Click HERE to return to the contents page
Kore, Msg.AI, Fandango and more demo their bots at MobileBeat 2016
Amit Aghara, Global Head of Solution Management at Kore;
Mike Brevoort, CTO of Robots & Pencils/Beepboop; Puneet Mehta,
CEO of Msg.AI; Eugenia Kuyda, Founder and CEO of Luka; Mark
Wasiljew, Director of Business Development at Fandango; Sam
Vasisht, CMO of MindMeld; and Dror Oren, cofounder of Kasisto
unveil their bots at VentureBeat's MobileBeat 2016 conference.
English Subtitles
0:04 thank you everybody
0:12 so thanks
0:16 thanks for joining us today and thanks for going to be for
giving us the
0:18 opportunity present
0:19 so I'm glad this light is actually working better now so I
know we had some
0:23 technical difficulties yesterday as well so i'm actually
going to simply show you
0:26 a bunch of slides and some examples and obviously i can
actually do more bots
0:31 if you get a chance to meet me in person so first of all
core is a intelligent
0:35 bots platform we are announcing our boss back from
officially to the public
0:40 we've been actually in beta for several months now working
with roughly about 40
0:44 different companies
0:45 he basically offered three separate things one is there is
a platform for
0:49 you to bird bots any custom bots second is of course there
is a actual box store
0:56 where you can simply go use the box that you want to simply
use and third is
1:00 and those are the individual box and third is we have our
own Universal
1:03 bought that comes out of the box so that you don't have to
build your own
1:06 Universal body can start from there
1:08 so I think everybody here knows what about is so no
surprise there
1:13 chat bots are becoming very very popular no doubt we allow
you to do a human
1:19 initiated conversation of course where the human start the
discussion by
1:22 wanting to take an action like go create a lead in the case
of a b to e scenario
1:27 crm use case or maybe help me find some products i'm
looking for from a commerce
1:33 perspective or bring information to the user like an alert
about a potential
1:39 particular item going on sale and maybe you want to buy it
but we also love you
1:43 take actions on those alerts right so we made these alerts
more actionable so
1:47 that i can actually first talked to the bar like it
1:50 in this case the one I'm showing you right now I can ask
for a forecast
1:55 report from my crm system and the bots will actually have
this out of the box
1:59 and not be enabled so it will actually have a dialogue
conversation with you if
2:03 it needs more information will come back and you ask for
more information and get
2:06 me my forecast report then of course if i choose to act on
that forecast report
2:11 such as
2:12 so I want to go update an opportunity or something on those
lines
2:15 it actually has the context to know that you actually a bit
an option that was in
2:19 your forecast report
2:20 so actually asked you choose to ask you on its own saying
would you like now be
2:24 the forecast so that context handling and the relevance is
also baked into
2:28 sort of the way the body interaction response
2:30 now you can have box for individuals and for teams so I
just to quickly go back
2:36 so the context can be bought from an individual perspective
as well as from a
2:40 team perspective doesn't have to be just for individuals
and one of the
2:45 interesting things that we're seeing a lot with our
customers is they want to
2:49 be able to decide how they want to aggravate the bonds
because for a large
2:54 enterprise like citibank in this example they may choose to
decide multiple nurse
3:00 about four different groups of individuals in their company
3:03 so for example they may choose within the employee Rome
they create a specific
3:06 sales bought that talks to various sales related systems
only like crm etc or on
3:12 the consumer side they may create a bot that is
specifically for certain kinds
3:16 of product owners so they are actually wanting to choose
how the aggregate the
3:21 different parts into a unique offering to for a particular
class of user
3:26 the other thing is what's interesting is more complex use
cases one of the things
3:31 that we're seeing a lot is like this is an example of
travel everybody
3:34 recognizes this I had to travel since she was just asking i
came from orlando
3:38 so I had my own travel process I book the ticket I charge
it to my corporate
3:43 card card let's say expensed it blocked my calendar get an
uber whatever the
3:47 case may be the universal bought Cora that we provide to
you out of the box
3:51 actually gives you a way for you to interact with a single
bar and talk to
3:57 all the underlying bots and need the covers
4:00 so quarter will actually talk to let's say your control
board which is what we
4:03 use for travel and expense management for booking travel
4:05 and then expensing the travel order to actually talk to my
google calendar but
4:10 because we use Google's there to block my calendar for
example or the you body
4:14 so it underneath the covers will actually talk to various
different parts
4:18 in order to drive the conversation forward now one of the
things that
