SPEAKER 1: All right,
good afternoon, everybody. Well, I wasn't
supposed to come here, but my team asked me
to come and be a cameo. And I'm trying to be very short
and brief about opening up this spot, like sessions. This morning, I mean, you
guys were in the keynote, and we had three new IoT
product announcements. And it actually shows insight
on what Google is really actually doing and the
vision, or strategy, that they were taking. The interest we are seeing
in the field, that a lot of the customers having
the IoT, the problems, and the applications. And we see this
is an opportunity, because its actually that a lot
of data are being generated. And Google is, essentially,
a data company, and our mission is
to organize data. And that we see as an area
where our core strengths, in AI and big data, and then
also a lot of the analytics solutions. And those things can
actually be a big instrument for a lot of our
customers, and also providing the access
to organize this data, and access to a lot of them. Our key strategy, as you
probably actually guessed, is really AI first, and
bringing AI as really front and center of everything. And so the first
announcement that we had was about the [INAUDIBLE] TPU. It's a custom
hardware accelerator for running
TensorFlow Lite model. That's actually really
optimized the performance per dollar and performance
per power consumption. It really enables a
lot of applications, as you can imagine. When this TPU got combined
with the microcontroller that actually makes, running
some processing, and doing input and
output, and combining with a security
chip that actually provided a little trust,
and then on the encryptions. And that actually really can
actually be very disruptive in the market, creating
new opportunities, like making things in the
world really intelligent and providing brain
at a very low cost, and with very high
power efficiency, without compromising
the performance. And that's really a small
peek into what we are doing. But actually, we are
building, really, the core end-to-end story of the
cloud infrastructure for IoT, and then also the [INAUDIBLE]. And we're doing this
not just by ourselves. We're actually in partnership
with many partners. And there's certain areas
that Google can do a good job, because we know our
algorithms and we know how to optimize for them. And then we also make
those solutions actually available over to our
partners for them actually to make a product out of it. We also recommend our partners
to our customers, and vise versa. So actually, we thrive on this
ecosystem with the partners, as well as our own
technology innovations, and really pulling our
own weight as Google. So today, we have a
really beautiful session and a lot of good information,
more detailed information, about Google strategy, about
IoT, intelligence, security. And we have a good
set of customers going to speak at the panel. And so I'm very excited
to have all of you. And without much due delay,
I'm going to open it up to the next speaker. Thank you very much. [APPLAUSE] ANTONY PASSEMARD:
Thank you, [? Anton. ?] So last session of the day,
so thank you for being here. How excited are we about IoT? There were 17 sessions,
yeah, 17 sessions on IoT, the big keynote announcement,
a lot of good news. I'm really excited
to be here and talk about our vision for IoT. I'm Antony Passemard. I head the product management
team for Cloud IoT offering, and really excited to be here. So we all know we're living
in a connected world. There's really no surprise,
now, that IoT's real. What's really changed
is that I could talk about the
billions of devices that are going to be
connected, but the reality is we are moving from
connectivity to intelligence. Because without
real intelligence, without real insights,
or actionable insights, on the data that's
generated by those devices, it's really hard to get a
real return on investment. So intelligence is key
to those investments. And the problem is those
devices are generating a huge amount of data. And when we're looking
at how much data that is-- an [? ITC ?]
study said that in 2025, there's going to be about
40 billion zettabyte of data generated by IoT
devices, and most of it is actually in real time. And 40 zettabye is about, what,
40 billion petabyte of data, if I'm not mistaken. That's a fairly
sizable amount of data. But the problem is
with that much data, how do you get
insight of the data? How do you analyze that
data to really make meaningful decisions? All that data and
all that insight is really about
connecting devices with application and
people in a seamless way, in a nondisruptive way. It's really that data that
is the glue between all those pieces of the puzzle that
we're trying to solve here. Oops. I'm going the wrong way. So Google has been in the
IoT for quite some time, especially working
with intelligence on devices, so intelligence in
the thermostat, intelligence in the car, intelligence in
how you interact with your home devices, with a home gateway. And we've really learned
from all those experiences, internal experiences,
and brought last year a new platform called
IoT Core, Cloud IoT Core. Cloud IoT Core is
a managed service. It's available in
your console today, if you go to your GCP console. It is a managed service. You don't have to configure it. It scales up and down for you. It will allow you to connect
millions of devices securely to the cloud very
easily, from little POC of a few devices to millions
of devices, with no issues. We're really trying
to productize IoT Core the same way, bring
that innovation to the public, to our customers, the same
way we did with big data tools in the past. Like you've seen
Dremel, Hadoop, HBase came from internal projects
that came out and were delivered to our customers. So a little bit of that same
philosophy here for IoT. We've seen great adoption
since our launch. I mean, some great customers. You'll hear about Smart Parking. Aker BP, as well,
will be on stage. So we've seen that traction of
customers, really transforming their business, with IoT, really
