BMW has made a bold statement this week, breaking free from the shackles of Silicon Valley. In launching its own virtual assistant, BMW is confident it can deliver a better experience than Google, Amazon or Microsoft.

From March 2019, the BMW Intelligent Personal Assistant will be powering the automotive experience. Whether its checking traffic conditions, learning your habits to create journeys which incorporate a McDonalds Drive Thru, telling you how certain features work or operating functions through the voice interface, it’s the futuristic dream which has been promised by artificial intelligence.

“Software is one of the main investment areas in our company. Not only for the connected car or digital services, but also in the classic IT, with autonomous driving it becomes very important,” said Dieter May, SVP of Digital Products at BMW

“I think our assistant needs to have the capabilities to operate the car at the end of the day and that is why we believe for safety and integrity reasons we need to manage that customer interface. Overall we have a two pronged approach. For BMW ecosystem related topics and plus the car operations, we have our technology stack, but then we are also able to dispatch depending on the type of questions to third party assistants.”

In terms of what the virtual assistant actually is, it’s nothing out of the ordinary. Of course, a functional virtual assistant is certainly something to be applauded, but BMW don’t seem to be doing anything which is already on the market aside from being able to rename the assistant whatever you want. It is specific to the car environment, but we suspect it wouldn’t take long for the likes of Google or Amazon to figure this out. So why break away from the Silicon Valley experts?

As with any new software breakthrough, the manufacturers try to create their own proprietary offering. There is no problem trying to own the customer experience, they are customers of that business after all, and it does offer an opportunity to differentiate. In this case, it is also part of the Infotainment package being sold to customers as well. However, in most cases, the versions created by the OEMs is sub-standard to the version created by the experts, and we suspect it will be the case here as well.

The beauty and downfall of virtual assistants is the machine learning and personalisation opportunity. The beauty is the constant improvements in the accuracy and effectiveness of the software, though the downfall is exactly the same. If you don’t use it enough, there isn’t enough data to train the algorithm, and the assistant falls short of the promised personalisation leading to poor customer experience. This is where BMW could be kicking off a dangerous trend in attempting to create its own virtual assistant.

Should others follow suit and attempt to own the assistant, the space will become incredibly fragmented. Your smartphone could become littered with different assistants for different parts of your life. Not only would this be incredibly irritating and confusing, but it would impact the performance of the assistants themselves. As mentioned before, virtual assistants are only as good as the hours they are used. If you are splitting your time over several different assistants, it reduces the amount of data each collects and can be used to train the algorithm. The result could be several so-so or poor assistants, offering more generalised services, instead of a single virtual assistant which collects all information about the user and is incredibly personalised.

Both consequences would be very negative for the development of virtual assistants. If there are too many hyper-specialised assistants, they become irritating so people will stop using them. And if the training data is spread too thin, the personalisation promise is never realised, with users probably ignoring the generalist services which never quite hit the mark.

The BMW virtual assistant might well be a good bit of kit, though we suspect the software engineers at the firm are not quite in the same league as counterparts in Amazon, Google or Microsoft. But whether it is any good now is not the point. Over-saturating the virtual assistant segment significantly lowers the glass ceiling of personalisation and will damage the long-term potential of the technology.