Ford introduced an AI assistant able to report the vehicle's fuel level and tyre pressure, reported in the second week of August.
The choice of scope is what makes the case interesting. It isn't an assistant that drives or decides: it's a conversational interface to data the car already measures and that today requires a menu, a dashboard light or the manual.
This kind of application has a quality rare in AI products: the answer is verifiable. If the assistant says pressure is low, there's a sensor-measured number confirming or contradicting it, which limits the room for invention.
It's the opposite of the pattern dominating the category, where the assistant answers on broad subjects with no identifiable source, and the user has no way to check.
For anyone building products with AI, the example points to a practical and unglamorous direction: starting with what the system already knows precisely, and simply making it accessible in natural language, delivers value without inheriting the reliability problem.