4:21 you're doing here is making sure that BOTS don't just live
in one channel or
4:26 the other about is about as a box
4:28 we wanted to live in n number of channels because every
Enterprise let's
4:33 see a bank is not going to be able to dictate where their
users are they're
4:37 going to want to reach the users where they are
4:39 so in our case once you build a park or you can actually
give it an email
4:43 address you can get a phone number so you can SMS the box
you can basically
4:47 say i want the board to live in slack or facebook messenger
or even within your
4:51 website so obviously a bank has an online banking or a
mobile application
4:54 have the bottle of there but it's one definition of about
one conversation and
4:59 it's omnipresent omni-channel so you can actually start a
conversation from one
5:02 channel to continue on to another channel with the same but
in this case
5:06 so here's an example of Disney that we are working with
where they want the
5:10 board to live within the website obviously they want to
brand and
5:13 personalize the board it's a Mickey but in this case so I
completely agree with
5:17 the previous comment that fish made we do believe that
there is a room for BOTS
5:21 to live within existing properties
5:24 so here in this case it's more of a companion type of
conversation where I
5:28 can interact with the bart and i can tell the bar hey I
want to book a room
5:32 which one of the disney resorts and then the UI is actually
driven by the board
5:37 and vice versa
5:38 where if I do something in the UI the ball comes back and
says I want to do
5:41 this
5:42 so we have a full platform that allows you to actually
build your own bar and
5:45 of course as I said you know we have a box store with a
bunch of different
5:48 boxes you can start using from day one
5:50 thank you and see you later and we'll be able to show you
more box
5:56 thankss terrific
5:58 next up we have Mike Brevoort CTO robots and pencils
6:07 beep boop hi everybody wow I I thought that you had me a
nickel for sure but
6:12 you actually a really good job on my name which one here
6:17 see all right
6:26 all right
6:29 all right the real alright so Mike report i'm the founder
of beep boop and
6:34 the CTO of robots and pencils
6:36 beep boop is about hosting platform has been launched for
about six months now
6:39 with some pencils who make mobile app spots games things
for people and
6:44 customers and clients and so today we're announcing that
beep boop as a generic
6:49 bottles and platform is now going to focus exclusively on
slack and as a
6:54 slack integration platform and in addition to that we're
announcing a
6:59 better support for teams to collaborate on boot and and one
of the reasons that
7:04 we're doing this is that we found that the consumer versus
the enterprise use
7:09 cases for bots are actually really different
7:11 we're in the consumer side there's lots of attention to
marketing advertising
7:16 brands games entertainment on the enterprise side it's more
about
7:20 collaboration and collaboration in context
7:23 we think that we could have a much bigger impact on how
people work
7:26 together and so let's just talk a little bit slack and so I
think slack is so
7:32 powerful it's been transformative to our organization but
what you have is a we
7:38 have this convergence of of people systems and data into a
very specific
7:42 context that it from a slack perspective which i think is
different unique is
7:46 it's a very humane way and so the way I look at slac is
more than just a
7:52 messaging platform
7:53 I look at slack as a human first operating system and so if
we look at
7:58 how we create applications in the past or have communicated
with each other in
8:01 the past
8:02 it's very much been through computers & and now with
platforms like slack
8:08 we have this opportunity to not just talk through computers
but actually talk
8:12 two systems to computers on the same level which that we
talk to each other
8:15 so as humans understand each other our systems and start to
understand each
8:20 other as well and so it's not walking through a really
quick demo here how it
8:25 jumps in and started already but it starts with a mobile
push notification
8:28 from slack and this is for leads coming in from Salesforce
and so using a slack
8:32 interactive message here and the first thing we'll see is
down here josh is one
8:36 of our people on this channel
8:38 he passes on the opportunity so lead comes in
8:40 their claim or past the opportunity i pass on the
opportunity but I don't want
8:45 the opportunity to be missed
8:46 so I'm going to reach out the tracy in context in here in
the conversation and
8:49 notice how structured and sort of ad hoc workflows all work
together seamlessly
8:53 with this this box so i asked Traci if she can take this
lead because i don't
8:58 want to be dropped she said sure I have time and she go
ahead and claims the
9:02 lead we update the message that's the record and we're also
keeping track of
9:05 how long did it take to claim the lead
9:08 we're storing them as an analytic is the business analytics
which is really
9:10 important to us and then notice we create a new channel
