Gasgoo Munich- On September 18, Liu Xinlong, head of automotive AI solutions at Volcengine, took the stage at Gasgoo's 4th AI-Defined Vehicle Forum. In a speech titled "Practical Insights on AI Transformation for Automakers," he detailed how large-model technology is being applied across the industry.
Liu noted that as agent architectures standardize, AI is now ready for practical integration into enterprise production environments. He suggested companies pursue three "compound interest projects"—structuring business operations and codifying execution manuals—to lay the groundwork for future AI value.
On the productivity front, Liu argued that deploying "digital employees" requires four pillars: model capability, secure agent architecture, IM platform context, and the accumulation of AI know-how. Citing Doubao Work as an example, he outlined its use in cross-system operations, user research, data analysis, video creation and replication, travel management, resume screening, and financial reimbursements.
Turning to product capability and consumer experience, Liu shared case studies on intelligent cockpit assistants, 3D avatar generation for desktops, and personalized vehicle wraps. Notably, the cockpit assistant is already collaborating with Tesla to offer features like persona-based interaction, information queries, and linked navigation. Meanwhile, the cost of generating 3D desktop avatars can be kept under one yuan.
Liu also highlighted Volcengine's exploration into embedded development toolchains. The company is working to integrate with various R&D tools, aiming to provide automakers with enhanced support for their development environments.
Volcengine has been deeply entrenched in the automotive sector since 2019 and currently collaborates with most domestic OEMs across a range of dimensions, according to the company.