finding new business model, finding efficiencies, finding
optimization everywhere they can. So it's all about
operational insights. It's all about
business processes and managing their assets,
transforming their business with IoT. And it's talking
to those customers. We're really looking at
how we are going forward. So talking about
customers, I would like to bring on stage
Aker BP and Kjartan from Aker BP, Kjartan Neese,
who's been using the platform. So Aker BP is an oil and gas
company based out in Norway. And they've been using some of
the platform, the GCP platform, through a partner
called Cognite. And they're doing predictive
maintenance, and all of that. But Kjartan, tell us
more about Aker BP. KJARTAN NEESE:
Thank, you, Antony. Aker BP is upstream oil and
gas company based in Norway. We are the fourth
biggest company listed on the Norwegian
stock exchange, so big in the Norwegian scape. Just short of 2,000
employees, and have a daily production of a
bit more than 150,000 barrels of oil. The Aker BP, as
we know it today, is the result of a merger
of a small Norwegian company called Det Norske, with
a very dynamic culture, and also the big enterprise the
BP side, BP assets in Norway. ANTONY PASSEMARD: It's a
pretty impressive growth for a year and half company. KJARTAN NEESE: Yeah, quite so. ANTONY PASSEMARD:
So what are some of the key challenges
that Aker BP's facing, and how is IoT helping
with those challenges? KJARTAN NEESE: I
think Aker BP has the same challenge as a lot
of the upstream oil and gas companies. On the top, we are an
industry that's quite heavy, and we have the potential to
have quite a bit of impact on the environment. So I would say A to Z,
really safety and environment is really always on top, so
that's something that we really like to use digital for. And of course, profitability. That's driving down costs
and increase production. ANTONY PASSEMARD:
That's important. So how is it helping? How is IoT helping with those? KJARTAN NEESE: I think there
are several different areas. I think that, actually,
one of the things that we see quite good results from
is actually very simple. It's just based on the
work that Cognite is doing. They are extracting
a lot of data. We are pulling up to
a million data points a second into our Google Store. And based on these data, when
we get them contextualized in the right way and push
it back to the operators out on the platforms,
that really helps driving better decisions,
not really being very advanced. Of course, also, predictive
maintenance, predicting events, is very important. I think, also, these data
really drives opportunities within moving people
out of dangerous zones, and also moving over to
an unmanned platform. That's something we are
leading on also in Norway. ANTONY PASSEMARD: That's great. And anything else you'd like
to add about your usage of GCP Platform, anything? KJARTAN NEESE: Yeah. I think we are really impressed
of the speed of the GCP Platform. And I think, also,
as an industry, we are, unfortunately,
a bit slow movers. And I think also the work that
we are doing with Google really helps us moving faster. I think that's really good. ANTONY PASSEMARD: Great. Kjartan, thank you very much. Thank you for your
trust in the Platform. Looking forward to innovate. So fantastic use case with
predictive maintenance, everything. We really love to
work with Aker BP. So as we're looking now to
the future and a little bit of our vision around IoT,
really looking at IoT through four main pillars. One is around
intelligence everywhere. You've seen this morning, we
announced our edge strategy. This is really about
expanding all the AI and ML capabilities down closer
to where it's needed. Serverless, scalability. I talked a little bit
about Cloud IoT Core, and how Cloud IoT Core is
serverless and scalable. We're really trying to make
that across the entire value chain of our offering. Security, always number
one priority for us. And security in IT is
actually complicated. There's a lot of pieces
that we have to deal with. It's from the manufacturing
process to the boot of a device to the communication to
the backend services, and we'll talk about that. And obviously, our
ecosystem of partners. IoT is not a game
you play alone. You have to play
with partners and you have to build an ecosystem in
every place of the value chain, as well. So we really value building a
big ecosystem for our partners and for our customers. So let's dive a little bit
into intelligence everywhere, and particularly, where
we go with the edge. So Google has been known,
and is known today, for a lot of the ML
and AI innovation and the tools that
we're providing. In this conference, you've
heard about BigQuery ML. You've heard about
oil ML coming out. You've heard probably
about Cloud ML Engine. I mean, all those
tools that we're building to really
democratize AI and democratize ML for our customers,
really trying to bring all that power to our customers. And this is particularly true in
image recognition, for example, with auto ML vision. That's a very
impressive tool that you can get to use very
quickly without being, really, a data scientist. So we've been glad,
too, this morning, actually, to introduce
our Cloud IoT Edge. There is actually another
session tomorrow morning at 9:30 that drills
into IoT Edge. This is really listening
to our customers, saying, I love your AI, but I have
some constraints at the edge, that I can't really use your
cloud platform to do that. Sometimes my internet
connectivity is down. In the case of Aker
BP, sometimes it's boats that are going
that can't be connected. Sometimes it's because
the bandwidth constraint. You heard about LG this morning. They're taking pictures,
hundreds of pictures all the time. They can't send
all those pictures to the cloud to do the
inference and come back. They have to do it on-site. Sometimes it's compliance. You have to get the data. You have to leave
the data on-site and you can't send it
outside of your premise. And there's a lot of reasons
why Edge makes total sense, and we're really trying to
bridge between Cloud and Edge together. So Cloud IoT Edge is
