Digital Transformation and the Journey to a Highly Connected Enterprise
Articles Blog

Digital Transformation and the Journey to a Highly Connected Enterprise


– This is gonna be a fun
presentation and really what I’m gonna end up
talking about is the use of neo4j and graphs,
generally across a variety of industries and then
how that applies to both, how graphs themselves
have evolved as well as where they’re going. So, the exercise here,
first introductions. We’ll talk a little bit
about the advantages and graphs themselves from the standpoint of what we’re doing, how we’re different from other database systems and platforms. We’ll talk about government. We’ll talk about finance. We’ll talk about these digital
transformation initiatives that are so compelling. Little bit about myself. I’ve been with neo4j for
about a year and in that time, I live in northern California
in San Rafael, that’s north of the Golden Gate Bridge. Neo4j is about halfway down
the peninsula in the Bay Area so I have about a 45 mile commute. It takes roughly an hour
fifteen, sometimes an hour and a half each way to get
there, so while doing that I’ve started listening to
audio books and I’ve listened to a lot of audio books and
these are some of my favorites that I’ve enjoyed, Steven
Johnson’s, Where Good Ideas Come From, and as it turned
out when I mentioned that to Robbie at Intel he
recognized it immediately which was great and then
referred me to the reference material prior to that. The Right Kind of Crazy is
a really interesting book about, written by a project
lead at JPL, the jet propulsion laboratory,
about getting to Mars, and putting the largest set
of rovers on top, onto Mars. Burglar’s Guide to the City
is super fun in that it talks about cities themselves are
laid out as mazes and how criminal behavior is such that
they don’t look at buildings and they don’t look at
roads and streets and sewers the same way as you and I would. They look at those as possibly
entry and escape routes. Win Bigly is very interesting
in terms of promoting influence or talking about influence. It’s written by the guy who
does Gilbert, Scott Adams, and he uses this idea of how
influence peddling operates in society by describing all
the major events of the 2016 presidential campaign and
so, he was one of the first people to predict that the
president was actually going to win the election,
solely based on his ability to influence people in a
variety of different ways. Then of course, Talk Like
Ted tells you to tell a story about yourself when you
begin your presentation. So, the interesting thing
about all those books is that they’re all have a common theme. They all talk about connectedness
and connectedness, so I’m putting up here, this is
for Rice who’s got the talk to the hand sticker on
his computer, where we put up here the social network of
Game of Thrones and you can see how every one of the major
families, the Starks, the Lannisters, etc.
are all connected across the Game of Thrones ecosystem. So, a really great example
of a social network and actually one of
the, and I thought about putting up Facebook and
Zuckerberg in this instead, but decided to really stick
with, rather than people who were going before Congress,
stick with HBO and some TV. Connectedness in social
networks is really the first generation of the graph
wave, as was, let’s say think applications systems
like LinkedIn or like PayPal doing money transfers or
Cisco in building routers and networks of connected routers. So this whole idea of
connectedness as it’s represented inside of graphs you can
see that graphs really are, or this ideal of connectedness
really exists everywhere. It exists in an organization. It exists in those social networks. It exists in activities that you might do like product subscriptions
or network operations. Long ago I was, got to Restin
and I’d been inside the MCI, what was then the MCI
network operations center. I think AWS probably runs that right now. Really, really interesting
kind of environment, huge network operations kind of facility. So, connectedness is
the theme and how things are connected is where we’re
going in our conversation right now. Let’s talk about what neo4j
is doing in that regard. We’re a graph platform. So graph is the mathematical
term for describing connectedness and how things are related. It dates back to, it’s
about a 300 year old set of principles, graph theory,
dates back that long ago and what we’ve done is we’ve
built an enterprise grade platform that lets you store,
reveal and then traverse or query all your data
relationships across your system and there’s really, really
clever things you can do once you understand how one piece of data relates to another to another
to another and then some. So, the whole goal is to
traverse and analyze any level of depth across all of
these different networks. It’s actually designed to
grow and build a variety of different types of
graphs and all the while you’ll hear us talk about,
in terms of the technical benefits of what neo4j
does is offers high degrees of performance, ACID
transactions so the data’s always gonna be reliable, super agility, right. Our customers, including the
ones that are speaking today probably took anywhere between
three months and six months to deliver to their first