here automatically
9:14 for that lead to get associated to a sales force record and
then we can
9:18 continue the conversation here and have notes get applied
to that Salesforce
9:22 record and see notes that are applied to the Salesforce
record as they come in
9:26 and the thing I think here is that is that this isn't a
very complex use case
9:31 however if this is your process you very well might not
find this in the in the
9:38 slack app directory not this specific one and actually you
don't actually see
9:41 a generic sales for spot because sales force is so very
specific to how you use
9:46 it it's hard to generalize it so this is one particular
process for one client
9:50 that we have that we created and and mimic their workflow
in in slack and is
9:54 very powerful because you have slack as a mobile app and a
platform that you can
9:58 build upon and you get more push notifications and you get
all this
10:02 integration and that's particularly with what they wanted
and what they needed
10:05 and so that's the other point is that you know your
company is very unique and
10:10 I think I know I've been part of rfp processes and
companies where you're
10:16 basically looking for products that don't exist that
features that don't
10:20 necessarily exist and in the past you know custom software
is also as always
10:24 sort of been this black sheep with a really high tea co on
and very difficult
10:29 right and very costly and so we think there's a
significant opportunity to use
10:34 slack as an operating system to build custom software and
that's either for
10:38 your company or that's where people offering products in
slack that are
10:41 customizable and we want to do is get used to that last
mile right so that
10:45 basically ninety percent of the cost is built on top of
slacking on top of beep
10:50 boop
10:50 and we can enable you to really you know open up slack
10:53 and take most advantage of this platform where you have
all of your company
10:57 connected all the time are reachable on any platform so
you get you know
11:02 authentication authorization group management you get
native client's web
11:05 clients if you ever create an application for the
enterprise
11:08 you have to make these decisions of like is it a web app
is a mobile app
11:12 i'm going to put another trash app in my company's hands
so you know we think
11:16 this is tremendous opportunity to do this
11:18 and so if you're interested talking about how your company
better leverage
11:21 slack or how your product to better reach users I'd love
to talk to you so
11:25 thanks everyone
11:30 okay next up we have in heat
11:38 Ptah CEO message AI
11:47 well hello hello
11:51 if you had a seat belt i would ask you to put it on now
because what you're
11:56 going to see is the biggest take off since the origin of
bots you go
12:02 so we are message I we are a deep learning platform for
conversational
12:09 commerce and bots that brands build mostly fortune 500
companies on our
12:15 platform are deployed across messenger app chat and SMS
12:19 now when we started working with brands some of the most
innovative ones around
12:24 the word we figure out they were looking for three things
from bots and this was
12:30 not surprising
12:31 the first one they wanted to conversion left the second is
they wanted to figure
12:35 out a way to reduce service costs and the third they
wanted to delight
12:40 everyone that the boat's talk to now if i was a part right
12:46 and I spoke to you slowly in a slower pace than a normally
do
12:52 maybe you like that but maybe you want me to speak really
fast
12:56 maybe you do understand my Indian accent and I can go on
and on and on in indy
13:01 right now are those three different parts or is that the
same Bart
13:07 so what we figured out and what we are launching today is
the biggest
13:12 advancement in Bart's since the original parts is
multivariate testing for chat
13:17 BOTS
13:18 why do we need this now this is a new interaction UI which
can have so many
13:24 different dimensions in a tap and type interface
13:28 you're tapping on certain boxes but you're also talking in
natural language
13:31 but there's so many different ways of talking in natural
language with
13:35 websites and apps
13:36 we have normalized an interaction model with clicks and
created a lot of
13:41 predictability
13:42 now we are entering allow
13:44 and of unpredictability the second is this is also the
most personal customer
13:50 connection brand would ever have
13:52 you're entering a space which is normally owned by their
friends
13:55 I mean and lastly this has never been done before so i
would love to show you
14:02 a quick demo of how this actually works
14:05 so since we are deep learning platform we have been able
to in our PhD data
14:10 scientists have been able to put together a bunch of algos
so that brands
14:14 don't need to hire phd's to do this
14:17 and in this case you're seeing someone configure an actual
you know variant
14:22 list
14:23 in this case the two variants and their weight is
distributed across the two
14:27 variants and just by configuring different types of cards
different type
