Under Lan Guan, Accenture has shifted from advising its clients on AI use to being more of an orchestrator, guiding them on how best to harness the technology. In 2025, the consulting giant expanded its Nvidia-backed AI Refinery platform with industry-specific agents and governance tools, and launched its Trusted Agent Huddle, which enables agents built by different vendors to work in unison across different enterprise systems.
Accenture works with many Fortune 500 companies, helping them deploy AI without becoming locked into a single model or platform. The company has already generated billions of dollars in generative AI bookings, while Guan says deployments for clients like the Australian bank Westpac have cut some workflows from months to days.
Now she’s tackling the cost of scaling those systems. Accenture has recently focused on tokenomics, arguing that firms incorrectly default to the most powerful—and expensive—models even when the work doesn’t require such heft. “Only about 20% of enterprise workflows actually deserve frontier models,” she says.