actually two components. One is our Edge ML
component, which is a runtime to run
TensorFlow Lite model. And to get a
TensorFlow Lite model, you take a TensorFlow model
that you train, maybe with an ML Engine, or with Cloud
TPU, using the cloud, and we'll compile
that model for you into a TensorFlow Lite model
and send it to the device. How do we send it? We have Edge IoT
Core, which manages the security, the
connectivity with the cloud, or the local processing,
local storage. Anything you need,
you would expect as an Edge device, that's
handled by Edge IoT Core. This is in alpha right now. There's a [INAUDIBLE]
access program that you can sign up for
today to get access to this. This runtime can run on
Android Things, which is a fully secure and managed OS. So that's a full OS,
managed the same way we manage an Android
device or phone, except it's for Things,
so it's been optimized. And we're also going to
support Linux OS as well. And the big announcement was
the hardware accelerator, so Edge ML will
support, obviously, standard CPU, general purpose
CPU, or GPU, or our Edge TPU. So Edge TPU was probably
the most exciting thing. I don't know if you followed
the news a little bit. There were so many
tweets, I was like, whoa, this is actually
pretty impressive. So people tend to like it. So you're seeing a little
bit of the size here. You can fit four or five of
them on a coin, very small. Edge TPU has been really
worked in collaboration with the AI research team, our
software team, and the hardware team together to really bring
a very purpose-built hardware for inference of
machine learning models. So it's really
that collaboration between AI research
and the hardware that make it possible, with a
very big focus on performance per watt and
performance per dollar. So trying to make
it cost-effective so it can be
everywhere, and trying to make it very low power so it
doesn't consume a lot of power. You can put it on
smaller devices. You don't have to have big
heat sink and things like that. So we really believe
that this Edge TPU will transform just
dumb data collectors into smart collectors
and smart devices, and open up a whole range
of new user applications. So this is really,
really exciting. So the goal of Cloud IoT
Edge, if I summarize, it's to enhance personal
reliability, making sure it can work any time. You don't have to worry
about loss of connectivity. With IoT Core, you can
really deploy those models to the cloud, and to the
edge, sorry, and apply them where you need them most in
real time, very low latency. We always focus on
security, obviously, for the secure connectivity
with Edge IoT Core. And keeping it
cost-effective at any scale, because the more
cost-effective, the more they're going to be out there. We also thought about how do you
marry Edge and Cloud in a very closed loop of learning? And that was very
important to us. We didn't just release
an Edge product. We wanted to make sure
that Edge product is an extension of your Cloud
in a closed loop learning. So this is a visualization of
the data flowing into IoT Edge. You can do some learning, or
you can do some inference here, if you want. But you can send the
data out through IoT Core to the cloud, put it in a
database of your choice, or data repository
of your choice. Retrain your model with
Cloud TPUs, potentially. Then send that model
back to that device through Edge IoT, IoT Edge,
to have a local execution. So that closed loop is very
important because those devices are not static anymore. It's not just you send an
ML model and let it be. You will improve it over time. Things can change. New problems can be
detected with images, and you need to be able
to really effectively move those models,
update those models, through this life cycle. So we've seen LG this
morning in manufacturing. Very great use case. This is a million-dollar
saving per line. That's pretty impressive, just
by putting that intelligence at the edge. In retail, you can think about
hyperpersonalization of offers. This is an augmented reality
view, where discounts could be given to you based on who you
are and your purchasing habits. In automotive,
collision avoidance, detecting if somebody is
looking at their phone. That would be pretty neat. That would avoid a
lot of accidents. There's a lot of
traffic crowding. There's a lot of
application for ML. So those are just a few examples
where we've seen traction with some of our customers. So if I show you,
if we showed you, the little TPU, if I
give you that little TPU today, you're going to
be a little bit in trouble. You won't know
what to do with it. So we created a little
SOM that I have here. That SOM is a System On Module. So it's a full module which
has a quad core CPU on it, has a Wi-Fi, has, obviously,
a root of trust and crypto from microchip. And it has the Edge
TPU, obviously. Don't forget this one. So this is a SOM that's
production grade, so you can actually produce a lot of them. This will be
available in October, so you can buy this in October
and start playing with it. The other thing that we
did, because, same thing, if I give you this, you
won't be able to do much. So we created that
development board. This is our AIY team. Created the development
board so I can just slap this thing on it, like this. And I clip it. I'm not going to do it. Oh yeah, I did it. That was good. And then you have a full
system with gigabit ethernet and USBCs, and all
the things, HDMI. You can connect a camera to it. You can TPIOs. You can really play with that
thing and really get going. This also will be available in
October for you to play with. So we also work with partners
on the Edge TPU sites, so really working
with some [INAUDIBLE] vendors and some of the
device and [INAUDIBLE] vendors to leverage the Edge TPU into
solutions for our customers. So some of those partners here
are building either gateways, or really leveraging the
Edge TPU for their use cases. So we try to really kickstart
that ecosystem of partners focused on Edge TPU and
application of the Edge TPU for our customers and for you. Let's talk about scalability. Now that you're all
