AI Coach Brian Evergreen Tells Corporate America: Vision First, AI Second

Companies will pour over $1 trillion into AI this year, according to Goldman Sachs — yet Deloitte found only 6% see a return within a year. Brian Evergreen, the former global head of autonomous AI co-innovation at Microsoft Research, has built a booming coaching business on that gap. His message to executives is blunt: "AI doesn't have a vision. Leaders do."

Since launching his practice in 2023, Evergreen's coaching business has grown 200% each year. His client roster includes Microsoft, Accenture, Amazon Web Services, Salesforce, IBM, and NASA. One client, a leader at a global building materials manufacturer, put his value at "at least 20x what we paid." Clients credit him with saving them billions of dollars.

What does Evergreen actually teach?

His trademarked framework, Future Solving, rests on two claims. First, companies must choose the future they want to build and design a strategy around the resources they have. Second, that future depends on humans empowered by AI — not replaced by it.

"McDonald's got rid of human cashiers at some locations, and people drove out of their way to go to a location with a human," Evergreen says. "We're all wearing out the 0 button on our phones trying to reach a human."

He also pushes back on generative AI hype. Of the 27 types of AI identified by Gartner, he notes that machine learning and reinforcement learning have delivered measurable ROI — for example, letting manufacturing engineers test process improvements digitally without physical waste. Generative AI has not matched those results for his clients.

Why is Lululemon his latest test case?

Evergreen recently ran a workshop in Vancouver with Lululemon's guest support team, led by VP of guest support Shadi El Baba. El Baba invited him after an example about Expedia charging a premium for human help inspired him to "fully lean into the human element of customer support" and turn the team from a cost center into a revenue driver.

The stakes are high. Lululemon's stock fell to $102 in early September, an eight-year low. Its chief AI officer, Ranju Das, left in August after less than a year. New CEO Heidi O'Neill, hired in April 2026, couldn't start for five months, leaving the company in limbo.

"Lululemon has faced some headwinds, but I'm excited to see if Future Solving can help them surge to new growth," Evergreen says.

How does the method work?

The workshop process, run with about 15 leaders and a fireside chat for roughly 100 employees, follows a cascade:

• Define a vision for the guest experience, condensed into a two-sentence mantra.
• List everything that must be true to achieve it.
• Map each requirement down level by level until a short-term plan emerges.
• Revise upward, or start over, if any investment doesn't hold up.

Hours in, the team converged on one epiphany: to deliver a premier guest experience, Lululemon must give its "educators" — its employees — the premier work experience. "I'm proud we got here," one employee said. "It felt like we were on different chapters of the same book, but had we done this a year earlier we probably would have been on different books."

AI barely came up during the day. That is the point.

What's the track record behind the method?

Evergreen, a chess national champion at 12, joined Accenture in data entry in 2014, moved through Amazon and Microsoft, and rose to his Microsoft Research role in 2021. There he saw the problem firsthand. "I noticed these incredibly capable leaders — with incredibly capable teams and immense resources — were struggling to gain traction with AI," he says. "They were trying to use the same playbook that worked for digital transformation, but it wasn't working for AI."

After meeting over 1,500 Fortune 500 executives, his verdict is harsh: "consensus continues to be the emperor has no clothes — or no brains." His second book, Future Solving: Leaders Create What the World Becomes, arrives with a launch strategy built around handcrafted mugs bearing that motto.

At Lululemon, the early returns are visible in El Baba's reframing. His team's AI assistant Scout, debuted in early 2026, initially focused on reducing handle times. "What we should have been asking ourselves is what does it look like when we have the best and brightest educators, and how do we scale what they do to everybody?" El Baba says. "That's when the light bulb starts to click."