Data, culture and people: The real barriers to AI in automotive logistics

The conversation around artificial intelligence in automotive logistics has shifted from whether to engage with it to how fast, how deep, and how do you take people with you in a span of one year.

At the Automotive Logistics and Supply Chain UK conference in June 2026, during panel discussion with experts from Ford, Scania, SEEBURGER and Aston Martin, a unanimous argument emerged that technology is not the constraint, but data foundations, organisational culture and human capability are.

Designing before building

Stuart Clarke, manager, material flow and packaging engineering at Ford asked a question that the industry has struggled to answer: how do you make logistics legible to people who didn't design it?

Ford's answer has been extended reality - XR technology that converts flat 2D plant layouts into fully immersive 3D environments, allowing engineers, operators and senior stakeholders to walk through a warehouse or production line before a single physical change is made.

The gains are significant and Ford has estimated that it can take 10% out of base logistics costs at the concept stage alone, with further reductions in rework, travel and iteration costs as the system matures.

A misdirected kit area, a mistake Clarke said has previously cost €20,000 to €30,000 ($22,731 to $32,731) to fix, can now become visible before installation. A forklift driver's sightline at a blind corner can be assessed before the corner exists. Packaging can be validated with operators wearing wireless gloves in a virtual environment, allowing the design to be shaped around how people actually behave rather than how they are supposed to behave.

The system is now in use across Europe and rolling out into Ford's international manufacturing group. Clarke has set an 18-month roadmap to integrate XR models into Siemens Teamcenter, Ford's global engineering data platform, so they are accessible and updatable across the entire organisation.

The longer-term ambition is automation of model updates, potentially supported by AI embedded within the software itself.

From EDI to autonomous logistics

Johan Lindl, head of digitalisation and AI at Scania Logistics, and Ulf Persson, senior vice president of strategic product management at SEEBURGER, presented a transformation journey that began with a far less exciting challenge: how do you scale supplier collaboration to 2,000 partners without scaling your headcount at the same rate?

Three years ago, Scania was establishing its third industrial hub in China with a goal of building volumes equivalent to their entire current global truck production. The problem, however, was that the company was running a one-to-one model between supplier collaboration and human agents, with 18 full-time equivalent staff managing B2B integration through legacy EDI infrastructure. Meaning, that model was not going to scale.

Scania and SEEBURGER built an EDI-as-a-service capability in cloud, with supplier self-onboarding, automated entity creation and an API-first architecture that absorbed onboarding tasks that previously required specialist developers. The result was a reduction from 18 FTEs to four, operating at higher capacity and lower cost.

"When we had the ChatGPT moment, we had the foundation," Lindl said. "And we had organised ourselves in a different way, where we invested in people who wanted to work with strategy and innovation. And then, after the ChatGPT moment, when we did our AI literacy and AI adoption push and put AI into the hands of these people, we really did see magic."

The first output was EDIT, an AI assistant for EDI first-line supplier support. When Scania went live in China last year, EDIT provided 24/7 support in any language to parts suppliers without adding a single person to the support team. But EDIT has since evolved. It now executes autonomous work in the EDI environment and is trusted to make decisions, not just answer questions.

Lindl described a view of logistics organisation that is moving away from process obsessiveness, what he called "limitations of humans at work", towards something more fundamental. "From process intelligence to decision intelligence," he said. 

"A lot of our process obsessivity is related to limitations of competence, functional silos and a lot of handoffs." Scania is now close to concluding a "proof lane" for Level 4 fully autonomous logistics within one business area - packaging. 

Persson cautioned that technology resilience, the need for software platforms to be self-improving and agentic by design, is now as important a consideration as supply chain resilience.

The scale of what is coming is not abstract, as IDC estimates there are already 28 million AI agents doing some form of business work globally. By 2030, that figure is projected to reach a billion. For example, BMW, for its part, has already deployed agentic AI to transform purchasing processes, with 80% of inventory and procurement activities now handled entirely by agents, with human intervention required only for deviations or critical escalations.

