Moving Beyond AI Experimentation to Enterprise Value

At the third annual Wharton Business and Generative AI Conference in San Francisco, keynote speaker Lan Guan, Chief AI and Data Officer at Accenture, shared her perspective on the evolving AI landscape and what separates companies merely experimenting from those capturing real enterprise value. Guan leads Accenture’s Center of Advanced AI and oversees a global team of 77,000 AI and data practitioners. Drawing on decades of experience and over 6,000 advanced AI projects this year, she highlights where business leaders should focus their attention as AI adoption accelerates.

Business problems stay the same, but AI tools keep evolving

Guan emphasized that while technologies, from predictive analytics to generative AI, have changed dramatically, the underlying business challenges remain constant: growth, efficiency, and customer experience. “AI is just a set of tools,” she noted, and leaders should focus on selecting the right one to meet clear business objectives rather than chasing the latest trend.

Agentic AI is the next frontier, but requires clarity

Every executive conversation now touches on AI agents, but many confuse them with older automation tools. Guan cautioned against “agent washing” – rebranding simple bots as AI agents – and stressed that true agents should plan, pursue goals, and collaborate as digital teams. For businesses, this means rethinking workflows to capture exponential gains, not incremental efficiencies.

Physical AI opens doors across industries, not just manufacturing

Guan argued that physical AI, which combines digital intelligence with robotics and simulation, will impact every sector. Even banks or insurers can use digital twins to model processes and de-risk investments before making costly changes. “It’s about bringing physical and digital work together,” she said, challenging the misconception that robotics is irrelevant outside industrial settings.

Only 10% of companies scale AI successfully, and they share three traits

Based on Accenture’s work with over 1,500 organizations, Guan found that a small fraction achieve meaningful scale. The leaders share three practices: (1) a ruthless focus on value, not shiny objects; (2) intentional business cases that balance ambition with cost; and (3) designing for a significant change delta, a transformation that looks meaningfully different from the “as is” process, not just incremental tweaks.

The biggest obstacles: performance, budget, and talent gaps

Many companies stumble because their models underperform with messy data (performance gap), experimentation costs balloon when scaled (budget gap), or teams lack the full-stack expertise to manage modern AI systems (talent gap). Guan pointed to Accenture’s training of its 77,000 AI practitioners as an example of how organizations can proactively close these capability gaps.

Trust and responsibility must be built in from the start

With agentic AI, risks such as data poisoning or unintended agent interactions are growing. Guan stressed that trust and responsible AI principles cannot be afterthoughts. “It’s like working with a sharper knife. You need to know how to use it,” she said. Businesses that treat responsible AI as a design principle will be better positioned to scale safely.