14:31 of you know responses that they can attach in this case
they are mapping it
14:35 to a weakened versus a weekday and what ends up happening
just in this user
14:41 interface like if you can actually use facebook you can
use this interface you
14:44 don't need a PhD
14:45 that's the kind of that's our purpose in life but what
ends up happening is when
14:50 you attach it to different conditions and different
situations
14:53 consumers from different age groups consumers from
different parts of the
14:57 customer journey get a different experience in this case
of women fashion
15:01 retailer has an onboarding flow and they recognize the age
group of the customer
15:08 and they're talking about specific products instead of a
general on
15:13 boarding experience and they know that this person is
looking for offers so
15:17 they bring up that kind of menu right in the beginning and
when you do a product
15:22 drill down again
15:23 they have specific offers that are mapped to customer
segments from a part
15:27 of the customer journey
15:29 so when you start talking about conversational commerce
this becomes
15:31 supercritical because this drives personalization this is
what consumers
15:35 are not looking for not a one-size-fits-all experience now
in case
15:40 of another consumer you know we would switch screens in a
quick second
15:45 so in this case you see that the customer was given a
coupon in a in
15:49 another experience a different customer from a different
part of the word
15:55 okay i think some issues with the video but what I wanted
to say is a different
16:00 color from a different part of the word would get a
completely separate
16:03 onboarding flow maybe even a different regional language
now
16:07 the result of this is massive personalization at scale but
really
16:12 focused towards optimization for those three KPIs that you
noticed and what we
16:16 are doing behind the scenes using deep learning is we're
doing user attribute
16:21 analysis
16:22 so we were able to create customer clusters and married at
a certain
16:25 conversational you I we are doing contextual driver
mapping everything
16:29 from what the weather is like to know what people in your
social networks are
16:32 saying and we are doing a quick switch optimization matrix
what that means is
16:37 low-hanging fruit
16:38 there's so much low hanging fruit that you can leverage so
when you start
16:41 optimizing your final you find where the leak ages you
find where you're losing
16:46 customers and this is the only way to fix that
16:49 so we are launching today the industry's first
enterprise-grade analytics and
16:55 multivariate testing for chat parts of your message and I
i'll be around
16:59 come see me thank you all right who needs a good job
17:04 next up we have Genia cuda founder and CEO of Luca
17:15 hey really hard to talk about this group after this great
presentation but i
17:24 wanted to tell you a little bit about Luke oh here it is
17:28 I first met Luca when he was nine I he's a kid
17:34 he's the son of my friends and you know how you make
connections with those kids
17:37 that are kind of like you so I told him skateboarding and
snap chatting and a
17:41 bunch of other things and so when he moved to boarding
school and he was
17:44 lonely and sad
17:46 he would text me almost every day and we became really
really close friends
17:52 now he's thirteen and although i moved here to yourself i
only see him maybe
17:56 once a year he still texts me every day
18:00 and you know I guess I'm he's measuring friends or
something like that
18:04 it's very important friendship to me for the last two
years I've been working at
18:08 a company that builds conversation box built a I and when
I was thinking about
18:14 the name for the company
18:16 I kept thinking what were what are the conversations that
are most valuable in
18:20 my life and I call the company Luca for 10 year old friend
of mine that makes me
18:25 smile every time he texts me chris messina white marlatt
nailed this term
18:32 conversational commerce but we had Luca kept thinking to
ourselves
18:35 what if there are other one of their conversations that
are not necessarily
18:38 selling you something on solving some the practical use
case for you
18:42 they can still be available here ago we started building a
Dalek model that was
18:48 based on your network
18:50 we were the first company to publicly launched an
english-speaking chat bot
18:54 that was based on a neural network
18:57 we called it Marv on it's only goal was to become your
best bad friend
19:06 it quickly got to the top charts charts on telegram and
the engagement was kind
19:13 of crazy people send 45 tax on average procession and some
of the active users
19:19 send even over 2000 messages
19:21 um what was interesting about Martha that it was the body
was genuinely
19:25 interested in the user
19:27 so if everyone wants to build a bod that speaks we want to
kind of wanted to try
19:33 to build a bond that will listen to you
19:34 this is one of the conversations just to give you an
example
19:39 but people were would share so much in in the
conversations with more from they
