From chatbots to control towers: How OEMs and suppliers are using AI to strengthen automotive supply chains and manage risk

While the automotive industry has excelled at technology adoption in the past, particularly in the form of physical innovations within plants and production lines, it has been slower to catch up in software solutions, data quality and sharing, and integration of pilots into the supply chain. But some OEMs, logistics providers and suppliers are leading the pack and have been piloting data analytics and AI solutions that could be scaled up and shared across the automotive logistics landscape.

While some are still in their infancy, these real-world applications of AI are being tested now to iron out flaws in data quality and management, spot potential gaps in performance, and most importantly, to train people in the supply chain on their benefits.

Real-time supply chain risk management with AI

AI is moving beyond theory and into practical applications and use cases – one of the most exciting being risk management. There has been a shift from reactive problem-solving to proactive real-time logistics orchestration, with the goal of faster decisions and greater decision coverage, ensuring AI tools can handle a high percentage of issues autonomously and freeing up planners in teams for more strategic tasks.

At our recent Automotive Logistics & Supply Chain Europe 2025 conference, Dr Gisela Linge, vice-president of global logistics at Autoliv explained how what was once thought of as a dream future is now possible using AI. She said that Autoliv is using machine learning (ML) for demand forecasting, predictive maintenance and quality control. Autoliv has built a control tower digital twin which it is enhancing daily with more use cases and growing data, and has an AI chatbot which summarises standards across functions and the company to answer questions, help to build training materials, or process descriptions. Similarly, Fabian Pobantz, vice-president of operation digitalisation & IT, supply chain and purchasing at Schaeffler, said AI has been a part of Schaeffler’s forecasting algorithm for many years, and the company is now searching for how this forecasting accuracy with AI can be shared with OEMs.

Pobantz explained that the use cases for AI in forecasting accuracy are growing more important, whereas previously emphasis may have been on things like inventory optimisation.

“This year, it’s about execution and taking it to the broader scale,” Pobantz said. “There is a focus on data quality, and forecasting accuracy which is a common pain point, because that drives the entire supply chain.”

Giving advice to the industry based on Schaeffler’s experience, he said: “Make [your use case] more granular and adapt to your business condition. Small successes with iterations will give you good results, and engagement of the leadership, but also make you aware of the upcoming challenges in regards of processes, standardised systems, connectivity, data quality, accuracy or data acquisition. If these are challenges, work on those first.”

Linge said that Autoliv has been using its control tower and AI tools to help deal with recent supply chain disruptions, such as the tariffs enforced in recent months.

“When we talk about these control towers and digital twins it’s a lot about risk mitigation, getting transparency and being able to react fast, for example we used it when the whole tariff topic gained momentum,” she said. “It’s really good if, within a short time frame of a couple of hours, you can really tell your CEO, ‘This could be the annual impact,’ because then you get more clarity and what we all hate is uncertainty.”

Giving a recent example of real-world AI application at Autoliv, Linge said: “We recently dealt with an incident at a key supplier and the digital twin helped us to immediately see which customers are affected. We could see where, if we didn’t get the inventories out from the different regions, there could be a potential production line stop. It’s a great foundation then for the task force team to work on and get the details.”

“When we look into the planning systems it’s more about improving our decision making, being data enhanced, and for us in logistics, is it a way in demand forecasting to reduce our inventories, our obsoletes, so that’s also a big goal and then in the end making the whole process more efficient.”

We have also seen OEMs begin to make use of AI tools for risk mitigation use cases. Since 2023, JLR has been using Everstream’s AI to monitor for risks such as natural disasters, strikes, data breaches and export issues that could delay shipments. The technology uses AI, predictive analytics and machine learning to help avoid disruption, in combination with “human intuition”, as part of a wider strategy at JLR to build end-to-end visibility throughout its supply chain. 

Similarly, BMW has been deploying AI technology across its plants, including at its largest global plant, Plant Spartanburg, where AI is already used daily in its body and assembly shops for quality checks. BMW has also integrated AI-driven systems in its San Luis Potosí plant in Mexico, using historical data to make predictions on the supply chain.

And GM has been using machine learning to give the OEM parameters of the highest risk suppliers in its network, allowing GM to engage with them proactively to put mitigating actions in place.

Human-AI collaboration in automotive logistics

In her AI dream world, Linge imagines AI chatbots with human names, working together with humans in logistics teams to collaborate and share information. “Is this really a dream, or how far away is this?” she asked. In reality, it’s pretty close to how AI is being viewed across the logistics industry.

“I give them names, because AI would be involved as real team members,” she said. “That also means that, like with human team members, AI has to be trained and introduced, and you have to learn how to interact with it.”

She said that she envisions using AI agents to focus on visualising and explaining data trends, while human members of the team focus on parts of demand forecasting that the machine learning cannot reliably predict, using human interaction with customers. She said that the team members and AI could work in tandem to enrich the data foundation that the AI would be working on, with more cross-functional and external collaboration.

