Why are carmakers racing to build smarter in-car AI?

It’s easy to forget just how quickly technology in our cars has changed. Not long ago, talking to your vehicle meant having to memorize and use a handful of pre-defined phrases — “Call home,” “Navigate to work”— and hope the system would understand you. But over the past two years, thanks to generative and agentic AI, the world of in-car voice assistants has stepped up a gear and the pace of change shows no sign of slowing down. Carmakers, suppliers and tech companies are jumping in and the technology itself is getting smarter, more capable and easier to use.

Back in January 2024, TomTom introduced its own in-vehicle AI Agent, aiming to make interactions with your car more like natural conversations. Since then, the market has become a hive of activity. Major automakers like Stellantis, VW, Audi, Mercedes and Kia have all launched or announced their own AI-powered voice assistants. China-based carmakers are moving fast too, integrating advanced AI into their vehicles at scale from the outset. The result is a landscape that’s busy, ambitious and fast iterating to develop and find incremental improvements.

Industry analyst Roger Lanctot, who’s been following these developments closely, describes this moment as a “transitional phase.” He points out that, while the technology is advancing, “the uneven execution and variable implementation means that we are still in a transitional phase of understanding the technology and finding the best ways to implement and explain it to consumers.” In other words, industry players are having to balance the uncertainty of experimenting and learning for themselves with confidence and clarity required to educate their customers, from partner businesses to end-users.

Going through these potentially uncomfortable and challenging times is important for carmakers and their suppliers — not only must they extract value from the significant investment being made into AI technologies, there is also a lot at stake.

For automakers, voice assistants are about more than just convenience or gimmicks. They’re becoming a key part of how car brands connect with drivers. Lanctot explains, “In-vehicle assistants will ultimately integrate with dealers, emergency response, CRM systems and smartphone solutions. AI agents will come to define brand engagement with consumers.”

If these systems don’t actually improve the driver’s experience or strengthen the relationship with the brand, he says, “the car brand will essentially be considered a failure.”

There’s also a practical side: smarter assistants can help spot and even fix vehicle problems, they can even directly educate drivers on how to maintain their vehicle and address minor issues such as changing a wiper blade or filling up screen wash, potentially saving automakers money on warranty claims or small servicing queries.

However, as Lanctot says, the industry at present is uneven. Not every carmaker is at the same stage, and as with driver assistance systems, they all offer differing levels of functionality. Some carmakers rely on off-the-shelf voice assistants that we know from smart home technologies. These can respond simply to basic, pre-defined commands. Others, like TomTom, are pushing the envelope to create systems that can handle natural language requests that might come with multiple commands nested within one statement such as, “Turn down the volume of the radio, turn on my heated seat and find me an Italian restaurant on the edge of town to stop at on the way home.”

Some suppliers are connecting these language functions with productivity apps like calendars and email. These allow the car to make suggestions using incoming information or pre-program navigation based on the driver’s calendar appointments. They can also tolerate being interrupted with what Lanctot calls “barge-in commands” — an addition to a previous command that’s made after the previous command has been actioned or is being responded to.

Lanctot says these are some of the cleverest solutions and feature we see on the market right now, but with new developments coming every week, that’s all but certain to change.

With continued development comes the ever-present challenge of Big Tech’s encroachment into our vehicles. As Big Tech embeds itself in our dashboards, it comes with pros and cons. For drivers, it can deliver the powerful features of web services and applications they use every day on their smartphone. For carmakers and companies like TomTom, it makes it harder to deliver to drivers what they want and expect, Lanctot explains.

He says the direction TomTom is taking with products like its Automotive Navigation Application and In-vehicle AI Assistant, is the right bet. “For carmakers and companies like TomTom, the opportunity lies in offering something different [from Big Tech] — navigation and alerts that use real-time vehicle data for example, or other features that are deeply integrated with the car’s own systems.”

Both of the aforementioned products are designed to integrate deeply with the vehicles they’re embedded in. This allows them to operate at a level of depth that other “plug-in” style solutions cannot. In an electric vehicle, the map can read the battery status to plan charging stops and the car can read the map to see how long it is until the next stop to pre-condition the battery for optimal charging.

[TomTom Automotive Navigation Application also allows carmakers to customize the look, feel and branding of the interfaces and map, helping them stand out and bring their brand to life. Read more here: Customize, integrate, deploy in weeks: TomTom’s new in-vehicle navigation application]

For voice in the car, the more deeply it is embedded, the more deeply it can interact with and control the vehicle and the richer the experience it can provide. A regular assistant might only be able to tell you things about your vehicle’s status or respond to basic commands. TomTom’s In-vehicle AI Agent can alter car settings, change the route, interpret context from traffic information, perform functions on other embedded apps such as playing music and much more.

Indeed, according to Lanctot, the most successful in-car AI tech, and by extension IVI systems, are those that feel most like a natural extension of the car. They can control vehicle functions directly, integrate with productivity tools, and — perhaps most importantly — understand the context of the drive and act with that knowledge.

The goal is for the assistant to anticipate needs, provide timely information and make life easier for the driver. The better it can do this, the more it will be used by drivers, and the more they will love driving.

As AI tech continues to proliferate our digital lives and as consumers become more accustomed to it in their smartphones and computers, their expectation to see it in their car is only going to rise. Lanctot predicts that drivers will soon expect their cars to be much smarter and more self-aware than they have been.

“Cars should anticipate driver and passenger needs: doors should lock or unlock at appropriate times, hatches should open automatically, HVAC systems should regulate the in-car environment automatically based on driver/passenger preferences, previous routine trips should be remembered, charging or fueling requirements should be anticipated, contextual and relevant weather and traffic alerts should be displayed for the driver and the car should route to where parking is available in real time,” Lanctot says. This is not an exhaustive list, but an illustration of the kinds of things drivers expect and what might be possible with deeply integrated AI in-vehicle assistants.

In short, the car of the near future will be expected to do more, with less effort from the person behind the wheel. With AI technology embedded, the car will be able to do more.

For TomTom, the focus since launching its AI Agent has been on making voice interactions in the car feel natural and genuinely helpful. The aim is to help drivers complete their journeys safely and easily, whether that’s finding a destination, adjusting the climate or reacting to real-time alerts.

Voice assistants in cars have come a long way in just two years. There’s still work to do, but the progress is real — and the potential is making itself clear. As Lanctot’s insights show, the industry is learning and developing technology quickly, despite wider uncertainties about what will prove successful in the market.

Whatever shape the technology must take, the best solutions will be those that simply make life easier for drivers.