Fortune54%
The secrets of an unheralded AI success story 48%
By Jeremy Kahn46%
7/14/2026, 7:14:11 PM
Topics: AI Implementation, Business Strategy, Operational Efficiency, Employee Productivity, Cost Management, Innovation, Company Culture
Keywords: Artificial Intelligence, C H Robinson, Dave Bozeman, Logistics, Supply Chain, Productivity, Earnings, Freight Broker, Lean Management, Automation, Customer Quotes, Revenue Growth, Margin Expansion, Employee Productivity, Supply Chain Consulting, Smes, In House AI Models, Open Source Models, Engineers, Domain Knowledge, Token Costs, Socratic Method, Failure Mode And Effects Analysis, Fmea, Cultural Transformation, Risk Mitigation, Failure, Success, Operational Design, Culture
BS Summary: This article contains 27 faulty reasoning types, including Ambiguity (Equivocation), Confirmation Bias, and Optimism Bias, with Post Hoc (False Cause) as the most egregious example at 27.7% saturation with 313 hits. Analysis detected 2,059 faulty-reasoning hits from 1,128 analyzed words, generating a BS Score of 48.8% and a BS Rank of 48% (11,577 of 21,887 articles). This article is better (less manipulative) than 52.90% of the article peer group.
Hello and welcome to Eye on AI.
In this edition:
How Fortune 500 logistics firm C.H.
Robinson became an AI success story
Apple sues OpenAI for theft of trade secrets
Economists urge policymakers to take the threat of AI seriously
A new method for making frontier AI models safer
Is data the new bottleneck to AI progress?
There are a lot of Fortune 500 C-suite executives who still complain about not being able to get ROI from AI.
Dave Bozeman isn’t one of them.
The CEO of C.H.
Robinson Worldwide, a 120-year old logistics company headquartered in Eden Prairie, Minnesota, says the company’s use of AI has resulted in a 45% uplift in employee productivity since 2022.
Its use of AI has helped the company deliver double-digit earnings-per-share growth since 2023, despite a post-COVID slump in global shipping that has seen the company’s revenues drop some 34% over the same period.
Robinson, as the company is commonly known, is primarily a freight broker, specializing in what the industry calls LCL (less-than-container load) freight.
The company now deploys hundreds of AI agents across different aspects of its business.
A believer in “Lean management” —a system initially developed in Toyota’s manufacturing plants that focuses on maximizing customer value and eliminating waste—Bozeman, who has been Robinson’s CEO for the past three years, deployed teams to map out workflows and processes.
Any tasks that didn’t add value were eliminated.
Those that were essential but highly-routinized and repeatable, they’ve automated with AI agents.
For example, these agents now deliver quotes to customers, a process that once took human specialists 20 minutes, in just 31 seconds—and they operate around-the-clock, 365 days a year.
“It provides us not just productivity,” Bozeman tells me.
“This is revenue growth, margin expansion, productivity as well as customer advantage.”
He says by speeding up the time it takes to give customers quotes and providing more information to the customer, customers are more likely to submit jobs for quotations to Robinson, giving it more chances to win business.
Moving employees up not out
Like many executives, Bozeman is at pains to say his company’s embrace of AI isn’t about replacing human workers .
He says the company has been moving the shipping specialists who once provided quotations into higher-value work, like helping customers navigate shifting tariff regimes.
But that doesn’t mean there hasn’t been some labor savings.
Bozeman said the business had a natural employee turnover rate of 11% to 14% each year, and the use of AI agents means that Robinson has not had to hire new workers to replace those who have left.
The AI agents mean that for certain aspects of what Robinson does, such as providing those customer quotations, headcount is now largely divorced from volume in a way that was never possible before.
AI is also letting Bozeman contemplate strategic moves that the company might have struggled to execute previously.
Ultimately, his vision for Robinson is to be more than just a freight forwarder and shipping broker.
He wants the company to move towards being a supply chain consultant, and perhaps ultimately taking on the entire supply chain function for its customers.
“Think about it as ‘supply chain in a box,’” he says.
“I want to get to the point where a customer would say it’s going to be irresponsible not to do business with C.H.
Robinson, and it will be irresponsible for us to actually have a supply chain department.
Why do we need that when we have this company that can really do that, do it better than us, and allow us to focus on our core?”
Bozeman is also focusing more on serving small and medium-sized customers, an area where Robinson has lost market share in recent years.
Now, the CEO sees an opportunity to grab some of that back, with human sales reps assisted by AI agents.
In both of these domains—the high-value supply chain consulting and the servicing of more SMEs—Robinson is hiring more employees, Bozeman says.
It’s just that those workers have AI assistants helping them surface the insights their customers need.
How has Robinson been able to deploy all these AI agents without incurring crushing token costs?
The answer, Bozeman says, is that it has built almost all of them in-house using its own AI models or open-source models.
The company employs some 450 engineers, most of whom are steeped in the shipping industry—domain knowledge that Bozeman says has enabled the company to build better models than any third-party vendor could ever supply at a fraction of the cost.
Bozeman says that the company is currently “getting hundreds of millions of dollars of benefit with a token cost of less than $2 million.”
“This is a deep, wide moat,” he says.
“We calculated that if you wanted to replicate what we’re doing here, you would have to partner with 15 to 20 different entities to do that.”
A key to Robinson’s success in building these in-house AI models, he says, has been the operating mode he’s brought to Robinson.
When figuring out what agents to potentially build, Bozeman assembles cross-functional teams consisting of engineers, operational domain experts, and people from business departments like finance and legal.
He poses questions to them using the Socratic Method and has them debate solutions.
“That’s priceless when it comes to discovery.
It’s priceless when it comes to ingenuity,” he says.
There’s no success like failure
He also credits the AI success to other aspects of a cultural transformation he’s tried to implement at the company.
His teams use a FMEA (Failure Mode & Effects Analysis) methodology to game out how the AI systems they are building might fail and to mitigate those risks.
Bozeman has also pushed Robinson’s employees to embrace failure as a waypoint on the path to success.
“Failure is part of what we do,” he says.
He notes that when his teams report progress towards goals, they use a modified “traffic light” methodology that only allows two colors: green (on track) or red (off track.)
There’s no yellow; Bozeman says ‘yellow’ is usually really a red but the manager is afraid to say so.
Instead, he has tried to take the fear out of reporting a red.
“We say we celebrate the red.
If you’re red, you get the full weight of this organization to get you back to green,” he says.
“But you have to think about it, and how you problem-solve to get back to green is super important.”
It’s something I hear a lot from executives who report success deploying AI at scale: success is never just about the technology or about engineering talent.
It’s about operational design and culture too.
With that, here’s more AI news.
jeremy.kahn@fortune.com
Speakers
2speakers48%attributed speech582writer words
Voice mapSelect a segment to jump to its words
Selected voice
100%flagged-word coverageC.H. Robinson Worldwide
63 attributed words12% of attributed speech74% writer coverage
Attribution is sentence-level. Pattern percentages are calculated only from words assigned to that voice.
Loading…
Loading…
Loading…
Loading…
Analysis
Hover over highlighted words in the article to view the associated bias or fallacy analysis.