project in neo4j and then, they seem to all be iterating
on about a quarterly basis. Every three months they’re
cycling through new stuff, new applications, so that
kind of agility is what we’re talking about. We’ll talk a little bit,
my folks will talk about graph algorithms this
afternoon and the benefits of developer productivity,
how speedy the environment is. It scales like crazy, etc. So, what I wanna end up talking
about is setting the stage or setting the premise of where
we’re seeing and involved in connections and we’re seeing
the rise of these connections in data across a variety
of different types and we originally cast this is
recognizing that of course, social networks, there’s
networks of people, right, but then there’s also
networks of activities that those people perform. That’s our business processes. And then, there’s networks
of knowledge that they build over time and they wanna
keep referencing over and over again. And finally, the one
I don’t put on here is then there’s the network
of technology, right. The digital systems upon
which you’re performing all of these different exercises. So keep thinking in terms of that way. People, processes of the
things they do and then networks upon which they
perform those activities. That’s really the core sets of
that’s principles that neo4j is operating within. So, in the age of connections
here, what we’re really seeing and this is really
truly starting to happen, you’ll see in the use cases
that I’m about to show you, connecting disparate silos
is a universal challenge. I’ve been in computing
for 30 years, right. The activities around, how
do I connect my CRM system to my customer support
system to my marketing system to my supply chain? Those are universal problems
that in my early world of business intelligence
were trying to solve always the themes that you hear
from every vendor is, connect your silo data. So, graphs are a new take
on this and the beauty of it is graphs are a
non-intrusive take on this. They don’t require you to
throw out anything that you’ve got right now, they’re
actually complimentary to all of that and then
as you apply them and roll them out they reveal how
connected all of your different data elements indeed might be. And again, that’s a theme
you’re gonna keep seeing throughout the course of the day. So, the next wave that we’re
seeing, the current wave that’s happening right now,
is organizations are figuring out that graphs are a
vehicle to drive innovation and really build off of,
determine their next major competitive advantage. This is why the theme of
digital transformation is so interesting is because
we know that there’s disruption going on all over the place. We know that there’s the Ubers
and the AirBnbs and the like are all crashing into
Amazon, are crashing into traditional businesses and
the traditional businesses have to react quickly, so
graphs are a fantastic way for that to happen. So, we talk a little bit
about how that happens and what we’re doing here. So, in a typical kind of
environment your typical systems that you have, this
is for the 40 percent of you who have never seen graph
or don’t know very much about neo4j, the others,
you’ll say yeah, that’s exactly what they do. So, you’re typically
storing and retrieving data as tables, as documents,
as key value pairs, any way that you wanna store
it on the left hand side. But then, if you put it in
neo4j the cool thing neo4j does it starts to reveal the
connections in that data. And so, what’s most interesting
about that is when we reveal the connection and
continue to work through that process the connectedness
actually increases the value of the data that
you’re evaluating and looking at. So, we’ll keep kind of
operating against that theme and of course remember
neo4j, we 20 years ago, 18 years ago, we invented
this label property graph or this property graph concept of how data was interrelated roughly
at the same time as Tim Barrensly was developing
the RDF framework and core pieces of the internet and
it’s a really simple concept. You take your data entities
and I have two people, Dan and Ann, they each
have birth dates, they have Twitter handles, they happen
to be married, they happen to cohabitate and Dan drives
the Volvo, but Ann owns the Volvo. See, all we’re doing is
illustrating this idea of connectedness and the
relationships amongst people and objects, things that
they’re doing and objects upon which they’re doing it or
those digital networks. We’ll talk a little bit,
the terms that we use very frequently, nodes
are the big circles. Those are your almost like
nouns if you refer to your data as sentences, right. Nodes are the main data elements. Relationships are the verbs
or how these things interact. And then, properties
are just the attributes that you might apply to these. The beauty of this whole
thing is that’s as much schema as you really need to
understand or know in order to be productive in the property graph. Now we can apply, if you’re
looking at existing data that you have and you have
to make sure that a customer ID number exists in
every one of these lists of properties, you can do that. You can enforce that. So, we have schema enforcement
kinds of capabilities built into the system. But, that’s fundamentally
