Xfinity xFi & User Personalization with Graphs — Jessica Lowing, Comcast

Xfinity xFi & User Personalization with Graphs — Jessica Lowing, Comcast


– Hi. My name is Jessica Lowing and I am a Technical Product manager at Comcast. Today I’m going to be walking you through my adventures in
personalizing the smart home. A little jeopardy. Smart home for 200. Who here has a connected device? Lots and lots. That’s an easy one. How many of you have ever fallen asleep with the lights on in another room? Most. How many of you have ever left the house wondering if the oven is
on, if you left windows open, did you leave the garage door open? Okay. Smart home for 600. What about leaving the house without taking an umbrella when
maybe you should have, or a jacket for that matter? Run out of toilet paper? Wondered if there’s chicken in the fridge while you’re at the grocery store? Smart home for 1000. Do you actually live in a smart home? Truth be told, if our
home was actually smart, it would do all this stuff
for us and be amazing and actually be smart, as smart as a little kid could tell you if you’re running out of toilet paper. There’s a lot of stuff that our homes could do for us today that they are not currently doing. The progress that we’ve made is that we’ve connected a bunch of stuff. That’s really it. In reality, we really
have a long way to go to make our homes truly
interactive and truly helpful. When it comes down to it, our homes are still not very smart. To help me illustrate this, I’d like to introduce you
to my BFFs River and Lily. Take the following example: turn off the lights in Lily’s room. Not so bad. Simple, right? A kid could do that, your smart home might be able to do that. Alexa, turn off the floor
lamp in Lily’s room. Lily’s floor lamp. Alexa has it covered. Start the washer and dryer
after this movie is over. Getting a little tricker. Probably can’t really
ask Alexa to do that. What about this one? Where are the girls? If you live in a giant house, dad wants to know where are the girls. Your smart home doesn’t know
that and cannot tell you that. Let’s look at an example: turn the lights off in my kid’s room. This really has a strong relationship to semantics and social relationships. This is actually a pretty
tricky thing to answer. Who is speaking, who are your
kids, where’s their room, do they have lights that you
can control in their room? There’s a lot of open questions that you don’t really know and again, let’s look at this from a
computer’s point of view because it still comes back to those semantic and social relationships. Who am I, who are my kids,
what is this room I talk about, what subset of devices in the room are lights that I can control? If we were in a truly smart home, you would be able to
speak naturally to it, it would know you and your family, it would understand the context of your real life and your world that you live in, and it would just do the right thing that you want it to do. As a parent, you want to
feel like you are safe, that you are comfortable,
you are in control of the sea of devices in your home, you’re connected to those
outside of your home either friends, your digital lives. The experience of a smart home should be easy and rewarding and not a burden where you have to really work hard to get your home to automate for you. This is a big problem that we’ve been looking at at Comcast for a long time. I want to tell you the story of how we go started and how
I ended up here today. For those of you that don’t know, Comcast is a large
telecommunications company. You might think that we
just do cable television, but we are also an ISP, we have products for home security, mobile phones, telephony, we’re the owner of NBC Universal, and we’ve got syndication
partners across North America where we go outside of
the Comcast footprint and provide cable service there as well. Back in 2013, we started a new team based out of Sunnyville, California, with part of the team in Philadelphia where Comcast is headquartered, to research and prototype
this smart home of the future. We named our team Jarvis for none other than Tony Stark’s smart home in Iron Man. I joined this team in 2013 as
a Technical Product Manager right as the team was forming in 2013. My Master’s from MIT’s AI Lab in Distributed Robotic Control Systems helped me bring insight
into this connected home and really the problem space
that we were working in. We started by taking a long,
hard look at the question, what would the connected home of the future really look like? We’re all guessing so
we gotta figure it out. We studied up. We based our prediction off of research coming out of
universities, trends in the market, products that different
companies were coming out with, and all of this led to a few core themes that we wanted to explore a little more. Connected Devices, where all of the devices in your home naturally can interact together and really all the devices
that you can plug in. Similarly, all the people in your home. How can we take advantage of those natural relationships that
exist in the real world and integrate those? Rich interfaces, this is voices in interface,
all screens on all devices, how can all of these be
used in a fluid manner that people would actually want to use in the smart home of the future. Lastly, automation to
control those devices. In order to gain insight
into this problem, starting with a set of
functional prototypes. We can explore the problem space and learn more about those core ideas. PorchCam on TV. When my doorbell rings, I’d like to see my PorchCam