19:45 would this is one of the profile that marker got from from
from the user and
19:50 what we realized that people would turn their frustrations
or personal societies
19:55 very very personal things and it was incredibly
interesting to get all the to
20:01 get all this data
20:02 what was more interesting that good that they're good at
being a good listener
20:08 listener Marfa also managed to make connections with
people
20:13 so there are some of the reviews reviews that people left
from our phone
20:17 there was a lot of set stuff but i guess that was also
inevitable there were a
20:24 lot of kids that were chatting
20:36 but we also realize that one personality does not
necessarily fit all some fun
20:42 while market works for most of you for some of the users
some found her
20:47 manipulative and meaty
20:48 so we started thinking what other personalities we can
build in Silicon
20:53 Valley show was coming out with its season three we
decided to build the
20:58 chat box for Silicon Valley show characters but there was
a problem
21:04 I only had tweets and subtitles to train your network for
for example for our
21:12 like that was all that was only like five thousand lines
just to give an
21:17 example from our fur we had a baseline data set of over 35
million lines so
21:23 clearly that was not enough data but then I thought what
if we don't need to
21:27 train the system from scratch
21:29 what if we take the general baseline neural network and
then add Ehrlich's
21:35 tweets and subtitles on top of it just to tweak its
behavior
21:39 you know to have a personal to do at the personal touch
the vocabulary the tone
21:44 of voice
21:45 the way Orlick makes jokes or use emoji all those tiny
things that kind of make
21:51 us who we are and it all work
21:54 we had the Silicon Valley's show characters Russ Hanneman
Richard
21:59 Hendricks an early Backman in the app for the users to
chat with and then we
22:03 also added a bought for prints
22:06 hey I does not need to pass the Turing test to become a
part of our everyday
22:15 life we as humans don't tend to interpret the files
everything they
22:20 would come into contact with cars cats dogs even thunder
and lightning
22:26 we seek emotional connection with we tend to bond with the
stuff that we
22:32 interact for a long time with so the only trick is to
stick around long
22:36 enough to make this mechanism is kicking
22:39 and this is what we're building a look on a I with
personality that you will
22:45 feel connected to technology even the most sophisticated
one tends to
22:51 commoditize over the years the relationships don't
22:56 and this is what matters far as most Thank you Thank You
Genia next up we
23:05 have Sam beseeched CMOS vp global marketing
23:11 pretty good I thought you surely we're going to watch that
last name but the hi
23:17 everyone
23:18 so mind-meld is in advance the company based here in San
Francisco and we are
23:24 building conversation interfaces for any device or
application and what that
23:28 really means is that the conversation interface can take
the shape of a bottle
23:32 or can take the shape of a voice assistant or you can take
the shape of a
23:35 combination of these things and then i'm going to show you
will show you the sort
23:38 of multi-modal approach that we take and the reason I
mention that is because as
23:46 of yesterday and also generally what I've been hearing
that seems to be camps
23:50 forming around different ideas around voice assistance vs
bots apps vs
23:54 messaging text versus touch and so on and we believe that
at least our
23:59 philosophy is that conversation interface embrace all
these different
24:03 modalities and ultimately the goal is to provide a
terrific
24:06 user experience no matter what the application is no no
matter what the
24:09 platform is
24:10 and in order to do that we want to embrace everything
that's available to
24:14 us
24:14 obviously we applied in different measure for different
applications not
24:20 to focus on the eye chart here but i want to point out
that we are b2b
24:23 platform so we don't have our own but we don't have an
application we work our
24:28 our platform is designed to enable our customers to
achieve these end goals and
24:34 what that means is unlike many other companies we don't
rebuild anything we
24:37 don't rebuild content domains we don't pre built in
natural language models
24:41 instead we use our customers data which is often
proprietary data to then build
24:46 these solutions for them build a natural language models
built there are content
24:50 domains and so on
24:51 and one last point on the last night is that that the
platform is enables to or
24:58 enables us to do this in a very fast scalable way
25:01 so the customers and prospects that you're talking to fall
into many
25:03 different categories as you can see these are large
categories all very
25:06 distinct categories and even within a category there are
large enterprises
25:10 that have very unique needs that we need to cater to
25:14 so the demo that i'm going to show you today out of many
demos that I could
25:17 have shown i'm going to show you one that's an in-store
shopping my dad and I