the type of data model that you’re building and dealing with. The beauty of that is that
that’s exactly how we think. That’s exactly how we were taught. Like I said in the data
modeling world of 20 years ago when we were taking ER
diagrams and modeling those out or data flow diagrams and
modeling those, the idea of a data flow diagram is very
similar to a property graph. The beauty of it is it’s
looking at how the data elements relate to each other and the
great thing with this is, with neo4j is that it never
forgets the connections themselves, the relationships
and how important they are. When you’re operating in
SQL you’re making a table and the closest thing that
you’ll have to these connections are your foreign key
relationships to other tables. Now, those relationships
are often manifested when you run the query in SQL,
but then they disappear after that. There’s no context for you,
the context is not preserved for that relationship and
that’s one of the real big differences that we do within neo4j. It’s all about when you’re
conceiving your idea, you draw it out on the whiteboard. I have a gimmick that I use in sales call. The gimmick is I’ll walk into
and when we’re introducing people around the conference
table I’ll get my SE to go up and draw the
relationship, so it puts it in the account, wherever
your company, and then as you introduce yourselves we
just jot down your initials and your roles and then the
relationships between you and your other constituents
and then we can jot down neo4j and my role and
my sales engineer’s role and connect those all up together. The beauty of this is
I’ve just taught that room the most initial elements of graph ideas and I built an org chart for myself. So we snap a picture and
then the rest of the team knows who we were talking to. So, connectedness is this
idea that differentiating neo4j from everything else
and when I talk connectedness I am talking about how
we conceived the idea of the relationships amongst
my data, how I code it, and I have a really interesting
easy to use query language called Cypher, how I compute
it, I’m always thinking about how to optimize those traversals, right. Those relationships as
it’s actually executing in the environment and
then how I store it, right, all the way down to bare
metal, we’re thinking in terms of relationships and making sure that that connectedness exists. We’ve been doing this for like
I said, about 18 years now. The Community Edition for
neo4j was released right about 10 years ago and
has the uptake of that is astronomical, 10 million
downloads of Community Edition thus far. About every year we’re
holding events worldwide in about 400 different locations. We’ve trained more than 50,000 people. We’re executing more than 50,000 meetups. We’re operating and right
now seeing that there’s more than 50,000 instances of
Amazon of AWS that are actually powering and
running neo4j graphs. So, let’s talk about a
little bit how it fits in your architecture
and then I’ll get into some of these use cases. Oftentimes just revealing
connectedness out of, out of SQL information very, very easy to do. Connecting different
disparate silos and evaluating what those connections and
those relationships look like from these silos, again, very,
very simple, not intrusive kind of exercise, but
reveals those connections. And then, in the largest
case revealing those kind of connections from data links
and from, and capitalizing on your existing investments
in things like the dupe or other no SQL systems. These are ways in which we,
the installation patterns that we’re seeing in neo4j. Our favorite sets of use
cases really, really, these become actually kind of
no brainers once you start to think about it, real-time
recommendations, fraud detecting, network
management and IT operations, master data management, we’ll
talk a little bit about that, knowledge graphs, right,
maintaining the institutional memory of an organization
and then using that to your advantage, identity
and access management, another great, great one
happening and very appropriate for the intelligence agencies,
for financial services, etc. Okay, and then here’s some
of my favorite customers that are actually supporting
those kind of applications. Walmart, we’re powering
their, one of their websites. Itau Unibanco, so the largest
bank in South America. They’re actually looking at
building a system that has their four million account
holders, but then they wanna see how everything is
related to everything else. They’re, by 2020 they’re
expecting to have 93 trillion relationships identified in their system. They’re really tryin’
to map out their entire customer constituency and
it’s all to make sure that your credit, that their
customers are credit worthy, that they are, their
identities are valid and that they’re trustworthy to do business with. We’ve got HP, Cisco, I’ll
talk about Airbnb in a sec, UBS on the metadata, lots of
organizations are investing in these kinds of use cases. So, in the software
space, financial services. I’ll talk a little bit
about the IoT use cases for Comcast Intellia. We’ll talk a little more about some of the retail use cases, etc. I think the big thing we’re
trying to describe here is we’ve got over 250
different organizations who are operating projects and tens
