thank you everybody for being here this evening
we really really appreciate it we know all of you
have busy jobs many of you probably have families
at home and there's a million things that you
could be doing but the fact that you would spend
an evening talking about and learning about AI

and how the world is changing with us we don't
take for granted we're very humbled by it thank
you so much for being here we we we really really
appreciate it before we dive in uh just a quick
bit about me I consider myself the luckiest guy in
the world I've been married to the same woman for

18 years we have two wonderful daughters they're
16 and 13 everything I do at the end of the day
is for them we're die-hard Buffalo Bills fans
in case that wasn't clear from the picture up
there did I did I hear did I get a clap from that
oh that's for my team so that doesn't count but

uh we took off The Undefeated Chiefs on Sunday
night I was there for that game it was awesome
yes I did get a woo from somebody um I've also
been very lucky in my career I've always known
what I wanted to be when I grew up I got a TI-99
that my uncle gifted me when I was 8 years old

mostly because I wanted to play all the video
games that it came with but in the bottom of
the box that he had given me there was a book
with Texas Instruments proprietary programming
language and a bunch of programs in it and I
wrote my first computer program when I was 8

years old got completely hooked very good at math
and science growing up crappy at everything else
frankly which I which I now regret but studied
computer science in college started my career at
Bloomberg where I spent the first 11 years writing
systems for for financial professionals and and

people in the sports industry I know there's
some Bloomberg folks here thank you for coming uh before spending a couple years as the CIO at
Dow Jones where we led a pretty big cloud-based transformation and I got religion on public Cloud
which is what led me to AWS where I spent the

last eight years before I joined Google two years ago
leading their Enterprise strategy and running their Marketplace before I landed at Google a
couple years ago to to lead our Marketplace and all of our Tech Partnerships so companies
like arise and many that I might refer to

over the course of the next 10 10 minutes or so
uh we we manage the Partnerships with those teams and make sure that they're not just really well
built on Google Cloud but really well built in order for you to use on our Marketplace and and
to take advantage of as you as you build your own

workloads AI AI agents or otherwise on on Google
Cloud uh and one of the reasons I joined Google is because it became abundantly clear to me that
moving to the cloud had become table stakes after eight years or so at AWS every company in
government is going to end up moving a material

amount of their workloads to the cloud because
the value proposition is so great but now that everybody is going to be doing it just moving to
the cloud is no longer the key Difference Maker it's really what you all and what our customers
are going to do when they get there that's going

to make the difference and I felt that nobody
was better positioned than Google given some of our strengths in analytics with tools like Big
Query where we basically reinvented the category and then specifically nowadays in artificial
intelligence where the people at Google some

of whom you're going to hear from today wrote the
paper "Attention is All You Need", which invented the Transformer model back in 2017. Uh, and really, um, is pervasive throughout everything that we we we do at Google. It became very clear to me that we
were in the best position to really help customers

navigate through this change. And I think there's
no other way, better way to demonstrate that than if you look at um Gartner's Magic Quadrant over
the past couple of years, which had Google still on the top right, but in the bottom left-hand portion of
that uh top right quadrant, really moving up and to

the right further and further over the last couple
years. So again, I'm the luckiest person alive. I feel very very lucky to to be here helping customers uh
move through that. And one of the things that's fueling our growth and that movement up to the
right is the ubiquity of generative AI and our

position um here. So if it wasn't already clear to
you, we believe that gen AI is going to transform just about everything. We're seeing huge amounts of
budget and focus shift from some of the largest companies and governments in the world to harness
generative AI to improve their business. And if you

think about a few years ago, machine learning
wasn't necessarily new. You'd see a lot of use cases around predictive maintenance and what
you know, maybe to decide which product might fail on an assembly line or looking for faulty
products in a big e-commerce site. But but over

the the last few years, as you see the ubiquity of
large language models in generative AI from folks like us, Google, Mistral, Cohere, Anthropic and others,
we're seeing a litany of use cases where customers are now creating or generating lots of content,
whether that be uh code, job descriptions, we work

with some HR companies where they're giving their
employees tools to create job descriptions or the
first beginning of an employee review let's
say product descriptions marketing campaigns
travel itineraries we're working with a training
company who's creating educational lessons and

