Postcard from Auto Shanghai: Intel, ModelBest Deliver a Personalized LLM

To: Everyone who drives

From: Auto Shanghai 2025, Shanghai, China | April 23, 2025

What if your car could understand the complexities of human language and the behaviors to complete full requests while you are out and about in a way that’s personal and secure? An example: You’re driving on your daily errands and you tell the car to text your friend tonight’s movie reviews and times. To do this, the vehicle goes into a movie fan app; gathers the links; goes to the texting app, pulls up the friend’s contact and sends all the details.

Here at the Auto Shanghai event, ModelBest shows off what you need to do just that. Its GUI Intelligent Agent brings to life a true in-the-vehicle large language model (LLM). Powered by Intel’s software-defined vehicle (SDV) SoC and Intel® Arc™ graphics for automotive, it enables offline, AI-enhanced voice controls and personalized interactions.

It works by accurately understanding the way we speak naturally, what we are asking for, and completes the task – without the need for fiddling with complicated buttons while driving.

The agent also runs locally in the car – without online connectivity to the cloud (unless it’s needed for the movie app) – saving precious data, reducing round-trip latency, and ensuring your privacy and security concerns are met.

P.S. Intel Accelerates Software-Defined Vehicles with Next-Gen SoC and Partners | Intel at Auto Shanghai 2025 (Press Kit)