excited about Edge TPU, you're going to have a lot
of devices with intelligence. You're going to need some
scalable infrastructure in the backend. So IoT Core was,
as I said earlier, was really meant to be
completely serverless. And you can start
with one device, connect it, and
ramp up to millions of devices in production,
without really having to shard, having to plan for
memory CPU, anything. Everything is done for
you and you're just going to pay as you go. And this allows you to
go from POCs to pilot to a small deployment
to your production, really, with no effort. And that's really
a big game changer, we think, in terms
of how it helps you in your digital
transformation journey. So we innovated in the Edge. Obviously, we also kept
innovating on the IoT Core itself, and those are
some of the features. Some are better. Some are a little more early. For example, the Cloud
IoT provisioning service, that's an early access. So this one is about enabling
you to onboard and provision, literally, millions of
devices we've had to have, without having to go one
by one and provision them. So if you're buying a bunch
of crypto, for example, with microchip, you
buy all those cryptos, they're going to give you a
real ID and some information. You're just going to put that
in the Cloud IoT provision system and all those IDs will
be moved to your account, so you don't have
to do that manually. We did things like
gateway, which is a way to pass through
the authentication of IP or non-IP devices, pass that
authentication to the cloud directly. So the gateway is not
the proxy for all. You really authenticate the
devices behind the gateway, even though they may
not be IP devices. So that's an interesting one. We do groups. We do high-speed
messages to the device. If you need 100 messages per
second down to the device, we can do that now. And we added some extensive
logging capabilities to debug your system. So a lot of features. We're going to keep
going like that throughout the rest of
the year and next year. This is kind of the Core
system being improved. Now, I would like to welcome
another customer that is really changing the game in farming. So Jon Friedman
from Freight Farms. You all know about Airbnb
disrupted hospitality. We kind of view Freight Farms
as disrupting the agriculture. So I'm happy to
welcome Freight Farms. I think before we start,
we're going to see a video. JON FRIEDMAN: That's right. ANTONY PASSEMARD: So I don't
know if I click this or not. No? [VIDEO PLAYBACK] [END VIDEO PLAYBACK] ANTONY PASSEMARD: Excellent. Well, thanks, and welcome, Jon. So tell us more
about Freight Farms. JON FRIEDMAN: Yeah, sure. Happy to. I did notice a few people at
their session description. I know you're dealing
a lot with containers as it relates to the GCP. ANTONY PASSEMARD: That's true. That's true. JON FRIEDMAN: This is a little
bit different container, So you're going to be
finding out about that today. But we use that as a pivotal
piece of our infrastructure. We feel that shipping
containers and IoT are actually great building blocks for
the next level of food. We're looking to lower
the barrier of entry for anybody who's trying
to get into farming and make food supply
a lot more accessible. So getting into this, we
started looking at the data, but we were surprised
to find there's not a lot of data in agriculture. In one of the biggest
industries in the world, really hard to find consistent
data to build around. And the reason for that is-- there's a few reasons
for that, actually. One, there's not a
lot of environments that are exactly the same. So if you have a
farm in one place of the world and
another farm halfway across the world, the
results they're getting and the things
that they're doing, they're not going to match up. So all that data is very siloed. Another reason is as farms
have been consolidated over the years, that
generational knowledge hasn't been passed down. So we're in a very centralized
global food supply system, and the things that
have been leveraged are the crops, the chemical, and the
genetic attributes of the crops have been really focused on. And the ones have been
prioritized are the ones that can withstand
things like drought, are resistant to pests, and
they can travel really far and hang out on the
shelves for a long time. So you notice, I
didn't really say anything about taste
or variety or flavor, or anything like that. So if you are a chef
and you're looking to get a unique
thing on your menu, and you want to
make a dish that's just very signature
to you, that's going to be a really hard task
with our current food system. If you're in sourcing, or you're
in any type of food business where you're trying to
bring innovation to it, it's going to be very
difficult for you to get something
that's only grown in a specific part of the world
shipped all the way to you in the quantity that you need. So it's very familiar to
minimum order quantities you see in global manufacturing. So what we were
inspired by was actually 3D printers, and
that technology, and how that shifted
a segment that really couldn't get into the game
of prototyping or small run manufacturing. And we put that in their
hands so a lot more innovation could happen. So that's where we
came in and said, hey, is there a way that we can
build a platform that's accessible to everyone, uses
some infrastructure that already exists in the
world that everyone has some logistic support for? That's the shipping container. And IoT, which allows us to
get a lot of that transparency and traceability that
everyone's looking for. ANTONY PASSEMARD: So tell
us more about IoT then. How is that helping? JON FRIEDMAN: So
IoT is what creates those environments in the farm. So it's internet of the
environment for us, right? When our customers are
looking at the farm, they're looking at
all the components that make up the perfect
scenario inside the unit. So let's say you're
just getting started. IoT is going to be
what, basically, sets the environment around
a certain crop attribute and heightens that
across the way. So let me give you an example. If you are looking