helping folks learn foreign languages by being
able to converse with avatars and sort of score
the results of that conversation and real world
scenarios like ordering a meal at a restaurant
or taking an Uber or talking to a cab driver but
perhaps one of my favorite use cases is we're

seeing AI companies also rise up to create dating
profiles for any of you online dating fans out
there I thought that was a sort of clever uh
clever use case that we're seeing as well so
um look I know it's still a pretty new space and
there's lots of experimentation going on and there

are some who think that it's a bit of a hype
cycle that's yet to pay off but I continue to
believe that we're in the dawn of a really new era
uh where we're going to continue to see lots of

specialized agents prop up and if you think about
how we think about the world and the technology
that we offer to help our customers navigate that
um this transformation we're thinking about it
and offering services and capabilities at every
single layer of the stack to meet any one of our

customers where they are whether it's training and
building their own models building applications
on top of it the tool chain around to support it
supported by an ecosystem that I'll go through in
just a moment but at the lowest level we have our
infrastructure and our chips uh some of which we

custom create including the tensor processing
Unit or TPUs that some of you here may use uh
to maximize the price performance that you get
for training workloads on top of that we have
our model Garden which has over 130 models that
customers can choose from whether it be from

Google or from some of our partners we have a
way for customers to build their own models
whether that's to fine-tune you know their own LLMs
from the ground up or to create custom adaptations
of existing models by either prompt training it
or tweaking it with some of their own data or IP

which can all remain in your VPC or your tenant
where you get to control kind of the security of
the data um and the keys then we have an agent
Builder to help customers build their own AI
agents on top of it and then a number of more sort
of fully managed services that are powered by AI

to help our customers transform their contact
center their healthcare systems and uh Etc and
from the very beginning of Google Cloud we always
knew that we were going to need a robust partner
ecosystem to help our customers many of whom
are relying on us as consultants as part of a

much larger transformation to meet them with the
tools that they're already using so when we think
about uh our ecosystem we're 100% committed to
delivering the world's most open and innovative
ecosystem again at every single layer of the
stack so when you think about AI in addition

to our own chips we've got GPUs an AI toolkit
software from folks like Nvidia Intel and AMD
on top of the chips like I mentioned we've got
over 130 models in our model Garden including
some of the most well-known models from folks like
anthropic Cohere AI21 Labs and Mistral on top of

that JetBrains and GitLab are bundling some of our AI capabilities into their IDEs so that they can meet developers where they are and we're super excited to be partnering with Jason and Arise as well who's also building the observability and the monitoring that you need when you're building AI

workloads on Google Cloud, MongoDB, Pinecone, and Elastic. They're integrating all their database and their search technologies into our stack for customers who want to continue to use those tools. Data providers such as Moody's, Eleg Dun, and Brad Street are making data available for customers

to train and tune their models as well on Google Cloud. And then we have a number of higher level SaaS applications from folks like Workday, um, SAP, and Salesforce who are really weaving some of these AI capabilities into their products to to help make them more efficient for customers

as well. And we're really seeing that AI, which is Arise, that's funny, that's a pun unintended, uh, arise in um customers creating really specific AI agents for specific use cases. So for instance, Dialpad, they've created customer service agents that their customers buy as a SaaS model to

really help transform their contact center work. Day in, UKG, 2R Information Services companies, um, and I mentioned some use cases there before where they're creating agents to help employees and the companies who use their service create job descriptions or employee reviews or to get started

on um uh, you know, filling out let's say a W2 or W4 form. Replit, GitLab, and Pinecone all have created coding agents that allow companies to write code faster and build gen applications on to on top of them. Jasper, it's a cool company building creative agents to deliver on brand and data optimized

marketing assets very quickly and then you have
folks like Palo Alto Networks and Exabeam that are really creating security agents to help those
folks who work in a security operations center, um, really find the needle in the haystack and help
them narrow in on and focus on where

threats are actually real and they should focus
their time. Just as a couple of examples and just some quick customer examples just to make this
real before I pass it on to our next speaker, uh, Wayfair. You've probably heard of Wayfair, one of the
world's largest e-commerce companies. They've been

using Gemini Code Assist to not just help all
of their developers be productive with, you know, more active code complete, but I suspect there's
a lot of developers here, right? How many folks are developers here? All right, keep your hand up if you
love writing unit tests, right? All right, one person

