantage of concentrated data and short decision chains. Information that lives in a handful of spreadsheets or the inbox of a small leadership team can be easier to expose, clean, and iterate against. That accessibility can lower the barrier to testing ideas, tuning applications, and learning from results quickly. Sheen believes the most meaningful early wins may come from these actors. "Small teams can try an idea on Monday and learn by Friday," he says. "The speed of learning can beat the initial scale." Practical examples help illustrate the point. In one small manufacturing shop that Sheen advised, a simple language model was used to summarize long project proposals. What had once been a slow, time-consuming task became quick and automatic. The owner could now review more opportunities and spend more time talking with clients instead of being buried in paperwork. "The owner called the change transformational, and I agree," says Sheen. "Even small time savings can create big effects. Gaining back just an hour a day can lead to more client outreach, product improvements, and steady operational gains." Indeed, in a small business, those incremental advantages often have a much bigger impact than they would in a large organization, where results are spread across many layers of management. Sheen notes that agility and a lower risk profile also matter. Small teams can accept imperfect tooling as they learn, iterate quickly, and avoid the layers of governance that slow large-scale pilots. "They are not immune to challenge, but their environment often rewards rapid testing more directly," he states. The creative freedom to fail fast and refine is, in many ways, an engine of practical innovation that large firms may struggle to replicate without deliberate structural change. PPI's role, as Sheen describes it, is to bridge these realities. This means helping larger organizations cultivate the data practices that make AI a tool for meaningful decision support, and working with smaller enterprises to scale promising experiments responsibly. For leaders exploring practical approaches to AI, Sheen advises: "Start where your data is, design experiments that answer a clear business question, and let learning, not hype, set the pace."