of thousands of open source users, many like yourselves
working on neo4j right now. Graphs are, graphs are really here. Graphs are really
operating very, very well. It’s an interesting question though. I do a lot with market
analysts and things like that and my question that I pose
to them very, very regularly, ’cause I walk up with 25
different customer use cases over and over and over and
over again, eBay, Comcast, Scripps Network, I’ll show
you that one in a second. And what I end up talking
about is the connectedness not only in the kind of
things that they’re doing, but the connectedness of the
people in the organizations that they’re doing it with. So, developers are building these real-time graph applications. That was kinda the first
commercial set of use cases at neo4j was chasing down. Generation one kind of use cases. Generation two gets us to
data hub and what we’re seeing there is both modernization
and migration of old systems, legacy systems into
cloud-based systems like moving from people softer workday. We’re seeing and University
of Washington is a use case there where they’re migrating
six hospitals worth of data in addition to the school
that has 80,000 students and employees associated with it. Get to the data hub, right, and make that much more convenient. We’ll talk a little bit
about GDPR and data privacy regulations, right, and how
that effects or how a graph can actually help with that. We’re getting into data science. We’re getting into AI situations
where the data scientist is looking at, what can I
do with the new sets of data that I wanna add into the graph? And so, the really interesting
things are, as I navigate through an organization
with this kind of use cases there’s lots that pass
through an organization. The second, the biggest
observation we have though is the graphs themselves
are really hungry for data and when I mention that you
know, the iteration speed that my customers are going
through is like three months or less, it’s because
they wanna add more stuff to the graph. Now, map this idea back
to one of the books I was talking about earlier,
this idea where innovation comes from. Innovation comes from
marrying up two disparate sets of ideas and then seeing
if they actually click of fit together making
a new node and making new connections,
identifying what they call, what they refer to as the
adjacent possibilities. That’s what’s happening in
here as graphs seek more data. They wanna build more connections. They wanna add more nodes. And so, we’re seeing so that
strong appetite for data, it loosely fits towards you
know, the decades old now idea of Metcalf’s law of the network. The value of the network
increases at the square of the number of nodes or number of users. Right now they’re proving that
out that sometimes it’s more, sometimes it’s less,
but it’s certainly not, it’s still an exponential
kind of, kind of concept rather than just a linear one. And like I said, we’re
seeing iteration at like the three month cycle right now. So, let’s look at a
couple of use cases here. These are gonna be, they’re
kinda fun and ideally somewhat localized to, to the area. So, historically I was
raised on the Space Program. These are photographs from
my house and my dad’s house. So, I’ve got every
mission patch from Mercury on through the first
instances of the Shuttle including the Apollo saw
use missions and every one of the Apollo missions right there. That hangs in our living room. And then the other picture
here, the Aviation Week picture, when I say I was raised on
the Space Program, my dad worked for Hamilton
Standard and they built environmental control
systems for the Shuttle as well as the suits for
both Apollo and Shuttle. So, the funny, the funny
thing that my father would end up doing, and he’s
actually the guy on the left, so he ran the Space
Department, but he also has a really ridiculously big head. So, he was actually, for
internally, he’s got a 95th percentile head, so he was
the one that if the helmet fit over his head it actually passed QA. If it got stuck on the bridge
of his nose, it didn’t. I’ve been to Edward’s Air
Force Base, that 1977 launch of, or approach and landing test. I’ve been, saw the fourth
liftoff of Columbia down in Cape Canaveral and
laying out the use case here is going to be NASA as a
graph use case, but I’m fairly well invested as well as
understand many of the use cases and situations that NASA’s
gone through in their entire history. That’s a huge bucket of
knowledge that they’ve got and guess what they did. They put it in a graph. It’s called the Lessons Learned Database. So, they’ve taken all this
historical information about mission information,
about successes and failures and trials and errors
that they’ve gone through. This half a century of this
NASA engineering knowledge and they applied it to
interesting problems like, let’s see, if we wanna go
to Mars, how are we gonna get people back and oh yeah,
remember the command module? The early iterations of the
command module, especially in Mercury, when they would
splash down and the Astronaut would open the hatch it
would fill up with water. One of the astronauts almost drowned. So, they didn’t want that
to happen again and again and again. So, how do we make sure
