Automotive AI expands expectations for the smart device - by BYTON | Disrupt SF (Day 3)
7:50PM Sep 10, 2018
Hello, how's everyone doing today. So my name is Abe Chen, I'm the head of our VP of digital technology at BYTON so what we see here is our K byte. And I think downstairs in the expo hall, if you haven't already seeing already, we have our m byte. So we launched our m byte the car downstairs, originally at CES this year, and we gave approximately a little bit over 100 people rides in this amazing car. So what I want to do today is to talk to you and tell you a little bit about our story and how AI really became a major part of her company's DNA. Now, traditionally, are when, when the company was started, we it was mostly made up of automotive experts from BMW, Nissan, infinity, and many of the other car companies from around the world. Now, as we all know, car companies are not always know for innovation. And so what they were as a company from the decision was made to move engineering into Silicon Valley. And that's kind of where myself and my team kind of came and became part of the bite and family. Okay,
older years, there's been many, many innovators and they're kind of tread the path to take us to where we are now today. So electrification, right, really? Well, our goal, when we electrify cars, as we worked in the vehicle industry, going from technology into the electric car industry or into the vehicle industry, what we were trying to do is really to change the world. I think many of us here, we are innovators, we're here at TechCrunch today, to really to focus on how do we enrich the lives of everyone, our fellow human beings, as well as to figure out how we can push technology forward at some point in time, we realize that with electric platform and bring together the people that
inspire to build these powerful, you know, these fast, powerful accelerating cars, we can actually
we have a very good point platform to make the car smarter, we added the ability to do over the air updates, which opens up a whole new windows and capabilities. And then today, as we know, especially the demonstration on k by our design elements, we integrated this autonomous capability with our partnership with a roar. Okay, so now the artificial intelligence story, okay, how do we end up looking at artificial intelligence as part of the automotive companies DNA.
what happened next is that we had all these concepts and
ideas as a organization or team said, Okay, one a car that would actually create this sense of transparency from end to end, where when the customer gets in our car, where they take, you know, they can transfer from their phone, their, what they do on their phone, into the car, but how great would it be, if you get in the car, then you don't have to always take out your phone, put on the dash, or, or if, let's say, for example, if
you're, you need to make a change, you don't have to reach for your phone and make that change, or rely on programs like Siri and so forth to me, you know, to, to change the settings and, or, or look for information, one of the car was intuitive
enough to know it's time to go get some milk, or it's time to go charge. Or perhaps, let's say
you're you go to the gym every day at 6pm. And
just so happens as there's a charger block
away that you didn't know that was there, one of the car can tell you, hey, you can actually save some time the day if you were to go park your car a block away before you went to the gym. Okay, so all these questions came to be. And we realized somewhere along line that
what potentially we were
looking at is to build this this portfolio of capabilities.
And many of these capabilities are available with, you know, that we can acquire from third party, many of these abilities are pleased that we could potentially build in house or there are things that we've done by ourselves
before in the past.
And so we can't have this concept of this ecosystem that was really API driven, okay, which look very different from automotive company.
And this API driven capability within our
Biden smart or hybrid cloud allowed us to integrate and exposed to third party partners, potentially,
some of you in this audience today, as well as, as well as push and pull data
from within our system and our partner systems.
So we came with this problem of like, Okay, so how do we get all this data together? Because we are a startup, like many of us here today, we didn't have that legacy. I'm sure many of you who worked with data before have this problem of data, old databases here, databases there, but how do we get all the data to get in one place? Okay,
so sounds simple
on paper to say, okay, we're going to create this
centralized data, like, we're going
to put all the data into one storage location, we're going to go to different business units, and say, Hey, you need to surrender your data for the greater big of greater of the company. But the reality is that there are many barriers that keep
companies from be able to achieve this, for example, perhaps the other business
unit who, for whatever reason,
does not get along another business unit. So you have politics and companies, right? So what happens is, there's a resistance
to move that data or free that data, or just give up that data to a centralized group. And so you spend perhaps, maybe two weeks at least three
weeks, six months trying
to negotiate to get that data moved over into this big, powerful data lake. And so what we did is that we made a decision earlier on that
we were going to not just
we're not going to have those issues, we're going to
move aside, push aside those innovation
killers and bring together ourselves as one company and one group inside the organization that we call digital technology. So what is this, what do we do so we knew that oftentimes,
team or the hardware team inside the car would argue with the data scientists to to just you know, as to whether what kind of data can actually be sent from the car, we knew that our product security teams are security teams are very worrisome about the amount of data or the type of data that we collect from our
customers and where that data is kept.
We also need our cloud teams oftentimes had issues working with from past experience, had issues with working with the engineers that were working on the car as to
how the car connects, right? So we had all these challenges
that we knew that they were engineer, there are harbored
challenges, for example, the ability
to send data out of the car, okay, so what we ended up doing was we took the organizations that produced the, what we call it bite and smart gateway, see heart and brain of our car, we took the organizations that built the seller telling Matt capabilities, we took the organizations that built the antennas on the car, the cloud products, security, and we put them together into one organization. And we also integrated it.
So now there's, there's no excuse your one group, there's no politics. And so really the many ways that's actually our secret sauce, because
we knew that we had to be able to achieve and deliver our cars by end of,
and as a company, we've only been around for two years.
