Volkswagen is moving generative AI deeper into vehicle engineering. The plan aims to shorten development cycles and reduce repetitive work. It also raises serious questions about control, safety, and accountability.
For telecom engineers, the story feels familiar. Modern cars now behave like rolling software platforms. They depend on sensors, connectivity, cloud services, and regular software updates. That makes automotive development closer to network and software operations than ever before.
Werner Tietz, Head of Group Research and Development at Volkswagen Group, stated: “At Volkswagen, AI is not an add-on or a standalone project – it is an integral part of how we work.”
The group says it now runs more than 1,200 active AI applications. These cover engineering, production, supply chains, cybersecurity, simulation, and internal knowledge sharing. In Technical Development alone, over 100 AI-based processes entered live use during 2025.
The target is ambitious. Volkswagen wants to reduce vehicle development time to under 36 months. That would cut around a quarter from its current baseline. However, the company has not yet published enough data proving that target is achieved.
One practical example is GHOST, Volkswagen’s internal testing tool. It tests infotainment software, including touch interactions. In simple terms, it can press buttons like a human tester. Volkswagen says this improves repeatability, reduces documentation errors, and supports faster releases.
This matters because testing often slows engineering teams. Skilled staff spend valuable time repeating checks and preparing records. AI can remove some of that burden. Engineers can then focus on unusual faults and product improvements.
Meanwhile, Volkswagen is also working with Microsoft and PTC. The project brings Microsoft Copilot capabilities into Codebeamer, PTC’s lifecycle management platform. The platform helps teams manage requirements, tests, and validation records.
Robert Kattner, Head of Volkswagen Group IT Engineering, said: “By having a copilot in the Codebeamer software, it can assist with creating new requirement specifications and test cases using our specific data and business context.”
That approach points to a wider industrial shift. AI is no longer just writing summaries or searching documents. It is entering controlled engineering systems. These systems shape products with long service lives and strict safety demands.
Yet the risks remain clear. AI can produce convincing but incomplete outputs. It may miss a safety condition or repeat flawed source data. Therefore, human review, version history, and traceability stay essential.
Tietz later added: “We often refer to ‘AI-accelerated engineers’ – professionals who see AI not as a substitute, but as an amplifier of their capabilities.”
That message may resonate across telecom, manufacturing, and software-defined infrastructure. Automation can speed work, but it cannot own responsibility. Engineers still need to validate final decisions.
Volkswagen’s plan shows how enterprise AI may mature. The strongest value comes from focused workflow improvements, not broad promises. If the company proves faster delivery with maintained quality, other industries will watch closely.