the command module doesn’t tip over and sink after it hits the water? So they went back to their
knowledge graph and said, hey, wait a minute. If we make the command
module more buoyant, we put the hatch further up
to the top of the cone and we give it uprighting
kind of balance capabilities then it won’t tip over. Then we won’t jeopardize
the safety of the astronaut at the last possible minute
when something could to wrong when they return home. So, the impact of this particular graph is pretty substantial. We’ve measured, David
Meza is quoted as saying, we saved probably two
years of investigative work on that exercise and well
over a million dollars of taxpayer funds in
solving just that problem. But, they continuously
use this knowledge graph over and over again as the look and plan for future missions. Talk about ICIJ, one of
our favorite, very public, very popular use cases customers. ICIJ of course, is the
International Consortium of Investigative Journalists. That’s the first acronym you
need to learn when you join neo4j. So, ICIJ is a team of about
300 journalists and what they’ve been working on, or
they periodically are working on are leaks of information,
usually financial record leaks, and their first major, major
reporting out of this was the Panama Papers in 2017, or 2016. And, Panama Papers of
course, figured out, it was a Snowden-Style leak,
like somebody walked out with a hard drive, 11 and a
half million documents on it. It had information like
people who own accounts at a bank and then it
figured out stuff like, well, those people live
at a particular address. They live with somebody
else, probably a spouse, who is an officer of an off-shore company, the holding company
with accounts at a bank in the Bahamas. So, simply tracing with these
six hops from the paycheck of a, of a popular professional
person, channel through their spouses actually
revealed where they’re hiding their taxable income in offshore accounts. That was the Panama Papers leak. As a graph it looks exactly like this. Person A with their
account, there’s my node, there’s my relationship
and you can see how I would traverse around this
to go from their income all the way to their, to their tax haven. Well, in 2017, ICIJ and
journalist from McClatchy, The Herald, New York Times,
they won the Pulitzer Price for investigative journalism
because until Brexit and the election, this was
the biggest story in 2016. And so, the other interesting
things that had happened with this though is that it
identified the holding accounts of celebrities, of gangsters
and of very high level politicians, including oligarchs,
including prime ministers. It brought down the Prime
Minister of Iceland, he resigned after this was revealed. It brought down last year the
Prime Minister of Pakistan. ICIJ’s investigations. Well guess what? They did it again. Paradise Papers, different
leak, instead of being the Panamanian law firm,
this is a 100 year old law firm called Appleby
and Appleby is even more documents, 13.4 million
records about where companies and people are putting their tax shelters. One of the interesting
ones that it revealed was, remember when Apple was
brought up in front of the EU about repaying their taxable
income in the European Union and at the time Apple had
held all of their corporate accounts in Dublin, Ireland. Well, what they did subsequently
is after the EU got ’em for something like $14
billion they moved all of their accounts to the
Isle of Jersey in the middle of the English Channel. The Isle of Jersey has a
population of about the size of this room, but very, very
favorable taxable income kind of, kind of regulations. So again, ICIJ did it again
and one of the really, the other cool things that
has gone on is they were recognized, not only did
they win the Pulitzer Prize last year, this past
year they were granted a million dollars endowment
from the Hollywood Foreign Press Association. Those are the Golden Globes people in order to keep their activities going. So, the Paradise Papers,
that’s kinda the interesting, human interest side of the story. The graph of Paradise
Papers is really simple. It’s just officers of a
company, they’re connected to something else and
connected to another entity which might be an address,
might be a different type of, an intermediate entity,
but that’s the structure, the metadata model of the Paradise Papers and Panama Papers leaks database. When you draw back the
lens on that this is what it looks like and you can
see Appleby is the big firm there in the lower right
hand side, but then there’s other clusters of
information that are revealed simply by putting all that
information in the graph and zooming in and out of the graph. Let’s take this idea of detecting
fraud and carry it forward a little bit more, and why
graphs are so appropriate for that. So often times traditional
fraud detection measures, they look at things like
you know, what’s the final end point of the transaction? Did the transaction match where I think that particular user is? So, remember your bank
is constantly calling you with hey, somebody in Cleveland
just ran your account. Is that you? It was only for $6, right. That’s the kind of end
point centric analysis. There’s navigation centric analysis. Is there any kind of