on my X One television and be able to say to my
smart home, unlock the door. Where’s Waldo, or in
this case, where’s Lily? Being able to ask your
smart home where is Lily and see a map of Lily’s
coordinates within the city displayed easily on my TV which is the largest screen in my house. BYODevice. Take any device, bring it home, how can you quickly and easily integrate that into your
smart home of devices? Kidwatch. Tell me when my youngest daughter, well in this case they’re twins. One’s a little bit younger. Tell me when my youngest daughter comes home from school or
fails to arrive by 4pm. Jawbone Granny. For those of you that
don’t know about Jawbone, not like FitBits. They had to open APIs. Tell me when my grandma who doesn’t live at home with me haven’t moved very much so that we can plan a family walk towards dinner time. We held a series of internal
demos within Comcast where we demoed to senior
executive leadership and senior product management. The response was very positive, specifically now go build it. That is a different problem to solve. We have a lot of work to do. Again, back to deep introspection. This time, what is the gap
between prototype and production? We needed to evaluate the gap to see what services were
available at Comcast today, what services are available
in the market today, and what would it take to
really start to bridge that gap? The place to start was Xfinity Home. Xfinity home for those that don’t know is a home security product with a suite of devices and sensors
for managing your home and controlling those devices. Camera, lights, door locks, you name it. This makes no sense if you are going for a socially
embedded functional network. You’ve got a bunch of
other devices in your home and Xfinity Home really only supported just the subset of devices that could come off a truck with a technician. In 2015, we launched
works with Xfinity Home and this let you bring home devices and plug them into your Xfinity Home. Turn off all the lights could now include Your Philips Hue
and your Lutron dimmer. We were inching towards a smarter home. In 2016, we tackled automation by building the Xfinity rules engine. Think of it as Ift on steroids. It’s really powerful and
it’s been open sourced so I recommend you go to check this out. This now let us do things like, tell me when River comes home from school. Let’s see how we actually accomplish this within Xfinity home. I wanna know when the kids
come home from school so, let me set it up to get
a push notification, or an SMS when the door opens, or when the system becomes disarmed. Great, okay. No sweat. I’ve got two rules that I manage. All is good in the world. I live in a smart home. What happens when you also
add facial recognition or voice recognition to your smart home when we’re able to tell when specific devices join your home wifi network? No human in their right mind is gonna wanna manage this
number of device-based rules. Really when it comes down to it, device automation is
not the solution here. It’s where we’re at today
but it is not the solution. Again, we’ve built a
lot of connected stuff. In other words, we still haven’t really built a smart home yet. What were we trying to do in that example? This was telling me when River comes home. That same rule could be set off if it were the dog
walker or the babysitter. It’s not really saying anything
about River as a person. To support this, we
need a richer definition of what a person is within our systems. A person is not just an ID. A person is a set of relationships to personal information, to a specific house in a specific place,
people that they live with, people that they interact with commonly, devices that they use, and
services that they use. That’s the rich definition of a person that we started to use. Similarly, we need a
rich definition for home. What is a home? It is a physical building with a specific and unique topology. It’s in a specific
place, people live there, people come and go from that home, there’s devices that are
permanently in the home, devices that come and go from the home, and services that support those devices. Why have this rich definition? It’s because it’s about the people. That’s really what we’re trying to do. Mom didn’t wanna write a
rule about the door sensor. No. Mom wants to write a rule, tell me when my kid
comes home from school. That’s what really matters is my kid coming home from school. What we need to do is make people be the center of automation. Again, we’re back to modeling semantic and social relationships. Everyone’s family is unique and different. They are not all the same. People have inherently
richly connected lives when you think of the
relationships between people, devices, rooms, all
of this is by it’s nature, richly connected and unique
for each individual household. That brings us to the
Xfinity profile graph. This lets us model real life
relationships of our customers. It provides the context
to Xfinity applications so that they can provide a more personalized experience for these users. Before I get into the details, I wanna cover a few core ideas, a few design principles that we stuck with as we were figuring out
specifically what we should build. Big idea number one, there is a product need that we build a shared platform that can be
used by any Xfinity application. This is at the household level. This needs to be a shared platform that all applications, all interfaces can present the same set of information. Big idea number two, we need to model this set of relationships as a graph data structure. This data is richly
