Gasgoo Munich- "We've been married all these years, and you still don't understand me—why should I believe a car can understand me any better?"
On September 17, at the fourth AI-Defined Vehicle Forum in 2026, Ren Yi opened with a rhetorical question. Ren Yi is the product director at RIVOTEK Technology Co., Ltd. To him, this captures the current reality of AI cockpits. AI is already close, but a cockpit that truly "understands you" remains distant.
Ren Yi stated that RIVOTEK is transitioning from a traditional integrated Tier 1 supplier to an AI cockpit solution provider. The focus is not on the number of large model interfaces. It involves layering AI capabilities onto existing mass-production systems. This builds skills that are executable, integrable, and ready for mass production.
Currently, RIVOTEK has established an AI intelligence base centered on the Zhixin Domestic Computing Platform, Zhiwei AIOS, and Zhirun Edge-side Model. Public data shows mass-produced smart cockpit installations have surpassed 1 million units. Adaptation is complete for mainstream domestic chips like UNISOC, AutoChips, and Horizon Robotics.
During his speech, Ren Yi revealed data from an AI Coding trial. A five-person team, supported by an AI server, completed Agent OS-related engineering within two months. The codebase grew from roughly 35,000 lines to 140,000 lines. Test files were built from scratch to about 150. The system now covers 150 in-car scenarios.
But he emphasized that "AI Coding isn't about letting AI write freely; it is an engineering system." RIVOTEK deploys different AIs for development, auditing, and cross-validation, establishing a tiered review mechanism. Modifications involving high-risk architecture, customer value, or compliance still require final human review to ensure accountability is traceable.
Cost is also an unavoidable hurdle in engineering AI cockpits. Ren Yi explained that high-frequency applications should be handled by edge-side models as much as possible, while complex, context-heavy scenarios should trigger cloud capabilities. RIVOTEK has set an internal red line. The cost increase introduced by AI must be kept within 10% of the total product cost.
Meanwhile, AI is also penetrating the company's R&D and operations systems. According to Ren Yi, AI has driven a 60% efficiency boost in standardized tasks. These include documentation, code commits, and test cases. Meanwhile, non-R&D labor costs have dropped by roughly 40%.
"AI cockpits must retain Agent capabilities. If current cockpits lack this, they forfeit their entry ticket to the next round of competition," Ren Yi stated. RIVOTEK ultimately aims to upgrade intelligent cockpits from functional control systems into genuine in-vehicle intelligent systems.