navigational behavior like, what’s the IP address that
that customer is using with that account? So, that’s always tracking
the cookie your bank is tracking the cookie of
your machine, like your mac address, your IP address, etc. Remember, it always says,
do you want me to remember this machine? That’s that kind of fraud detection. And then, account centric
is comparing you know, transactions and vetting identity, really kind of basic things, right. The bummer is it doesn’t
really examine things like rings of people, right,
coordinated operation amongst individuals, IP address
faking, hijacking devices, synthetic devices. So, if we apply the
connectedness model and do connected analysis then we
could look at things like cross channel, find the
anomalies and weird behavior of coordinated individuals. We could look at things like
linking entities, so let’s look at whether the relationships
between one account holder, another and what do they
have in common, what do they don’t have in common? Is that right? This is what that looks like. So, modeling fraud transactions
as an organized crime ring. At first glance I’ve got
my account holder, favorite credit card, they have a
bank account, they have an unsecured loan that
they’re all operating with it. Three people have set up
similar situations here. They have multiple types of accounts, etc. What makes it really
interesting when you model it as a graph and you look
at, hey, wait a minute. They have a common phone
number amongst them, so that could be location
identity information. They have a common Social Security number. That’s really a red flag,
amongst a couple of them. And, there’s a common address. So, they’re operating
out of the same location, perhaps the same apartment
and using similar information and then just dispersing
it amongst a whole variety of different accounts. Really, really easy to
understand types of identifying relationships amongst
people, the weird activity that they’re doing and
the networks that they’re doing it on. So, that was fraud. Let’s look at our other
favorite topic of the month. GDPR and tracking regulations
and activities across, this is again a financial
services kind of firm, but most of these are
either identity or financial traceability kind of
transactions and activities. So, if I’m looking at internal
risk models I have things like in a big financial services firm. I’ve got my trading desks,
I’ve got my fund managers. I’ve got the actual trades
and activities I’m performing, core banking activities like
loans and savings accounts, checking accounts and
then, everything that’s going on in the market. When I look at even just
tracing you know the risk lineage of the activities
that the trading desks are performing against the
baskets that they might be performing and operating
against, like my technology desk is trading in cloud computing. It’s got a lotta equities
associated with that and then I wanna see what’s
happening in the market there. So these activities and all
of this risk analysis here, what’s required by the
SEC is to be able to trace forward and backwards
all these different kinds of activities and how
they’re interrelated. So you can see, evaluate
the interrelatedness of all of these things in order to verify that. You know, the risk that they’re carrying and performing trades. They have enough equity
and trading capability, trading power to actually execute those. And another great one of
the role of ontologies. These are catalogs of information. FIBO, the Financial
Industry Business Ontology is a great one. Usually it’s performed with
RDF technology, triple stores, right, because they’re great
at organizing catalogs. Well, FIBO has a, as a
data model has about 13,000 triples in it associated with
it, but when we move that into neo4j and a property
graph while maintaining all the same fidelity that was
required of the FIBO model, we brought it down to 1300
nodes, so we shrunk the size and complexity of the RDF
store down to about a tenth the size of what it was previously. These operate against all
kinds of different, different regulations and different types
of financial trading models and financial management models. Let’s talk about knowledge
graphs for a second. Got a great customer, Scripps Networks. These are the HGTV, the Food
Network, Cooking Channel, Travel Channel, right, the
channels that you watch when you’re in a hotel room
or you know, late night TV and wanna see alright, where
are people moving, right. Where’s the Property Brothers, etc? Right, so developer of
lifestyle-oriented content for digital media, they
make a knowledge graph of all the media properties
that they happen to own and now they’re owned by
Discovery Networks who happens to be a local organization
here, Arlington. Is anyone from Discovery here? Just for you guys, just for you guys. Okay, so Scripps has this
content management environment where it’s figuring out not
only every one of the episodes of Rachel Ray, but it’s also
figuring out and tracking, what are the cameos that
Rachel Ray is making in other TV shows? And, where’s the, what’s the
most popular set of videos or YouTube content that’s
getting played off of these, off of these episodes? So they tag all the content
with who the characters are, what the legal rights to