Skills, people and the marriage that matters

Robert Gammon, director of planning, vehicle logistics & SC transformation, Aston Martin, was direct about where the real challenge lies. Finding strong supply chain talent is already hard, he said, and the pace of AI adoption is compounding the problem. "Not only does somebody have to have supply chain skills, they have to have IT literacy skills, good business acumen, good analytical skills. So it's a huge, huge challenge."

Andreas Neumeister, global account director, DP World, pointed to a parallel gap in infrastructure investment, citing the UK's ambition to increase rail capacity by 75% by 2050 as an illustration of the scale of commitment required, and the planning that must precede it.

The procurement model, Neumaster argued, needs to change alongside the infrastructure. OEMs too often source logistics transactionally, looking at single parts of the supply chain in isolation. 

"We need to take a broader view," he said. "Looking at the full supply chain and trying to find the right solution with the optimum, not just looking at one single piece of it, but combining it together."

Aston Martin has been building exactly this kind of holistic visibility. His team has developed a cost-to-serve model that factors in CO₂, packaging and logistics flows together, replacing the traditional procurement-led cost-per-piece logic.

The result, he said, is smarter decisions like fewer instances of moving large components back and forth across Europe when sourcing closer to build makes more sense. "People are considering: this costs money, source close to where you build."

On the people question, both Gammon and Neumaster circled back to the same conclusion that digitalisation does not eliminate jobs; it changes them. The Waymo autonomous taxis proliferating in San Francisco, he noted, have still created a new job, someone must close the doors that passengers forget to shut.

"The system doesn't work without a different sort of skill set," Gammon said. "It's that marriage for me."

What it really takes to embed AI in logistics operations

Following the presentations from Ford and Scania, Stuart Clarke, Johan Lindl and Ulf Persson took to the panel stage to interrogate the harder questions behind the transformation stories about not what the technology can do, but what organisations need to become before it can do it.

Leadership and the danger of technology for technology's sake

Clarke and Lindl were unambiguous on one point - passionate, committed leadership is non-negotiable. Without it, digital programmes stall at the proof-of-concept stage or become what Clarke called "a glorified business toy" that delivers no return on investment and no credibility with the shop floor.

"You have to understand the why," Clarke said. "We run the risk of technology for technology's sake rather than technology because it solves a problem or fixes an issue." For Lindl, the answer was structural after Scania reorganised its logistics operations away from function-based silos towards cross-functional teams built around business value and a single, business-prioritised backlog. "Material planning is material planning," Lindl said, "regardless of what application is supporting in the background."

Scaling without losing focus

Persson, drawing on SEEBURGER's work across the automotive sector, identified that organisations are securing AI budgets and trying to do too many things simultaneously. "Typically, seven or eight out of ten initiatives fail," he said, "because you don't start with something that makes sense." 

Clarke echoed this from Ford's own experience. The XR software his team uses covers far more capability than they currently need, and that, he argued, is entirely deliberate. "You run a real risk of using the software to 80 or 90% of its capability but it doesn't actually solve your problem in the first place," he said. "It comes back again to the why."

Experimentation needs permission to fail

In logistics and manufacturing, the stakes of getting things wrong are high. But Clarke argued that a culture of risk-aversion is itself a form of risk. "Fail could actually stand for first attempt in learning," he said. "You have to get that culture where people aren't afraid to have a go. If they're afraid to fail, they'll never try."

Lindl added that AI literacy is now a mandatory white-collar capability at Scania, embedded across the enterprise, not confined to a digital team. "We want the people to drive the shift," he said. "Not watch the shift happen."

What comes next

Both speakers pointed to agentic AI as the defining territory of the next phase, not as a future concept but as something already in live deployment at Scania, and rapidly approaching for the rest of the industry. Persson noted that the vendor landscape is evolving so quickly that the question of what software to buy has become inseparable from the question of how fast it will become obsolete. "Whatever assets you buy," he said, "they need to be self-improving."

For heavy transport specifically, Lindl made the case for electric more concretely than most: a Scania battery electric vehicle (BEV) with a gross capacity of over 60 tonnes can now charge for a 400km range in 35 to 40 minutes, with a break-even point for going electric at 4,750km per month - a threshold, he noted.