into IoT as a way to orchestrate all the
different components, you can, in any environment,
create an environment that's halfway across the world. So we have customers in
Puerto Rico, Dubai, Detroit, all over the place. And inside that farm, it's the
perfect day of summer every day of the year. It's a very unique
environment in there. We can also create
environments that are not of this world with IoT. So you could take the
nutrients composition of Italy and you can match that with the
air quality of Salinas Valley, pair that with the
CO2 density that would be next to
an active volcano, as well as giving the
plant the light spectrum that it really wants. ANTONY PASSEMARD: I really
want to taste that lettuce now. JON FRIEDMAN: Don't you? So yeah, so IoT is a
really central piece for us to build these
environmental recipes, to match what's out
there in the world, but also to create things
that aren't of this world. ANTONY PASSEMARD: So
your customers, the chefs or the farmers, they use a
mobile app to do all that? JON FRIEDMAN: That's right. ANTONY PASSEMARD: Can you show
us, maybe, what you have here? JON FRIEDMAN: Yeah,
I'll give you a little-- ANTONY PASSEMARD: That's
probably the demo god a bit. JON FRIEDMAN: Let's see. It looks like it's casting. ANTONY PASSEMARD: Can
we cast on the screen? Oh, yeah. Wow. OK. So what are we seeing here? That's a real farm, right? JON FRIEDMAN: That's right. So I will keep it
upright for this demo. So this is what any
farmer who has bought a leafy green machine-- that's
what a Freight Farm is called-- they would see this, no matter
where they are in the world. Inside the farm, they might
use it as their remote control. So if they want
to turn something on and off really
quickly, they're going to use this same app. So let's, for instance-- this is actually a live farm
in Boston, Massachusetts, right now. This is at our headquarters. So let's go ahead and we're
going to turn on the lights. So all the devices
are connected. We're just going to-- beautiful. We turn that on. And now if we do
a quick refresh, we should see the lights start
to turn on in succession. You'll also notice that
the front lights are linked to the back right
lights and the front left lights and the
back left lights. So that ability to pair,
link, and create relationships within the farm is all
within their fingertips. We can also set programs
and timers around that. We can do [? column ?]
recipes, yeah, and I'll talk a little
bit more about that. Now, let's say I've
turned a few things on and I really like what
I've got growing there, but I want to track what that's
doing to the rest of the farm. So you can pull up your
historical analysis and see what effect-- maybe one of the things
that you've turned on has on another
thing in the farm. So that's great,
because you can really start to tune the balance of
things based on your needs, and so on. So you'll be able
to look at things like water temperature,
pH, the amount of nutrients that are in the water, CO2 in
the air, humidity, temperature. And there is a whole
range of devices that you can pair, create
dependencies around, put on timers, things like that. So let's say I'm really
happy with what I got here. I'm going to go over
to the Program section. These are all of the things
that you can map out, different dosing schedules,
different lighting timers, set different day
and night times. And then you can
save that recipe. You can save that recipe
as a unique thing for you, or maybe you want to
get another farm that's halfway around the world. Say you're a chain
restaurant and you want to have a unique
lettuce on your bun. That can be something
that you save, import to all of your farms. But you can also pull from
a community of recipes that other farmers
have created, or we have created for you, to improve
your performance of your farm. So actually, last year,
we developed a recipe to improve the efficiency of
all the models, all the way back to 2015, and updated
all those farms at once. So with this, and with Cloud
IoT, we see this as a way that we can continuously
improve these farms over time, use machine learning
tools to basically do this updating for us, and
always choose the best path. ANTONY PASSEMARD: Wow. Jon, this is awesome. Thank you very much
for your trust. JON FRIEDMAN: Thank you. Thanks. [APPLAUSE] ANTONY PASSEMARD: Let's
talk about security now. Security is almost
my favorite projects. It's where I started
my career, actually. So I've always been keen on
having great security for us. And Google will take security,
obviously, very seriously. And trying to look at
it from the ground, that the Google
Cloud Platform itself is built from the ground
up with security in mind, and IoT Core is built
on top of that platform. So we really try to
think about security from an end to end standpoint. You're going from the devices,
the connectivity, the Edge, the platform itself,
and the application. But today, I really
want to focus more on the device side, rather
than the cloud side. By now, you probably have seen
a lot of sessions around that. And one thing we've
done differently, when we launched
IoT Core, is the way you authenticate
devices to the cloud. We use what we call a JW token. We don't call it a standard. It's a JSON Web Token. It's pretty common to
authenticate against APIs. So we use a JW token to do the
authentication of the device to the cloud. The cloud is authenticated
through a center TLS session, but then the device
uses a JWT that it signs with a private key
and authenticates itself with a corresponding
public key in IoT Core. It has the same
level of security as a TLS mutual authentication,
but it has a lot of benefits. It has no dependency
on the TLS stack. It's very small in footprint. You can easily
upgrade your TLS stock without disrupting
your application. And it can also
authenticate non-IP devices, because non-IP devices
can create a JWT, pass it to our gateway, and send
it to Cloud for authentication. So a lot of really
good benefits. And actually, to demonstrate
the power of this, we have partnered with