each one of these are, what the original television
format might be, who owns the rights to that? Those kinda things are built
into the Scripps Network knowledge graph. And then I’ve got Airbnb’s data portal. Right, this knowledge graph
is actually a metadata model that’s taking all of the data
that Airbnb is collecting worldwide and integrating
all those different silos and looking at the kind of
reports that any one of their Airbnb’s employees are actually
running ’cause what they really wanna do is maintain
high degrees of consistency of how they’re reporting
activity like, is this a really popular property? Why isn’t this one reserved all the time? So, and then also who the
information, the individual information providers inside
of Airbnb that are really the go to people who have
the most accurate information and the most popular sets of reports, etc. That’s what the dataportal does. And so, it’s serving about
3000 plus individuals inside of Airbnb and it’s
a really, really simple knowledge graph of you know,
what things are popular, and what things are not
and it’s all based on their data elements that they’re tracking. All right our final bit
here is let’s talk about where this is going. And as David mentioned
earlier, digital transformation is effecting every kind
of organization, right. Driving new ideas and
innovation is really really important for an organization, especially if you’re a
retailer and Amazon is chasing you down, or if
you’re a taxi service or a travel service and
Airbnb and Uber are tracking you down, all right. Coming up with new and
clever ways to deliver a highly engaging customer
experience, recognizing exactly where they happen
to be at any given time, the activities that they’re
doing, those digital networks that they’re performing
activities on, is a real key initiative in many, many organizations. How many of you guys have
digital transformation type activities going
on in your organization? I got a couple, that’s good. There’ll be more. So, let’s look, so a couple
of really interesting things here and this is
highlighting how graph can actually be applied to
this transformation exercises and the reasons why that’s
actually a good thing. One is, really the Gartner quote. My customers told Gartner
Group the reason they chose neo4j was to drive innovation. Remember what I started out with. I keep going back to that
adjacent possibility idea. Driving innovation requires
to identify new adjacent possibilities. Those are graph terms. Graphs help you drive that. So, there’s all kinds
of other ways to look at digital transformation in
these kind of megatrends that are happening
whether it’s innovating on things like drug discovery
in the farming industry, performing hyper personalization
in kinds of activities for recommendations for
healthcare tracking for media advertising, etc. our
Facebook-style use case, improving decision making, improving
data integration, supporting AI initiatives. The knowledge graph is a
great example of supporting an AI initiative. So, let’s talk about one of
my other favorite use cases, eBay ShopBot. This is a conversational
commerce application that operates in Facebook
Messenger and actually has its own app now in
Google Play as well as the Apple Store. What ShopBot actually does
is it allows yo to log in to it and say, either talk to it, talk to it, say hey ShopBot,
I want you to find me a new product, right. You can speak it, you
can feed it a picture or you can type in your
particular set of responses and this ChatBot on the back
side of eBay is figuring out whether they actually have
products for you or not. Really kind of basic thing
and then they present it back to you, but they
also communicate with you and ask you contextualization
questions while you’re in conversation with them. So, if we look at kinda
the context of things, of information that
customer might provide, instead of saying, I’m, I
need a new backpack, right, is this example. So, contextually you
might type out or ask, can you find me a brown
leather coach messenger bag under $100? And look at the context
that this is actually identifying here. Right, color, material, brand,
what the object itself is and price point of what the object is. This is what eBay is parsing
out and actually remembering about any one of their
products for sale and they’re actually mapping that back
to whatever you’re actually requesting as you’re
conversing with ShopBot. It looks like this. So, here’s my bags are in
the middle there and the questions that its gonna
ask me are things like, okay, do you wanna bag? Do you want it to be a handbag or backpack and I’ll pick backpack. Then it’s gonna ask me,
all right, what color brown or black, or what
material leather or canvas. What’s you pass point? If it’ll go through that
and the really cool thing, This is where the AI and the
advanced kind of customer experience comes into play. The cool thing that
ShopBot is doing and again, they’re iterating on
this three month cycle. What they’re doing now
is they’re remembering contextualizations about the
customer, attributes like what’s your shoe size, right. Attributes like, I’m left
handed, right, so when I shop for sports equipment, right,
or scissors, I need stuff like that supports my hand
in this and I want ShopBot to actually remember
