An Interview With James Sutton and Phillip Hall

If you spend any time on LinkedIn, it might seem like 2025 is the year entirely dedicated to artificial intelligence (AI). There’s a lot of buzz about AI’s potential—from writing code to writing blog posts, creating artwork to generating influencers. But how much is marketing hype, and where can AI deliver measurable results? Supply chain leaders, especially, are navigating this landscape cautiously, searching for targeted applications where AI can prove its real value. To help separate fact from fiction, we asked our experts, James Sutton and Phillip Hall, to share their insights.

AI in Supply Chain: A Practical Approach

Interviewer: Before we dive into AI in the supply chain specifically, could you briefly clarify what people mean when they talk about AI as a broader category versus generative AI specifically?

James Sutton: That’s a great question. When you hear about platforms like ChatGPT, Gemini, Claude, or Deepseek that’s generative AI—and that’s predominantly where most of the hype is coming from. The broader field of AI includes many other methods like machine learning, predictive analytics, robotics, computer vision, and natural language processing, each suited to different tasks. 

Interviewer: That distinction is helpful. Given that context, how do you see its role evolving in supply chain management?

AI’s potential to revolutionize supply chains is often overstated. Instead of trying to solve every problem at once, the best approach is to start with a problem where sufficient data already exists and then to build software specifically around that. Companies like EasyPost are focusing on one core issue: accurate time-in-transit predictions. We have billions of shipments and their transit events in our dataset, allowing us to reliably predict delivery times based on origin, destination, ship date, and carrier service. We’ve built a neural network that accounts for weekends, working days, and holidays—moving beyond outdated hard-coded delivery zones.

Interviewer: That’s fascinating. How is this being applied practically?

Sutton: This predictive capability is being integrated into our new software, Luma, to enhance rate shopping. Instead of relying on static transit maps, Luma ensures that shippers select the right service level to meet delivery promises. But AI’s role isn’t to make sweeping operational changes. For example, moving a distribution center based on AI predictions is a complex decision requiring deep consulting engagement with various factors beyond just data. 

Learn more about Luma.

Understanding AI’s Strengths and Weaknesses

Interviewer: Phillip, you’ve worked with AI across different business areas. What do you think are some common misconceptions?

Phillip Hall: One of the biggest misconceptions is generative AI’s ability to perform complex reasoning. People often ask me about Gen AI, and I tell them it’s not a computational engine—it’s a language model. It can predict answers based on patterns but doesn’t perform logical reasoning. That’s why Gen AI excels in applications like summarizing legal documents or transcribing conversations but struggles with analytical tasks that require deep domain knowledge.

Interviewer: So, Gen AI is great for some things but not for others?

Hall: Exactly. I recently had a discussion about Gen AI’s role in business intelligence. I wouldn’t trust it to fully explain what’s happening on a dashboard because nuanced interpretation often requires human context. However, it can quickly highlight patterns, suggest areas to investigate,  or help overcome writer’s block by suggesting starting points or potential questions to ask. It becomes a powerful tool when paired with human judgment.

Sutton: Gen AI is very good at summarizing context, understanding, and putting things together. It can help us understand complex documents, for example. But it all comes down to what it’s being fed. AI, generative or otherwise, is only as effective as the data feeding it. Many organizations still have foundational data problems, and AI won’t fix a flawed data pipeline. If a company isn’t capturing clean, accurate, and relevant data, AI can’t generate meaningful insights.

The Shift Toward Targeted AI Solutions

Interviewer: So rather than an all-encompassing AI solution, would you say the future is more in  targeted applications?

Hall: That’s the trend we’re seeing. One of our clients has a GenAI data product —a chatbot designed for simple data queries like “What’s my sales volume?” or “What’s my average check?” It’s tightly scoped for non-technical users who just want quick, accurate answers. 

That same client is also using it to great effect to accelerate sentiment analysis and help categorize text-based data and tie it back to strategic corporate goals. They’re now using this across Customer Support, Social Channels, Customer Surveys, etc.

Interviewer: That seems like a smart way to integrate AI without overpromising.

Hall: Exactly. By keeping AI applications focused, businesses can drive efficiency without replacing critical thinking. Blindly trusting AI to make decisions requiring years of experience is risky. But using it to enhance efficiency, improve language-based tasks, and accelerate data adoption? That’s where it thrives.

What’s Next for Supply Chain AI?

Interviewer: What do you see as the next big development in AI for supply chain management?

Sutton: As AI tools become more sophisticated, their adoption in supply chain management will continue to grow. However, success will depend on setting realistic expectations and integrating AI where it can truly add value.

Interviewer: So, AI isn’t replacing human expertise anytime soon?

Sutton: No, and it shouldn’t. AI isn’t going to replace supply chain strategists, but it will make us better at predicting outcomes, optimizing decision-making, and reducing inefficiencies. That’s the future we should be working toward, using AI to complement human expertise rather than replace it.

As the logistics industry advances, companies that harness AI for specific, practical applications—rather than sweeping transformations—will gain a competitive edge. The key is knowing where AI can add value and setting realistic expectations, rather than expecting it to solve every challenge overnight.

If you’re looking to harness powerful AI capabilities on your own data, Phil and Summit Advisory Team would love to help.