Microchip to release actually what is fundamental. This is the first 8-bit MCU that
is securely cloud-connected. [APPLAUSE] Yes. This is the first 8-bit MCU
that is cloud-connected. This is really a
revolution because 8-bit has been there for many years. 15, 20 years, people
have been buying 8-bit. People know what they are. They're cheap. They're really available. But until now, everybody was
saying, no, for JWT, 32-bit. Let's go big. Let's go big. Well, actually,
you don't need to. You can take this 8-bit. And because our
stack is so small, because you use a JW token
to authenticate devices, that entire JWT
authentication can be handled by crypto
that's on there, and the whole TLS is in
the Wi-Fi module itself. So because it's so
small, it fits in there. If you use mutual auth,
you cannot do this. This is only Google-specific. So this is really a cool thing. This is going to come
probably around October. It will be available around
October for you to purchase. But this will really put IoT
everywhere very securely, which is super important. So finally, let's talk
about our ecosystem. We talked about the
ecosystem of Edge partner. That's great. But there's really
a lot of partners that we want to work with. One noticeable
ecosystem that we like-- May this year, we
actually announced that we joined the LoRa Alliance. We really like that
connectivity layer, so we're working with partners
like the Thing Network, or Obgenius in France,
or My Devices, all really good partner in the LoRa space. That's something we really
like and we're going to pursue our effort into. But we really look at partners
across the entire spectrum, from device partners
to application partners to service partners, to really
serve and to add solutions to our customers. So this is just a sample of
some of the incredible partners we have, and we're
really happy to support. Our goal is to make
our partners successful so our customers are
successful, and that's really key to our
strategy when we partner. So now, I want to
bring Jen Bennett. Jen Bennett is part of
our Office of the CTO, and she's been working
with many, many customers, helping them in their
digital transformation. And so Jen's going to talk
about a few customer use cases and how she's trying to create
an ecosystem of businesses, actually. So Jen, thank you. JENNIFER BENNETT: All right. Thank you very much. [APPLAUSE] Yeah, it's great. Thanks to our customers. So Antony mentioned that there's
a number of industries that are being transformed by IoT. And to be honest with you,
there's not a customer that I've talked
to that hasn't been thinking about, or embarking
on a journey in, IoT. So it is transforming
all industries, and across the business,
from customer engagement to product lifecycle
management to risk management and even operational
efficiencies and effectiveness. So it's really this
idea of transformation within the industry. And as we think about
that, this is really a catalyst for new business
opportunities, this IoT data. And not only that,
creating ecosystems that can leverage the data in
new ways, to drive new value. So why you embark
on an IoT journey maybe for one particular
reason, but then that data starts to become invaluable in
a number of different business cases. And so this idea of
democratizing data is really, really key
to the future of IoT. Let's look at an example. So this is a customer of ours. So Midas out of Europe,
not too dissimilar to Midas here in North America,
provides automotive services to their customers, and
they do this across a number of European countries. But with the growth
of the connected car, they could see that their
market was being disrupted. And in their words,
they said, "We had an opportunity to be daring,
to disrupt ourselves and become part of this
connected ecosystem." And so Midas has
partnered up with Z, who provides an
OBDII dongle that attaches to the vehicle to
provide data from the vehicle. And then they partnered up
with a GCP, the Google Cloud Partner, Tellmeplus, who
specializes in building machine learning models. And these machine learning
models that they built were all around
predicting failures. In this particular case, they
started with battery failures. You're in your vehicle,
you're going to start it, it doesn't start. Not a delightful experience. So again, Midas
partnered up with Z who partnered up and created
this ecosystem with Tellmeplus to build a machine
learning model that could take this data
from the vehicle and predict battery failures. Of course, this
is only the start. There's many other things
that they're going to do, but what this enables them
to do was to become a driver companion to their customer. So now, they've built an app
that can provide information about the maintenance schedules,
that can provide information about predictions on what
might be coming and happening. It can also provide
geolocation information, and many, many other
things in the future. They've become the
driver companion and part of the ecosystem. Of course, they're
looking to roll this out to about 100,000 customers,
but it doesn't stop there. This data, you can
imagine, can be leveraged, for them to drive down their
operational costs, managing their inventory, having
that spare part available before you even get there. I don't know how many of you
have taken your vehicle in only to find out they didn't
have the part you needed. It just happened to me recently. And then not only
that, but who else could value from this data? The OEM. The vehicle manufacturers. All this plethora of
data that they can now leverage to understand
their customer better. It's an ecosystem. It's an ecosystem built on data. I'd like to welcome John
Heard from Smart Park, CTO of Smart Parking. Maybe he can shed a
little bit more insight on another example, all
related to smart cities. So John, welcome. JOHN HEARD: Thank you, Jane. JENNIFER BENNETT: Maybe
give us a brief overview of what you do
with Smart Parking. JOHN HEARD: Yes. Thank you, everybody,
for being here. And let me just tell you a
little bit about Smart Parking. We're a company based out
