Videos
Reinventing Product Identification: LKQ’s AI Journey
Insights
- Agentic AI enables real-time identification across hundreds of thousands of automotive SKUs.
- AI integration layers unlock value from fragmented legacy data without costly migration.
- Scaling AI requires data, infrastructure, and stakeholder readiness, not just technology.
This video highlights an AI pilot between LKQ, Infosys, and the University of Cambridge focused on solving large-scale automotive parts identification.
Bruno Zelić, Senior Director of Category Management Operations at LKQ Europe, explains how agentic AI improves speed and accuracy across highly fragmented markets. Rajiv Puri, VP and Head of AI for Manufacturing & Distribution at Infosys, outlines what it takes to move from pilot to scale. Professor James Fergusson, Director of the Infosys–Cambridge Enterprise AI Centre, shares how agentic systems connect legacy data into a flexible AI integration layer.
Together, the discussion shows how applied AI can deliver immediate operational value while laying the foundation for enterprise-scale transformation.
Bruno Zelić:
Category Management looks after products across Europe and more than 800 locations. An important function is data intelligence, which integrates AI agents in our daily processes.
Rajiv Puri:
LKQ is the leading distributor of automotive aftermarket parts. They are equally present in both North America as well as in Europe. They sell hundreds of thousands of SKUs for practically every single vehicle model out there.
Bruno Zelić:
The challenge that led to this AI pilot is to manage product identification, availability, and demand forecasting across many fragmented markets. It affects inventory, customer experience, and efficiency.
Rajiv Puri:
The research question that we are addressing jointly between Infosys, LKQ and Cambridge is all about helping LKQ’s sales associates identify the right SKU from the right catalogue for the right vehicle. You see, they have hundreds of thousands of SKUs linked to tens of thousands of vehicle classes and that is a huge pain point at this point in time. This is the big challenge that we're trying to address.
Professor James Fergusson:
We set up the Cambridge Centre really to look at research in areas that we think will fundamentally change the way science is done because we think that those areas have very wide applicability to actually how everyone does everything in industry as well. And that's why this collaboration is so interesting to us. So in this particular workshop, we're going to talk to LKQ about what I think is a really interesting and really critical problem for the AI transformation, which is what do you do when your data is fragmented across multiple legacy systems? And how can you use AI to get that data together into one place, essentially using AI to get it ready for using AI?
Bruno Zelić:
We choose AI because traditional systems can't handle our data volume and variety. AI learns from data, spot patterns, and gives smart and fast recommendations. This pilot uses machine learning and rule-based systems for real-time product identification.
Professor James Fergusson:
I think this use case is really exciting because it's essentially a way of avoiding this really painful transition that lots of companies go through. Because you can use data where it is by building agentic systems and putting model context protocols to connect to all of these systems. It becomes this really flexible integration layer between all of them that allows you to do real insights and real knowledge from existing data without having to move it or go through the expensive cost of moving it to one place.
Rajiv Puri:
The path from pilot to scaling hinges upon a couple of factors. First is data standardization across different categories. Second is infrastructure readiness. Third is technology readiness. And last but not the least is stakeholder readiness because ultimately this whole solution has to be used effectively by the stakeholders it is intended for.
Bruno Zelić:
The partnership with Infosys and Cambridge is important because Infosys knows our business well and brings strong industry expertise and tools.
And Cambridge adds cutting-edge AI research, being a world leader in advanced engineering. Together, they make the pilot practical, reliable, and innovative.
Professor James Fergusson:
And there's a huge amount of really interesting research that can come out of that. We work in physics, mostly fundamental physics, and we have a specific set of problems, but it's been really interesting to see what happens when you have lots of AI systems interacting. What do you get when you have like an ecosystem of AI, and what kind of emergent behaviors come out of that?