that when I bought the first baseman’s mitt, it
was a lefty, so that when I go and get golf clubs
soon, because you know, the masters was just on and
that was a really good masters this year, right. When I’m getting clubs I want
it to offer me lefty clubs. So, that’s the contextualization
that they’re doing now. What they’re going to be
doing later on this summer is adding even more context. Where are you located? What’s the climate like? What season is is? Is this a gift for somebody else? What’s the anniversary or holiday? So, it’s building out this
institutional knowledge, this contextualization
and then presenting that back to you as you shop with it. Their privacy controls are
actually quite strong here. Unlike the Facebook has been
watching my telephone calls for years. So, check out ShopBot. It’s very fun to play with. It is actually generating
revenue which is part of this transformational kind
of experience inside of, inside of eBay and we’ve
got a number of other similar types of use cases
like even as you’re operating inside of a, in your mobile
application, there’s so much contextualization that
you can actually derive and apply with any one of
these applications, right. And, the graph behind those
online stores, they have product information, customer
information, supply chain information, delivery information. All of those are built into
these variety of graphs. And as mentioned earlier,
Walmart is another one of these vendors who’s using neo4j. Let’s talk internet of things. Iot’s really big right now, right. Self-driving cars, smart
homes, xfinity xFi, a Philadelphia based project
here on the east coast. This is one of our
customers, presented a graph connected to it last year. xFI is the customer
experience that they, that Comcast is building to
integrate not only your router information like they know
who’s in the house at any given time, they’ll tell
you when your kids get home, stuff like that. They take the router information
and map it to your phone. They take the entertainment
information, your preference for shows that you wanna look
at and watch or who watches what types of programming
inside of your household,. All right, kids programming
versus, you know Game of Thrones, and then
they’re also mapping that up with security information of your house, whether you left your garage door up, whether somebody’s at the front door, whether odd things are
moving around, like you know, you never let the cat out. It’s tracking all of that
information and creating this very, very interesting
and you know, experiential kind of customer experience
under the xFi brand, right. This is what they’re quadruple
play or their triple play is really going for. And so, you can imagine
this is not only a customer mobile network device kind of environment. It really is the exemplification
of what you can do when you get into the ideas
around internet of things. Let’s look to the next set of technology. What’s comin’ up next? Blockchain. You know what Blockchain really is? Blockchain’s pretty easy to understand. It’s a public ledger. It’s tracking all the
transactions that might happen against anything and
then it distributes those transactions across a
networked environment so that everybody has the public record. It’s encrypted, so it’s
safe and you might not know the absolute identities of
everyone, but those transactions indeed, are available and
decentralized, so there’s no central certifying authority
that says these are not true transactions. You see this crypto currency
right now and that’s what most people are
associating Blockchain with, but this goes well beyond Crypto. This can actually handle
you know, your purchase transactions, your communications,
your things that you buy, etc., it can be used for
all kinds of different, different types of transactions
and different types of exchange eliminating
the need for you know, in real estate, all of
that paperwork you have to sign off on when you buy your house. So, that’s where graphs are going. Graphs are the foundation
for Blockchain ideas. Graphs are the foundation
for machine learning and deep neural networks. Guess what your favorite
neural network is. Right, it’s your brain, right. That’s a graph. So, we’re moving in that
direction and I’m starting to characterize this as the
third or forth generation of where graphs have been going. I’ve come from the origin of
the internet as a network, social networks, then into
the real-time recommendation space, into the metadata
management space, now into AI, deep neural networks,
Blockchain, internet of things, foundational technology for all of us. So, your assignment, this
concludes my prepared remarks. Your assignment today is of
course, connect with each other. Find an adjacent possibility. The great thing about this,
about the rest of this day, got awesome examples from
customers, from Dan, to plant in your head of, my
favorite thing about Dan’s presentation is just its headline. What’s Your Hundred Million Dollar Idea? Robbie’s actually next. He’s gonna show you how to assemble it. So, thank you very much
and I’m here all day so I would love to talk to
you and get to know you guys. (crowd clapping)

Leave a Reply

Your email address will not be published. Required fields are marked *

Back To Top