That AI should be treated as part of the team, but with a human-centric approach, is an increasingly popular school of thought within the automotive logistics industry.

Pobantz said: “We want to really move into the human-in-the-loop approach, where the AI agent is now taking decisions in certain processes and we just get the updates or the information on what decisions it has taken, because the human has to be always in the loop.”

He added: “It’s not that AI will replace people, it’s more that people with AI skills will be replacing the people without those AI skills. For me, it’s important to acknowledge that it is not only the new hires that we need to care about, we also need to upskill our organisation and train them into prompting.”

VW Group has been experimenting with AI use for risk analysis and network optimisation, and is looking at new opportunities in generative AI. One of the group’s most advanced examples of connected operations comes in the form of Seat’s digital control tower, which has helped the carmaker mitigate disruptions and better plan for capacity flexing.

Read more: Seat’s logistics director Enric Brau wants to inspire boldness in digitalisation and electrification

Seat’s IT and logistics team have been developing this AI further, including its proprietary genAI chatbot, known as Vortex AI. But a digital strategy without people will not get far, which is why the OEM is centering humans in its AI approach. Seat is ensuring thousands of its workers are being upskilled and trained to use AI and fully understand the challenges and benefits of the tool.

This is something that’s echoed by Linge. When piloting AI implementation at Autoliv, the only consultant she brought onto the project was a change management expert, who was there to ensure the team members were not only able to use and understand the AI technology, but also feel comfortable with this new way of working.

“On the human side, the only consultant we had on this project was a change management expert to make sure that we keep everybody on board,” she said. “Be curious, be collaborative and do it together, not only with your IT but with your users.”

Why data quality is critical for AI in logistics 

Of course, for AI tools to work, they need to have a solid foundation of supply chain data that is accurate, up to date, and governed correctly, and this involves lots of human-centred data management.

“If you really embark on this AI journey, you really must make sure that the data stays current and the quality is sustained,” she said. “These old ways that most of us have done in the past, like doing a big cleanup of certain master data and then not caring and after a couple of months you have the same mess, this will not work because when we are successful in this digitalisation we make people trust in whatever these chatbots or agents do and decide. People may become lazy and may not check data inconsistencies, so you have to find a way to set up governance that looks for more sustainable ways of checking on data.”

Linge said that she has data analysts and data scientists in her team because humans are needed to manage this data governance, support the roll-out of the technology, and take the right decisions based on whatever digital tools are used.

Mercedes-Benz also prioritises data quality as a strategic necessity, rather than a supplementary function. John Torres, lead senior data scientist at the OEM, previously told Automotive Logistics that the carmaker ensures processes are process and aligned with business objectives by monitoring real-time metrics of data completeness, transparency and analysis. “Data quality is actually one of the most important factors that we have in our organisation to actually leverage AI and analytics,” Torres said. “We have developed inhouse tools to actually monitor data quality in real-time.”

Linge said that one of the best ways to check your data quality is to start trialling AI using the data you already have – that way, you’ll spot any gaps or data inconsistencies. This is something that Paulina Chmielarz, industrial operations digital and innovation director, JLR previously told Automotive Logistics. She said the OEM has invested significantly in its data management and governance to prepare for AI. “Starting your journey with AI will initially show you how bad your data is,” she said. “When you address your data, the journey of AI is much easier. Proper platforms, proper management and governance of the data, and focusing specifically on the quality of the data and monitoring that quality and fixing the root causes is key.” JLR has been in a cautious test phase with AI before adopting it at scale. The carmaker is first testing safety and quality of data, particularly in the maintenance side of operations to support with additional information.

AI investment strategies and partnerships in automotive

AI integration into the supply chain is going to take a lot of investment and training. In Europe, government initiatives are gaining momentum and helping to accelerate AI implementation in supply chains. The European Union announced a €200 billion ($214 billion) InvestAI Initiative, which aims to provide a funding boost for AI projects in sectors including robotics, biotech, mobility and manufacturing.

But in the absence of government funding, supply chain firms can partner with one another, with AI software companies, or even with schools, universities and R&D specialists, to continue to expand and develop digital strategies.

For example, back in 2023, Renault Group’s supply chain division announced it had partnered with École des Ponts ParisTech, school of science, engineering and technology, to explore the future of AI optimization in the supply chain. The OEM signed a five-year agreement to strengthen the partnership and accelerate innovation in the supply chain using AI.

While the tools and potential of AI are growing fast, organisational culture and cross-functional integration are the real keys to adoption. That means upskilling the workforce, making data transparent and standardised, and establishing trust – both in the technology and between partners.

AI is no longer the new kid in town, as Linge put it, but it is evolving rapidly. Those who invest in people and processes alongside technology will be best positioned to lead the next phase of intelligent, agile automotive supply chains.