For the Optimization Catalyst podcast (listen here) Steve spoke with Suzay Passari, Head of Innovation and Transportation at UPS, and David O’Keefe, head of Alliances America at Gurobi. Following are lightly edited excerpts.
Sunzay noted that UPS, a 118-year-old company, presents unique challenges and opportunities for transformation due to its established legacy in technology, processes, and a long-tenured workforce. He views transformation at UPS as not solely about technology, but also about people and processes. Innovation must blend into a company’s local requirements, the culture, and the people.
David O’Keefe: Steve, you are an innovator and author about innovation. You are a pioneer in optimization, and you built your company from the ground up. What are your thoughts on innovation and transformation?
Steve Sashihara: Just last week a client asked, “What kind of background should we be looking up for in someone to lead our AI innovation effort that we’re starting?” It’s something of a coincidence that Sunzay mentioned his last position was as a product manager and owner. That job, which includes the new vision strategy and the new roadmap planning, brings together the thinking about what is going to be, rather than just moving groups. Right now in AI, the field is breaking apart very quickly. There is a high noise-to-signal ratio, but we all know that there is good stuff. I told my client that an effective leader will bring all the pieces together and achieve outcomes for the business.
David: Steve, what is your experience in transforming very large companies at scale?
Steve: Great question! The media loves to focus on disruptors, startups, and unicorns. These companies have bright people, but many are “pre‑revenue.” They may have few or no customers but they have great ambitions and private equity. In my view, the real high‑end stakes is where Sunzay and his peers are—at the “incumbents.” It is so much harder to change a large, successful company than a startup that can risk everything. At an incumbent, leaders demand something that is low-risk, guarantees results, and scales. Corporate leaders, in addition to low risk, seek high return. They are looking for big wins, big innovation—with low risk. That is the key to innovating at scale. For this reason, AI and LLMs do not replace mathematical optimization and solvers like Gurobi. Optimization brings proven methods to the AI game by powering solutions that reduce hallucinations and provide scalability and certainty.
David: In the logistics space, could you talk the challenges you face?
Steve: In Logistics, everyone is fighting for pennies. As Suzay noted, even if you are at the top of the food chain like UPS, you compete every day against businesses at the micro scale. That doesn’t happen in many industries, but it does in trucking and throughout freight transportation. It is critical—and challenging—that logistics solutions pay for themselves almost immediately and work at small, medium, or large scale. Massive upfront investments and years of hope are not welcome—solutions must work quickly and scale up.
On the positive side, a one percent or two percent improvement can be colossal because Logistics is so competitive and there is so much commoditization. If you could just do a little bit better, the whole world could be your oyster…
David: Do you see any risks of AI stagnation?
Steve: Not at all. I think a lot of people know “AI” has been slapped on a lot of things in the last 20 years that have nothing to do with AI. Shop carefully and make sure something works for your use cases. That said, you’ve got to seize the AI moment because the technology is powerful and wonderful. Use it to transform your business. Try to be in the front. You don’t have to be on the bleeding edge, but don’t be the laggard who gets pushed aside by the competition.
David: Steve, you’ve been modeling businesses your whole life. How do you build on failures to create success?
Steve: What do you do when you fail? I think the question is, “What caused the failure and should you hold people accountable because they screwed up?” If they did, then gently replace them with people that should be given a chance. In most cases, a lot of so-called failure is data. In a healthy innovative culture, people aren’t afraid of failure, they’re not afraid to experiment—the so-called failures are negative data. “We tried this, but customers weren’t interested so we closed it down and we’re trying something new. We’re trying and trying, and we’re collecting data.” Hopefully, wins outperform losses. If you focus on the losses and call them failures, you’re missing the bigger picture, which is that you are winning and making progress. I say to executives at large companies they should not fear failure, they should fear being so risk-averse that they lose the edge they earned all these years to get to where they are today.
David: Any last thoughts?
Steve: The first is that one of the most valuable things you have in your organization is your own data. That is what will bring a general purpose, a foundation model, to the highest level of value. It might be precisely your own data that gives you the edge, and you want to capture that and guard it. The second is that testing frameworks are critical. A lot of AI solutions today are in the Proof-of-Concept stage. As you scale them to the enterprise, do you have a rigorous testing framework that wasn’t supplied by the very vendor that you should be testing? It should be independent of a particular brand or chatbot or whatever that you are currently trying, so you can A/B against others and make sure the PoC won’t change its mind when you upgrade models.
