The Big Turbo

Cars, EVs and motorsport news

Latest / Tech

Tech

Talking Cars Get Proactive as Agentic AI Moves Onto the Dashboard

Autocar reports that in-car assistants are moving toward agentic AI, leaving European carmakers to ask whether drivers want proactive cockpit technology.

According to Autocar, in-car voice assistants have become far more capable, with Mini's 'Spike' cited as an example of how far the technology has come. The outlet reports that the next step is so-called agentic AI, systems that do more than answer questions and can carry out tasks on a driver's behalf.

Autocar says China is embracing what it calls "AI-defined cars," while European carmakers are weighing whether their customers want the same kind of proactive cockpit technology. The report uses an example of a driver asking a Mercedes system to quiz a child before a French exam, illustrating the type of multi-step request now being discussed in the industry.

The outlet links this shift to software-defined vehicles, pointing to the Volvo EX60 as an early model. These cars use fast, centralised computer chips that oversee almost all onboard functions, which Autocar says has changed the situation for car makers trying to add new digital features.

Autocar frames the central question as one of customer demand. It asks whether buyers in Europe will follow the lead set in China and on the American West Coast, where similar tools are already gaining ground. How the features would work in practice, and which drivers would use them, is not established in the reporting.

Background

Voice control in cars is not new. For years, drivers have used simple commands to make calls, change radio stations or set navigation destinations. These early systems followed a fixed pattern: a driver spoke a phrase, and the car matched it to a preset function. Most gave limited answers and often struggled with natural speech.

Recent advances in large language models have changed expectations. Assistants can now handle loose, conversational requests and follow up on them. Agentic AI goes a step further. Instead of only responding, such a system can plan a sequence of actions, such as searching for information, using a connected service, or completing a booking, usually with the user's permission at key points.

The car itself has also changed. Older vehicles relied on dozens of small control units, each handling one job. Software-defined vehicles consolidate many of these functions onto a smaller number of powerful central computers. This design allows features to be updated over the air and makes it easier to add services that depend on software rather than new hardware.

Regulators and manufacturers have long been concerned about driver distraction, and rules on screens and interaction differ between regions. Privacy is another consideration, since assistants that connect to personal accounts and calendars handle sensitive information. Any move toward proactive tools will sit within these existing expectations about safety, data and driver attention.

Why it matters

The shift from reactive to proactive assistants could change how drivers relate to their cars. A system that anticipates needs, rather than waiting for instructions, may feel more useful, but it also raises questions about how much control a driver keeps and how clearly the car explains what it is doing on their behalf.

For carmakers, the stakes are commercial as well as technical. Software features can become sources of ongoing revenue and differentiation, but only if customers find them worth using. Autocar's framing suggests that European manufacturers are cautious, partly because buyers in the region may have different priorities around privacy, convenience and driving enjoyment.

The reporting also highlights a split across markets. If China and parts of the United States move quickly while Europe waits, the resulting differences in features could influence which brands feel modern to buyers and which feel behind. That gap may matter as much as any single product launch.

What to watch

Watch whether European carmakers announce specific proactive features and how they describe the limits of those tools. Clear explanations of what the assistant can do without asking, and what it must confirm first, will be important to buyers and regulators alike.

Pay attention to how software-defined platforms such as the Volvo EX60 evolve in use, especially updates that add new assistant abilities after purchase. Consumer reaction, safety assessments and privacy disclosures will show whether the technology is becoming a standard expectation or remaining a niche feature.

Viewpoints

Commentary from our columns: three angles on this story.

The Skeptic open questions and weak points

It is easy to be dazzled by a car that can plan your evening. The harder question is who the car is working for. When an assistant books a table or makes a reservation, it is acting on a driver's behalf, and that raises practical issues: what it can spend, what it can share, and whether the driver is always told before it acts. Autocar's example of quizzing a child before an exam sounds harmless, but it also shows how many personal details a proactive system would need to touch.

There is also the matter of trust. Voice systems have a long record of misunderstanding simple requests. Asking one to handle a multi-step task introduces more places for things to go wrong, and a mistake in a booking is far less forgiving than a wrong radio station. Carmakers should be asked plainly how they handle errors, how they store data, and how a driver switches the features off. Until those answers are clear, enthusiasm for proactive cockpits should come with a healthy pinch of caution.

The Engineer how the hardware works

The interesting part of this story is not the chatbot. It is the architecture underneath. A centralised computer that runs most of a car's functions has to keep the assistant, entertainment and driver-facing systems separated from the parts that steer, brake and monitor the powertrain. That separation is the real engineering challenge. Critical functions need guaranteed response times and must keep working even if a software layer fails, so the safety-critical code should never depend on an AI service that needs a network connection.

Agentic tools also rely on cloud processing for much of their reasoning, which means latency and connectivity matter. A booking that stalls in a tunnel is an annoyance; a navigation prompt that stalls is a problem. Good designs will keep local fallbacks for essential tasks and treat cloud-based intelligence as an optional extra. Whether carmakers get that balance right will say more about their engineering maturity than any demo of a clever voice command.

The Commuter the everyday buyer's view

For most people buying a family car, the question is simpler: will this make my week easier or just add another thing to ignore? A system that can reschedule an appointment or find a place to park while I am already on the road sounds appealing, especially on a busy commute. But I would want it to speak up before doing anything that costs money or changes my plans, and I would want an obvious way to say no without digging through menus.

I also think about the people sharing the car. Children, partners and visiting friends all talk to the same dashboard. A system that acts proactively needs to know whose request it is responding to. Autocar's question about whether European drivers want this at all is a fair one. My guess is that buyers will be open to features that save time, but they will reject anything that feels intrusive or hard to switch off.

More in Tech