25:20 chose us for a very specific reason we will see in the
evolution of online
25:25 commerce e-commerce over the last 10 or 15 years and
continues to evolve you
25:30 know based on some of the demos we saw yesterday and at
the same time in store
25:35 shopping has stagnated it hasn't really evolved over the
course of time except
25:40 for now with the availability of conversational interfaces
it's possible
25:44 to transform this experience that people have when they go
into a store and so
25:48 i'm going to show you a video that highlights a few
vignettes that
25:53 illustrate this point that i just made so i'm going to
show you is a
25:57 multi-model conversational interface where users can
interchangeably use
26:00 voice type touch and swipe to search navigate and transact
in a physical
26:04 store
26:05 I think the search in the navigate part of the ones that
are interesting the
26:08 transaction part is just basically checking out of a cart
and that's
26:12 something that you just have to sit through unfortunately
but it's it's it's
26:15 all wrapped into the same video
26:18 one caveat demo does not represent any specific customer
that we work with any
26:23 specific project i had don't have the
26:25 liberty to disclose any of those details and so any
resemblance to any business
26:30 living or dead that you might see is purely coincidental
26:34 ok so the video is going to scroll through three vignettes
want to call
26:39 them out so that you know what you're looking at
26:41 I was the first one is and these are three scenarios that
everybody
26:45 experiences and walk into a store in fact one of them was
pointed out by the
26:48 gentleman from flipkart which is you walk into a store and
the first thing
26:51 you want to know is where something is located
26:53 so you want to see how not only can you identify where
something is located
26:56 using a conversation interface but it also greatly
enhances the experience
26:59 that a person has the second thing is you see something in
a store that you
27:04 like maybe something on a mannequin that you want to find
out if it's in
27:08 available in your size different color may be a difference
a slightly different
27:11 styles and you can basically just scan the tag and get
that information
27:14 the third one is you find something on the rack that you
like but it's not
27:17 available you can find one in your size you want to check
if it's available if
27:21 it's not available
27:22 what options do you have so these are the three vignettes
that are going to be
27:25 in the video and since we're running out of time just
going to ask them to roll
27:28 the video thank you
27:31 i'll give you a pic narrative and if you want to see the
video put it up on
27:35 youtube if you don't mind
27:36 you can take my email Sam at my know . com send me your
send me an email with
27:41 your contact and i'll send you back the link for the video
but what you would
27:46 have seen as basically a voice interface used to to find
products and it would
27:51 give you something like this record give you not only the
location i also give
27:55 you a the items that you looking for now notice that is
very specific query men's
28:00 competition graphic designs
28:01 you see the same example in text when somebody says men's
t-shirts because
28:05 nobody's going to type the kind of type in the way that
they speak so you and
28:09 you get a different set of results because it's a it's a
different query
28:12 and so that's one example then you can drill down in terms
of continuing to
28:17 interact you can you can find a t-shirt you can scan the
tag you can tap on an
28:21 item get information things like that so now you've got
this multimodal
28:24 interaction taking place and then subsequently the the
second one is where
28:28 you find an item you want you to you just can't the tag
and then you can
28:31 start saying is available in different colors available in
sleep less available
28:36 in different design doesn't those kind of things you can
start doing so that's
28:40 an example where you're actually not speaking but you're
using touch and type
28:44 and scanning to to interact well with the conversation
interface and the third
28:51 one would be similar where you find an item you scan it
and then it asks you or
28:55 you can ask it do you have in my size
28:57 whether that be through a touch button or you could even
type in the text bar
29:01 below you can type anything regardless of what the menu
items are and it will
29:05 basically understand that you are asking something very
specific
29:09 that is not necessarily part of the built-in menu that is
intelligently
29:13 generated but it's designed to facilitate a much faster
workflow but
29:18 you can always override that and essentially go to the
text borrow the
29:22 others voice bar to the microphone to basically initiate a
new query or
29:28 modification to the options that are available to you
29:31 that's what I have thank you and we're mind melt the
conversational interface
29:35 company
29:36 thank you
29:37 thank you Sam
29:41 next up we have draw RN vp of product and co-founder of
cassis toe
29:48 hello