connected by it’s nature and the real value provided is the set of relationships between
these bits of information, between these people
and places devices etc. Knowing that our client will want to crawl those relationships. We as a platform service needed to provide an API data model in the form of a graph data structure. Big idea number three, this
is an implementation need, we need a native graph database. A bad design would be to support graph semantics on a relational or non-relational database with indexing. We don’t want giant
crawls through databases to take advantage of these relationships. In that structure, the
graph complexity will grow as the data grows. In a native graph, we
kind of get to bypass all of those problems. The Xfinity profile graph is built with GraphQL APIs on NEO4J. GraphQL’s a natural fit
for NEO4J in our case. There’s a benefit to us
as platform developers, as service developers, in that we get to build generic expressive APIs for our clients. To the client developers, it’s great because it’s
intuitive and flexible. We open up the ability to quarry specifically for the information that each of those different
applications are interested in and the specific traversal
through that graph is supported. Finally, to the end customer. This lets us have applications that can support unique
experiences for each customer based off of that same data service, but with unique information
for each of those households. Let’s take a look at this again. Turn off the lights in my kid’s room. It was kind of tricky before. It’s not so bad when you
look at it through GraphQL. It’s alright. We can build this and give
that to our applications. Now we’re starting to get somewhere. Let’s take a look at how this can surface to an end customer. This is traditionally how everybody here in this room manages their
home wifi network today. You write your network name and your password down on a sticky note and you give it to your friend
and that’s really it. You have no insight into your network and you’re kind of flying blind. There is zero connection
to the people in the home, different experiences
for different customers. It’s really kind of all the same thing. Giant blackbox. Xfinity XFI is Comcast’s
personalized wifi experience that is tailored to cater to
customers’ individual needs. XFI has introduced user profiles which let you name people in your home and assign devices to them. There’s also a concept
of guest and household which are inherit to every household. XFI let’s you manage your network in terms of the people in your home. That’s pretty novel. That’s very different from a sticky note where once they’re on the network, they’re on the network and you really just don’t know anything else. For each person, you can dig in and see what’s going on with their devices. You can pause all your kid’s
devices when it’s dinner time. That’s the people-based experience that mom is really after. Pause all the kid’s stuff ’cause they need to get their
butts into the dinner table. For each device, you can go in and see individual usage patterns
and pause individual devices. Overall, it gives you an easy way to see what’s going on and control devices at the household person and device level. You can find out more
about this going online. I did not work on the
XFI front end at all, but they’re a client application and hang tight at CES this year. We will be announcing the newest advances that we’ve
made in personalization across a variety of
different Xfinity products. I can’t talk about them today. I’m sorry, but stay tuned. We are working on very fun stuff. My team is here today. Come and talk to us. I am happy to talk to
you about our experience building the Xfinity profile graph and our experience integrating with XFI as a customer application. Now it is up to you. Take the leap, take this information back to your own company. Apply this to your own industry. If you knew more about your customers, how could you change the products that you are building today? I’m arguing that you can
build better products if you know your customers. You are just as empowered as I was and my team was back in 2013 to take a hard look at the state of the world and make predictions on where it’s going and explore, try stuff, prototype. Go do it! You can do this just the same as me. Lastly, connect with your users. Let your product fit naturally into their lives because every customer’s life is going to be unique and different. Go build great personalized products. That’s what’s getting us
closer to a smart home. You are empowered to do exactly that. Last but not least, we are always looking for
smart people to join our team. Shameless plug. Please come and talk to us
and thank you very much. (audience claps) Sure. If there’s any questions,
I’m happy to answer them. Alright, come and talk to me. Thank you.

Related Posts

Atlanta BeltLine releases bid for lighting, security cameras along trail

Atlanta BeltLine releases bid for lighting, security cameras along trail

FORT FORT BENNING FORT BENNING WAS FORT BENNING WAS ALSO FORT BENNING WAS ALSO REPORTEDLY FORT BENNING WAS ALSO REPORTEDLY

Rola TX-103 Hitch Bike Rack Review – etrailer.com

Today well be reviewing the Rola TX-103 Three Bike Carrier for inch and a quarter and two-inch hitches, part number

Leave a Reply

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