Dave Bozeman is president and chief executive of C.H. Robinson, one of the world’s largest logistics companies, with 83,000 customers and 37 million shipments a year. He was interviewed by Asutosh Padhi, who leads McKinsey’s global firm strategy, for the series The Exchange. The conversation was recorded on 14 January 2026 and published on McKinsey’s site on 20 March 2026. It covered what Bozeman calls lean AI, the operating model behind it, the cultural change it required and what AI means for the company’s people.
A year of results
Bozeman said AI had moved in a year from early experiments into the company’s most important workflows, from pricing to execution and exception management, compressing tasks that took hours or days into seconds. He reported productivity gains of more than 40% since the end of 2022, up from the 15% he had described a year earlier. C.H. Robinson has automated more than three million shipping tasks, built more than 30 AI agents, saves 600 hours a day through its order agent alone and has automated appointments at 43,000 locations. He also pointed to eight quarters of market outperformance during a freight recession.
What lean AI means
Bozeman has practised lean management for more than three decades and calls its combination with AI the company’s competitive moat. Every initiative starts with a real customer or employee pain point, and teams go to where the work happens to observe errors before deciding what to change. He described logistics as a people industry, with technology there to help staff solve complex problems.
An operating model built on inputs
The operating model has shifted from judging outputs months later to tracking inputs in real time. An enterprise scorecard cascades into business-level metrics tracked hourly, daily or weekly, which roll up into enterprise KPIs reviewed every two weeks or monthly, such as win rates and truckload percentages. When a measure turns red, the team asks why, and the person is not blamed. Bozeman said the hard part is the work done up front to choose the KPIs that define the company’s right to win, and changing them quickly if they turn out to be wrong.
Culture and people
Cultural change, Bozeman said, starts with trust and with leaders setting the example. He spent time at employees’ desks watching how the work was done. The company had been "problem admirers", in his words, and now solves problems fast.
He said the company has no KPI for headcount. In the order-to-cash process, a quote that used to take 15 to 30 minutes now takes 32 seconds, at any hour and with more detail, which frees the same employee to talk to customers directly.
What comes next
Bozeman sees agentic AI, which can reach data outside the company’s own systems and reason over it, as the next step after machine learning and generative AI. He expects visibility across the supply chain and fast problem-solving to be what customers want most from a logistics partner, and expects the next two years to bring more change than the last two.
What it means for leaders
- Starting each AI project from an observed pain point, as lean practice does, keeps automation tied to work that matters to customers and staff.
- Input metrics tracked daily or hourly let teams correct course long before quarterly results arrive, provided a red metric leads to questions about the process.
- Productivity claims are more credible with a baseline and a date. C.H. Robinson states its gains against the end of 2022.
- Time freed from quoting and order entry pays off only when it is redirected to customers, which Bozeman describes as the intent.

