A conversation with Bas Groothedde, founder of verticallm, on George Lucas, a bitter truth, and picking a fight.
1) This interview was conducted as part of the inaugural issue of Vertical AI Magazine: Research, Practice, and Innovation in Applied AI, to be published in September 2025.
MIT SMR: You’ve spent over two decades working in the trenches of supply chain consulting. What drove you to launch verticallm?
Bas: It started with a very simple insight: time is always in short supply—especially in complex supply chain projects. Over the past 20 years, our team worked on everything from M&A integrations to network redesigns, lease transitions, capacity bottlenecks—you name it. And no matter how different the client or industry, the challenge was always the same: how do we understand the situation deeply and quickly, without losing strategic depth? We need to understand the context we are working in as quickly as possible. That has become our second nature.
So we built a methodology—using streamlined approach, survey-based, structured, but sharp. These surveys were never just forms; over time, they evolved into powerful diagnostic tools. They allowed us to quickly capture business context—one of the hardest things in our field.
You’re parachuted into a company and expected to grasp the language, the processes, and the dynamics as fast as possible. Only then can you move to the next step: gathering, processing, and applying data in a meaningful way.
‘We need to redefine consultancy as we know it.’
By building a platform to support supply chain professionals and leverage AI to transform context, data and input into clear actionable, individualized goals.
MIT SMR: So this data driven and giving-context became the foundation of the platform?
Bas: Exactly. The methodology grew with us. Over time, we developed versions focused on strategy, resilience, maturity, and organizational structure. These tools became an extension of how we think and operate.
I’m a huge Star Wars fan, so I use analogies whenever I can—and in this case, I really did feel like George Lucas for years. I had the script, the scenes, the structure—but the technology just wasn’t there to bring it to life at scale. Now it is. With Verticallm, we’ve embedded that methodology into a vertical AI model that truly understands the logic of supply chains.
It’s not generic AI—it’s built specifically for this field. Enclosed, safe.
'We didn’t pivot to AI—we built AI into what we do best'
MIT SMR: What does “vertical AI” mean in this context?
Bas: It means an AI trained with domain-specific data, language, and logic. We don’t want AI to improvise, so we guide it with structure and context, just like we would a new analyst or developer joining our team. And that’s exactly what we’ve been doing over the past year.
We made a deliberate choice not to invest in building our own LLM technology—that space is already moving fast. Nor did we focus on compute infrastructure; that will keep evolving. Instead, we invested in what we believe matters most: context. Our goal is to feed these incredibly smart systems the right data, in the right way, so they can actually use their strengths.
We use datagraph technology to deliver structured context to the LLMs as efficiently and effectively as possible—giving the AI background, structure, and business logic in a format it can immediately act on.
So our supply chain vertical combines three elements: our deep understanding of what matters in supply chains, the client’s real-world input, and the speed and scale of AI running on a modern infrastructure. That combination allows us to deliver immediate value. No need to wait for industry benchmarks or a consulting sprint.
Bas: Not long ago, I was at a kind of hackathon in Cambridge—an event for young, brilliant minds. I was there just to help out a bit and guide. But I was useless. The participants were given a dataset from one of the new automotive mobility car sharing company.
Then I witnessed what these twenty-somethings could do. They connected data, pulled in external sources, built models, spotted patterns, and pitching brand new business models. And none of it had anything to do with mobility. Their creativity was unbelievable.
Now, I’m an engineer. I’ve done a PhD. I studied at MIT Sloan. I’ve built machine learning models, worked for some amazing companies, spent 25 years in this field. So yes, I tend to lean on experience. I thought I’d seen it all. But that day? I was completely floored.
“It was funny—and a bit unsettling. I realized: something truly new is happening. Not just faster computers. A different way of thinking. It was the powerful combination of A datagraph as the enabling platform. An industry vertical capturing domain knowledge. AI to make it all intelligent and fast. And then the fourth—maybe the most important one: the human supported by AI. The augmented professional. That’s when I knew : this is the future of work.”
MIT SMR: How have your clients responded to this shift?
Bas: It’s been energizing. At first, clients loved the insights. But then they asked: “Can we have access to the engine itself?” And that made perfect sense.
Now, we have a group of close companies that help us develop our platform and we help them implement their own version of the vertical model—on-premise, private, and embedded in their teams. It interacts like ChatGPT, but it’s focused entirely on their supply chain, their data, their context. That’s what we call the augmented professional—human intelligence empowered by structured AI.
This has been so valuable for both, that we now opened up a Second Group for additional Development Partners. They get early access and the help us to improve upon our current platform.
MIT SMR: And where do you see Verticallm going next?
Bas: We’re just getting started. The vertical is already helping companies drive decisions today. But we’re working on even tighter integration—connecting the AI directly to operational data, making it context-aware in real time.
The idea is to create a supply chain intelligence platform that’s always on, always relevant, and always learning. We’re combining vertical models, datagraphs, and prompt engineering into something that feels less like a tool—and more like a strategic co-pilot.
MIT SMR: You’ve said this changes what consultancy looks like. What do you mean by that?
Bas: Traditional consultancy is linear: diagnose, design, deliver. It’s relatively slow and expensive. We take away part of this friction. Clients can now move at the speed of business—backed by insights grounded in research, context, and AI.
MIT SMR: Final question: what excites you most about where you are now?
Bas: Honestly? What excites me most is that everything is finally coming together. I feel incredibly lucky to be part of this moment, to work with such a talented team, and to collaborate with industry partners who bring just as much experience and energy to the table.
This feels like the right time, and I couldn’t be more excited about what we’re creating.