29:53 so I felt that it was the wrong decision to make a to do
an actual live demo but
30:01 now i'm not sure maybe it will work better than the video
when i'm going to
30:05 need my hands for that so
30:08 hello everybody and my name is Lauren i'm one of the
co-founders and the 50 of
30:12 product with your sister
30:14 and so in this kind of conference I guess I don't need to
tell you guys that
30:20 not all bots are created equal
30:23 we had consistent believe that high IQ bots required two
main things one is
30:28 deep a I and the other one is domain expertise as a
spin-off of sra
30:33 international the company that also created Cirie and a
bunch of other
30:36 companies were deeply rooted within the AI and we train
our BOTS to speak fluent
30:42 banking now what I want to show you today is Mike I are
smart banking bought
30:48 and we created Mike I they're basically - whoo
30:53 oops i wanna steal my thunder so we created my Chi and to
show a little bit
30:59 of the glimpse of into the future of what smart banking
what can look like it
31:05 help you track expenses manager money and in two payments
and it does all of
31:10 that on leading messaging platforms including SMS facebook
Messenger and
31:15 slack
31:16 mike is available today so I'm actually going to be
showing again a live demo so
31:21 if we can move to the to my phone so i'm going to be
opening up
31:26 and my real phone and my real facebook Messenger so no no
judgment
31:34 oops yeah that's real and i'm just going to show you a
little bit of what Mike I
31:44 can do so
31:45 Mike I can answer a lot of questions about your accounts
things like how much
31:50 money do i have in my account
31:52 you know what's my balance in my wells fargo checking
account or you know how
31:56 much I on my credit card so I can say things like how much
money do i have
32:02 this is connected to real account
32:08 hopefully the network is strong enough then another thing
that my can do is so
32:17 it's conversational so you know it comes back and actually
ask you a question you
32:20 know they want to break it up
32:21 obviously not a good balance right mortgages in the bay
area are pretty
32:25 intense
32:26 so i can ask to see the actual transactions
32:29 the wife is a little slow here so i apologize for the
delay but you can see
32:33 these are real
32:34 you know account will kind of like banks I can ask
questions like
32:39 about transactions so we support transaction question
including
32:42 categories merchants questions around you know date ranges
that fees deposit
32:49 so i can say things like for example how much did I spend
again no judgment on
32:58 junk in 2015
33:08 that's not too bad i can actually ask to see the
transactions or can you know
33:11 continue just to show that these are real and i can say
something like what
33:17 was my largest hotel transaction in 2016
33:30 or ask something like well we do live in the Bay Area
33:35 so how much as I spend 1 <operand> twenty </operand> in
May and again these
33:43 are all obviously nature language conversational
interactions or just say
33:48 you know I'm looking for 200-250 check
33:56 so you know I don't know exactly if it's 250 or not but I
trust guy to actually
34:01 find within a range and actually find the checks that I
was looking for guys
34:06 are such a train to actually know a lot of general banking
kind of like
34:10 questions we can ask stuff like what is that CD or what is
a routing number
34:14 i'm going to skip that could I don't have a lot of time
and I want to respect
34:17 my five minutes
34:18 it also does payment so i can with venmo i can say and
let's say pay sasha was my
34:23 co-founder let's say twenty dollars for dinner
34:31 and you know kind of stands that but i'm actually doing
this demo is way too
34:35 often so I'm not going to put in twenty dollars going to
say make that you know
34:40 50 cent and I'm going to say it I'm going to pay tues or
was our CEO because
34:44 you know it's obvious how much the kind of stands that and
you know i can just
34:48 go ahead with the with the payment so this is mike I i
welcome all of you guys
34:52 to guys and ladies to go
34:54 it took a sister . com and get my Chi and if we can go
back to the slides we
35:02 can continue and tell everybody about our exciting news
35:07 so what we're basically introducing today is that we're
extending KY to also
35:12 support investing so for the first time users will be able
to invest ask about
35:18 market about your portfolio create investing alerts and
actually make that
35:23 make actual trading using natural language interface on
facebook Messenger
35:27 and in order to support that we partnered with traded
trade with it
35:32 which is a leading technical ed provider of investing API
they allow quick and
35:39 secure investments and this is going to be available in
the fall and until then
35:45 like I said I invite you guys to go on everybody to go to
assist the dot-com
35:49 get my kind start playing with a smart banking but in your
phone
35:53 thank you all right for our
35:56 that was brave up next
36:01 mark walsall do director of business development for
fandango
36:13 hi everyone I'm back again