So which was nearly impossible feat. So what we
did was we
pump from the advice of
our product security experts,
we decided that early on that we're going to keep data within the region that it's created. So if you're in the US, your data stays in the US, if you're in Europe, your data stays in Europe, if you're in China, your data stays in China. Now, we knew that for data for machine learning to work, we need lots of data, right? So of course,
we work with partners like yourselves
to, to have some of this those data. And we also knew that we had to make sure that we meet the data privacy laws of these different
regions. So the election was to go for the most strict data privacy laws. So we integrated
GDPR as part of our model.
So what does that mean? Okay, so it's easy to say, Okay, I'm going to collect a lot of data from customers. But that's, it's that's to simplify, you collect the wrong data, the customer doesn't know you're collecting
the data, it can potentially create all sorts of legal ramifications downstream.
we collect data on customers, we consciously acknowledge, what we're going to do is to let the customer know that we're collecting data will let the customer know what that data is being used for. And then we'll also give the customer the right regardless of their region to be
forgotten. So, which is
in many ways, an innovative approach to data collection.
So what what happened next, when we
went down this road and decide, okay, all data has been collected, these groups are going to all work together, they're not going to be separating more, it's one team,
suddenly, we were actually realized that, hey,
we have a lot of data now, right? We have a, we have sensors, we have the ability to collect data, so
we could collect car data from the car the same to team that built the app
was the same team that builds the collection capabilities inside the car. And it's the same team that builds the antenna for the car, what a concept, right, and so also the same team that manages the cloud.
And so through this process, we realized that there were several
API's that we could develop
to advance our capabilities and allow us to integrate with our partners. And so, you know, here's just a list of them. But overall, at the end, what it does is that it allows us to power what we call our shared experience display,
which if you have time, please check it out, down on the the expo hall.
And so what the final end result is due to to realize this innovation or this vision that we had of sales, and after sales, security, and personalization, so from a sales and after sales capability, or service capability, where we really created as well lifecycle where the customer before they become official customer can download our app, we can, they can provide us with information, what their preferences are right, we can start with their acknowledgments are collecting information on
their restaurant preferences, the locations that they travel
to the their path
of travel, and so forth, that powers this capability,
where, let's say, for example,
on the aggressive driver, okay, and I break a lot
like, I drive like, you know, racing every single day. And on the other end, the car also knows what the or the lifespan
of a brake pad is, right.
So we know what the lifespan and brake pad is, the car knows how you drive
your driving style, right, that's to this multiple data points that allow us to predict or know when the car needs to be serviced when those brake pads need to be replaced. And that's actually from the for the automotive industry, that's actually very powerful. Because this way allows and also allows us to create a better customer experience where when it's time to go get your car serviced, we
can tell you in advance because of
your driving style, your break past when you replace within the next thousand miles versus this, nor with the current norm, which is
every 30,000 miles every 15,000
miles, you need to replace your tires or your brake pads, right. Or it's the same concept. So now this also kicks off because we're one
organization again, this the organization that controls the end and digital experience when our innovation
this allows us to empowers us to have this idea of where
when you have a
car that needs to be service inside the car you can display to the customer that is time to service, the customer inside the car can
select where they want to or when they want
to service we can make recommendations of where the cars should be serviced. And we also can tell downstream ourselves or our
fulfillment procurement capabilities
systems or CRM systems as well as our era p systems or brake pads and make sure that the brake pads are at the service center before the customer rights. Right. So now, this is something that this does not currently exists today. It does, there's no this capability doesn't exist today.
So how did this like how the AI come down
our vehicle initially, and I think
because passion, my background is security research, right. So this is actually another story of our company or
organization is that that
security by design is that our core most companies,
what happens is the the engineer and product engineering comes up with some ideas and some product r&d that comes becomes something and at some point in time security looks at it. And then it's too late product has to ship sure many of you I encountered on your career.
So you ship this product that is not optimal from the security perspective
works. But as you know, at some
point in time, someone's going to find a problem in the fix it. So what we did to kind of deal with that issue is what we went we move forward. And we made it so that the organization the security group, for our band security lab is the team that actually does the innovation, they come up with the security concepts. Now, security lab wanted to be able to do machine
learning inside the car. So they
dictated and they push for a higher compute, because we we did that it also allowed us to do what I
explained earlier, this whole concept of a smart diagnostics, right, smart
diagnostic, being able to detect
or predict when your car needs to be serviced using the same machine learning a similar machine learning algorithm that we use for detection of attacks, okay. And that same concept allows us to push and leverage personalization
side of vehicle to know
what your preferences are. So this is an example of where our
innovation and security ashy innovate allowed us in innovate in other areas.
So of course, the end result firm is really to power this large 50 inch LCD display. And because of this large display,
we had to be able to push a lot of data
to the car and that also inspired us are innovative to innovate in the space of what we call our bond smart gateway.
And the smart gateway is
is actually that's production we're
looking at well on course to 10 gigabits per second now, 10 gigabits per second might not seem like a lot. We talk about data centers that have an infinite amount of power, but inside electric vehicle, heat and power is your enemy. So 10 gigabit is unheard of today the fastest gateways on the market for that you can purchase for cars is around 100 megabit per second one gigabit gateways are just about to come out. But because of you know, our drive and our inspirations
or are we
are now in the phase of building our 10 gigabit gateway
and the overall
ultimately what we're Our goal is to really create that better trying to experience
our approach to develop
them cars is really thinking about what does a car in 2020 look like?
And so this is really our story as to how AI
and machine learning has become part of our DNA.
Thank you very much.