of Melbourne, Australia, but we have offices
in New Zealand, where we do our development,
and we have sites in 17 countries
around the world, where we are about reinventing
the parking experience for you, for various customer
experiences. So they are also the
customer experience for the enforcement officer. I know you hate them, but making
their lives efficient and much, much more effective. Changing the experience
of finding a car park that's in a city, that you
know that a car park is just around the corner. You don't need to keep
driving around and around, looking for that car space
that may not be there. We also provide
payment and interaction for allowing you to do
payment for your parking right there from
your smartphone, or also connecting to your
existing infrastructure, the payment machines
on the streets. And by the way, we do
both on the road parking and also off street
parking, and multi stories. And so when you see
some car parking spaces, where they have the red
and green guidance lights, that's the sort of stuff
that we do around the world. JENNIFER BENNETT: Great. Maybe tell us a
little bit about some of the challenges that led
you to partner up with Google. JOHN HEARD: Good question. As we have been starting
to talk more and more with our customers, we're seeing
a change in the conversation that we're having
with our customers. No longer just about parking. It's actually about how do we
change the city experience? How do we make our lives
within their environments more effective and efficient? And parking, because it's
actually a well-understood-- we know about paying
for our parking-- it pays for that initial
establishment of the IoT infrastructure for a city. And so what we're seeing is that
the Smart Parking is actually deploying and delivering the
initial pillar, or tent pole, which enables a city to start
to broaden their intelligence services within the city. So it reaches out
into other things, such as public broadband,
and smart rubbish bins or garbage bins, smart
street lighting, and so on. It's essentially unlimited. But that required us to look
at the characteristics of what that type of information
now requires. We need to connect to anything,
because if that central gateway on the street lamp
post is doing parking, it's also now going to need
to connect to other things. And so we needed a
new class of system that was not just about parking,
not just about just dealing with those transactions
for parking, but essentially
connecting to anything, communicating with anything,
and then processing that information in smart ways. So it's not just about
parking, but for smart cities. JENNIFER BENNETT: We've
talked a lot about Edge. Talk to us about Edge
and your thoughts on this new opportunities
at the edge. JOHN HEARD: Yeah. Edge is really critical
to our real world. As we know, sometimes when
you try to make a phone call, you can't connect. Sometimes when you have
really time-dependent things, the time it takes to
do a communication transaction to a backend
computer cloud system may be too long. An example is we process
many, many, many thousands of images per day,
and we recognize the number plates of cars. And that number plate is
used to raise a barrier arm, because you've preregistered,
for example, to go into that car parking lot. Now, if that
barrier arm does not open within about
two seconds, you're starting to create a queue of
vehicles outside that parking space, and that really
becomes critical, especially at 10 to 9:00
in the morning when you need to get into the office. Well, more importantly, it
creates congestion, also. So the Edge computing-- just one example of how the Edge
computing is really critical is to do that
processing in real time right there where you
need that information. And that Edge computing needs
to be quite intelligent as well, and that's actually what
we're so excited about, the HTPU capabilities,
because we can start to do a whole
variety of intelligence right at the Edge to guarantee
real time licensing. JENNIFER BENNETT: Yeah, I know. JOHN HEARD: That's
just one example. JENNIFER BENNETT: Fantastic. I was talking to a retailer,
and they always think about inside the four walls. And I said, one of my
biggest frustrations is finding a car spot. So when I talk about
ecosystems, it's like, how can we take, now, that
Smart Parking that navigates you through the journey that
you do in a day, where you're parking to go into a
building or going to the store. Right, so this ecosystem,
and I know that you probably face this quite a bit. It's really about
how do we start to enable these ecosystems? JOHN HEARD: Exactly. It was interesting,
John just commented about the internet of farm. And I was just thinking
as he was speaking, we're creating the internet
of parking for you. That's what we're doing. And actually, that's the mission
that I feel very passionate about. We're reinventing
that experience. And it's an experience that
we currently live with, and it sucks. And so the opportunity
here of not just the parking and changing
that dynamic, but making it more cohesive
and more personalized to what I'm trying to do. It's not just about parking. Usually I'm parking there
for some other reason, right? I'm trying to get
to a store, or I'm trying to get to the office,
or trying to do something else. And it's an
inconvenience right now, and that's what our mission
is, to actually do that. JENNIFER BENNETT: You guys
have been a tremendous partner. We thank you for
your partnership, and thank you for
joining us today. JOHN HEARD: Thanks, Jen. [APPLAUSE] So as we think about
connected things, one of the things that's
really critical in IoT is this idea of
location intelligence. And Antony mentioned
Google, we've done a lot with
connected things. But it's no surprise, I'm sure,
that we have a lot of work into location intelligence. So if we take the very best
of maps and routes and places, and we start to
apply it to things like asset tracking, or perhaps
efficient routing for some of the most complex
fleet itineraries, tracking or visualizing
your connected asset-- most importantly,