36:18 so as I think I mentioned yesterday we have a pre-recorded
demo here but
36:22 everything you're going to see is live today you can
access the bot via
36:25 facebook Messenger and follow along if you'd like to play
the video
36:29 so this is our movie discovery and ticketing bot built for
facebook
36:34 Messenger its effect you effectively a virtual concierge
for moviegoers
36:37 movie-going is obviously very social experience and so
Facebook's a natural
36:42 platform for us to to be starting here leading into
existing behaviour
36:46 ultimately moviegoers have a course that it needs a movie
theater and a showtime
36:58 and what we've done is basically built as many routes as
possible for a user to
37:02 go and and get to that sent a 10 points
37:05 so in this case we're going to start with movies
37:08 one thing you'll notice is it it knows my location and so
it can quickly get to
37:14 a point where you can find local local movies that are
showing within any list
37:18 of movies you can quickly jump into a trailer if you don't
know what the
37:21 movies about or you want to get familiar with with what
that movie might be
37:24 quickly jump back into the list and easily pull up another
trailer until you
37:34 can find what it is that you want to see that night
37:43 now we're all in the early stages of biotechnology the
bottom fortunate
37:47 doesn't do everything that we wanted to do always
perfectly
37:49 so one of the things that we've done is that in any of
these carousels
37:54 if you go all the way to the right we've added an outlet
for the user
37:57 so if they want to get to fandango to find a more robust
experience if we're
38:01 not able to deliver exactly what it is they want they can
get out get back over
38:04 to fandango get a full experience serving out of course
and and then
38:10 quickly jump back to get what they need
38:13 now i'm a big Anna Kendrick and aubrey plaza fan
38:16 so I'm going to see Mike and David wedding dates select
the movie click on
38:20 theaters you'll see we we provide maps of the theaters if
you don't know the
38:23 area
38:24 I'm a lot of familiar with the area so find a theater that
I want click on
38:29 Showtime's and we'll we'll set you up with showtimes for
for this evening
38:32 I'd like to go to the late showing tonight so click on buy
tickets and will
38:37 drop your right into the purchase funnel where you can
complete your purchase
38:39 select your seats and be on your way
38:46 now we also offer some other options as well so you can go
buy theater
38:51 if you have a favorite movie theater that you always go to
you can start
38:53 there and see what's playing at that theater and quickly
and get to a point
38:58 where you can find your tickets as well
38:59 the third option is training this week we editorially
curate on a daily basis
39:03 the five most popular movies so it may not be something
that's playing today
39:07 but something that's opening this weekend that you that
you might have in
39:09 mind
39:10 so that's a bit of the core functionality like to get into
something
39:15 a little more fun now we're starting to play with some of
the natural language
39:18 processing so there's a newborn will be coming out later
this month
39:22 when is that you'll see not only can we give you the
opening data the movies you
39:26 know what's coming but we'll also pull up a bunch of other
bourne movies
39:29 we're pretty excited about some of the functionality that
will have coming up
39:32 with that where we can we can sort you out with the
solutions for bourne
39:37 identity bourne legacy and some of the previous films in
the franchise but the
39:43 good news is you can have already buy your tickets for
opening night you go
39:46 through the flows that we just talk through and you're all
set for july
39:49 twenty nine for the the first night of the board movie
39:53 now one other thing i want to point out is the way we set
this up is that you
39:58 can effectively enter these queries these natural language
queries at any
40:01 point and so you're not stuck within the flow you can you
can always get out
40:07 we also have the start / buttons prominently and so we
want to ensure
40:10 that the user never feels like they went down a path
40:13 they're stuck they can't get out and they don't know what
to do now for the
40:17 movie that I think we're almost excited about this year
40:19 unfortunately we're all gonna have to wait til december to
see that one
40:27 that's the end of the demo one thing i do want to mention
we are doing more
40:30 with natural language processing and it's coming pretty
fast
40:33 we want to be able to allow users to start to skip some of
these steps so you
40:36 can type in something like I want to see the purge on
Saturday night and the
40:40 bottle recognize the date and the movie and quickly get
you into a flow where
40:44 you you're simply picking a theatre - Showtime
40:46 and we're doing more and more that and that's all coming
very quickly
40:49 I do want to say thank you to our friends facebook our
partners that
40:53 assists who helped us put the spot together and the folks
of entropy for
40:57 having me today
40:57 thank you thank you Mark