I got, I got somebody loves writing unit tests. Um, uh, Wayfair has been able to automate, I think it's 55%, something like that, half or roughly half of their
unit tests and their code base, which I feel better, is a consumer of Wayfair knowing that they have
better test coverage when I'm going to be giving

them my hard-earned money, um, uh, on buying things
on their site. So I think this is a pretty cool and transformative use case for them. Uh, captioning
from Warner Brothers Discovery, taking a voice model and listening to what's happening on The
Voice to produce the text on screen seems pretty

mundane, but that's that can be a very expensive
costly process, uh, and they've done this with an 80% improved efficiency rate on on captions
that they want to create for the vast amounts of content that they create. Uh, who owns a car? It's
another, do a show of hands, how many car it's in

New York so maybe there's a little less folks
the of those who own a car how often have you ever opened up your user manual I've owned a bunch
of cars over my life I've never opened up the user manual they Volkswagen created a their own
adaptation of one of our models to allow their

customers to basically interact with the user
manual of the car specific to them which of course gets smarter as the customer is asking more and
more questions around how do they change the oil or the brake fluid or you know what have you you
can tell I'm not a car person but you get the idea

uh and Snap who many of you or at least your
teenagers I can't get my teenagers away from Snap uh probably use on a daily basis and they've
infused a bunch of AI capabilities to help drive their engagement up two and a half times uh in
the USA uh for their customer base here but you

don't have to take my word for it I thought I
would show you a quick video uh of Evan who's the CEO of Snap and a big Google customer talk
a little a little bit about what they're [Music] doing Google cloud has been a partner
since day one for SNAP since Bobby and

I were working in my dad's house and scaling
from the first couple of users our friends to now hundreds of millions of people all
around the world my AI is a brilliant best friend powered by artificial intelligence it's
a chatbot where you can ask general knowledge

questions learn more about the world and
have a fun and friendly conversation and as we started partnering with Google and
experimenting with Gemini's capability what we found was an unbelievable ability to help
people solve their problems based on visual context my eyes great at answering
everyday questions and now that

it's integrated with Gemini, you can
just send a snap and ask something like, what's the healthiest snack?
Let's see what my AI has to say. It looks like Kind bars are the healthiest
[Music] option. So by partnering with Google, 

we can leverage the multimodal capability
of Gemini to understand what's going on in
a snap, even read the caption and reply. Snap's
community is distributed all over the world
and that's one of the reasons why Gemini's
multilingual capabilities were so exciting 

to us. We think that this new partnership with
Gemini and my AI allows people to learn so much
more about the world and to do it really
quickly in the moment. With my AI powered
by Gemini, I can easily practice my French,
translate this to [Music] French. Gemini's 

multilingual capabilities make it really easy to
translate text and of course, learn a new language. Ah, Google Cloud's flexibility and
ability to scale with our business
has allowed us to focus on what we love to
do, which is innovating for our community. 

With Google Cloud, Snap is building the
new way to stay close with friends and family. Super cool. I thought, uh, I showed
my daughter that capability, what's the
healthiest snack, and she looked in our cupboard
for the healthiest snack and then took the other 

one anyway. Your mileage may vary. Uh, we're
super excited about our partnership with
Arize. Um, I think it's a long time coming to
have, have a company with some purpose-built
capabilities really try to help customers
solve the observability and the monitoring 

problem it's a whole new frontier it's a new
set of capabilities it's a lot of uncharted
territory that we're you know that we're all
navigating here together and um Jason and his
team have been a great partner for us really
helping customers who are building some of

the most sophisticated and transformative AI
capabilities on Google Cloud um uh and doing
so in a way where they can feel that they're
doing it eyes wide open where they can quite
literally see and um uh uh what it is they're
building and and and how it's performing so

we're super excited by the partnership Jason
thanks for putting this event on with us uh
today and um hopefully you get a lot more out
of the other speakers to come and again thank
you for being here we don't take your time for
granted I hope you continue to enjoy the show

Google Cloud Keynote with Stephen Orban

Arize AI is the watcher, troubleshooter and the guardrail on deployed AI.