no matter where they are all around the world. This is geolocation, both
inside and outside, geocoding, distance matrix,
real-time traffic, real-time road conditions
all at your fingertips. Such a key asset to the
ecosystem of IoT, and we're very, very proud of
the work that we've done in cooperation with
our maps team as well. So no matter what
you're tracking, this solution, this
integrated solution, can really help you start
to manage your bottom line, grow your top line, and
manage the risk as well. Let's look at a few examples. So Vagabond is a company
that provides technology and operations for the food
and beverage service industry. So vending machines. And they embarked on this
journey of an IoT solution, where they connected vending
machines, getting information about stock, getting information
about the cash in the vending machine. But what they were
really looking forward to combine with that was this
idea of precise geolocation. And you can imagine these
vending machines may not always be in the easiest
places to locate. It could be in the
third floor dormitory or it could be in a
back alley, right? So this idea of having
world class geolocation was really, really critical. And now, they can overlay that
vending machine information on precise location
so that their drivers can be very efficiently routed
for restocking activity. And they've been able to drive
some tremendous improvements, in terms of both
the top line growth, as well as managing their
bottom line, and theft. So theft has been reduced
about 15% off the top line. So some really great example. Another example is a
company out of Australia called Fleetminder. And they provide vehicle
and asset tracking solutions in about 12 countries. And really, what
they wanted to do was be able to visualize the
location of these vehicles and combine it with some
of their GPS technology that they have, giving
dispatchers not only location information, but really
optimized routes. And so you can actually
see now even surrounding, the real-time
surrounding information, and really start to drive
a whole lot of use cases. For example, being able to
understand maintenance activity and where the closest
maintenance shop would be for that driver. Some really great
results from Fleetminder. About a 50% decrease
in overtime, which if anybody who's been
in the trucking industry, this is a huge problem. And a huge cause of turnover
is their time away from home. So some incredible
efficiencies that have been gained through
doing some of these things, not to mention really
reducing incorrect deliveries. So this has been really
game changing in terms of their business value. So we talked about IoT
delivering new insights. Antony mentioned intelligence. We talked about
intelligence at the Edge. We've talked about
scalability and really taking the opportunity
to drive new business. And we talked about how IoT
is enabling an ecosystem. I'd really like to pull
all those pieces together with you and share video. So can we roll
the video, please? [VIDEO PLAYBACK] BRIAN BEVERIDGE: West
Oakland is surrounded by three major
freeways, and it is downwind of the Port
of Oakland, which is the fifth largest
port in the country. The community is a real
blend of land use, warehouses next to homes. And it's only been recently,
in the past half century, that we've realized that this is
a bad layout for public health. MARGARET GORDON: We have had
clusters of cancers related to respiratory problems. MILLIE CHU BAIRD: What
makes airborne pollutants so dangerous is that
they are not always visible to the human eye. EDF and Google brought
together other partners to collect, measure, and analyze
air pollution data in Oakland. So Google brought an [INAUDIBLE]
who equipped our studio cars with their acclimated
environmental intelligence platform, which captures
scientific grade air volume measurements. MELISSA LUNDEN: What we've done
is miniaturized basically air quality laboratory into the car. So it's able to provide
high quality measurement at fast time resolution. MILLIE CHU BAIRD: EDF partnered
with academic scientists to analyze the data that we
receive from the Google street view cars. The measurements
and analysis has shown elevated levels of black
carbon and nitrogen oxide in Oakland. These types of air
pollutants typically come from the burning of
fossil fuels, cars, trucks, and other industrial sources. BRIAN BEVERIDGE: We're
really beginning to see, in much higher detail, what the
air we breathe on the ground, in the neighborhood, looks like. MILLIE CHU BAIRD: We hope
this new day can empower local community and
community groups to advocate for cleaner air. [END VIDEO PLAYBACK] JENNIFER BENNETT: So you can
see quite an ecosystem already being formed around
collection of data from vehicles and air pollution. But this ecosystem
is growing even more. In May of this year, Kaiser
joined into the ecosystem, and now what they've
done is take that data and combined it with
electronic medical records from residents who
live in this area, and did a study around
how that is impacting, a block by block
analysis of the health risks from air pollution,
the most detailed analysis of its kind. So we're really, really
proud of this partnership, this ecosystem, that's
forming to solve new problems, to solve old problems
with new approaches. Really, really
thrilled about this. So if you want to
learn more about IoT, we have a breakout session
tomorrow in the morning. Antony mentioned this
about the IoT Edge. We have a showcase on the-- sorry, we have an IoT area in
the showcase in Moscone South, but we also have a lot
of information online, and I encourage you to go
there and check it out. With that, our session
will be completed. I'd ask for you to provide any
feedback in the mobile app, and I hope and thank you
for joining us once again. I'd like to thank Jon
from Freight Farms. I'd like to thank John
from Smart Parking. And I'd like thank
Kjartan from Aker BP. Tremendous customers,
tremendous partners, and I hope you guys have a great
rest of the Next event. Thanks so much for coming. [APPLAUSE]
