- il y a 2 semaines
What You’re Looking For: How AI Is Changing Search Engines
Catégorie
🤖
TechnologieTranscription
00:00Hi everyone, having a good show?
00:03Okay, someone's having a good show.
00:06So, welcome to our panel.
00:09We're going to talk about search in the age of AI
00:14and how maybe things are changing there.
00:17You know, search was once thought to be a solved problem
00:21some times ago, and then
00:24Sergei Brin and Larry Page
00:26thought maybe it isn't quite solved, and they came up with an idea
00:29and I hear they built a pretty good company based on that.
00:34But until recently
00:38people thought that that was it
00:40that search was solved, but now we have this thing called Gen AI
00:43and that's under question. At least
00:47three of our panelists are
00:50seeing opportunities in this world
00:52and the fourth panelist is keeping a close eye on the whole field.
00:57So, please, I'm going to introduce them
01:00and then I'll give you a chance to welcome them.
01:04The panelists are
01:05I'll start the end
01:08Chirig Shah, who is a professor
01:10at the University of Washington
01:12Annalise Jansen
01:14the chief business officer
01:16of ProRata
01:17Okay
01:20Julie Bornstein
01:21founder and CEO of Daydream
01:25and Olivier Abacassiz
01:27who is the chief business
01:29no, sorry, is the CEO
01:31of Quant
01:32and also another company called Shadow
01:34but Quant is the one we're interested here
01:36Welcome them
01:39So, I'm going to talk to the
01:44panelists who have these
01:46you know, exciting new companies first
01:48so you can learn about them
01:49I'll start with Julie
01:50and Daydream
01:52you know, tell me
01:53tell me what Daydream is
01:55you know, we'll use the mic
01:56and
01:59and when you started
02:00how you're doing
02:01and what you're doing
02:02Daydream is a
02:03chat to shop platform
02:05so we're taking the spirit
02:07of ChatGPT
02:08and the other
02:09LLMs
02:10and applying it to shopping
02:12for fashion specifically
02:13so
02:14you can
02:15and
02:15it's a consumer facing platform
02:17that actually launches in a few weeks
02:19so we haven't launched yet
02:20and
02:21you can
02:22ask for anything you want
02:24in natural language
02:25my favorite search so far
02:27has been
02:28I'm looking for a revenge dress
02:30for a wedding
02:31to wear in Paris
02:33that has a salt burn vibe
02:36so people are looking for very specific things
02:39Was that customer satisfied?
02:41Yes, I think she actually got some pretty decent results
02:44yeah
02:45and so instead of
02:46you know
02:47sort of
02:47sending you off to links
02:48we show you all the product
02:49that matches
02:50whatever you're looking for
02:51and then we use a combination of image and words
02:54to help you refine
02:55so if you see something you like
02:56but you want it less low cut
02:58or you want it short sleeves
02:59you can
03:00say I like this product
03:01but I want it this way
03:02and you have a conversation with the agent
03:04to help you narrow down
03:05and find the products that you're looking for
03:07Are you sending them to your business partners?
03:10How does that work?
03:11So we have partnerships with brands
03:13all over the world
03:15we have about 5,000 brands on the platform
03:18when we launch
03:19and we'll continue to grow
03:20and once you find what you like
03:22you can save it
03:23you can keep track of it
03:24or you can click out
03:25to go buy it on the site
03:26of the retailer or the brand
03:28How do you intend to let people know about this
03:31and come to your chatbot to use that?
03:38Speaking at a lot of conferences
03:39I'm just kidding
03:42I'm hearing people click away
03:43sign up now
03:44Well you can sign up for the wait list now
03:46at daydream.ing
03:47I-N-G
03:48I-N-G
03:49Daydreaming
03:49Okay
03:51So we have a variety of ways
03:54that we're going to let people know about it
03:55we're working with a bunch of stylists
03:57who have created their own collections
03:59on the site
04:00and we'll be sharing those with their followers
04:03there's some sharing elements
04:05so if you want to ask your friend
04:07or your wife
04:09for advice on which of these 5 dresses you should buy
04:12you can
04:13or suits you should buy
04:15you can send it to them
04:16so we think that the product is going to have sort of natural viral growth
04:20in addition to some of the standard marketing that we'll also be doing
04:25You know sometimes chatbots give weird answers
04:29especially when they're daydreaming
04:31Yes
04:32Have you found that?
04:34How are you stopping
04:35you know when someone gives a specific request
04:38with something
04:38they come up with something totally absurd
04:41We have built
04:43instead of having kind of one
04:45large request that goes out
04:48and brings a model back
04:49We have sort of an ensemble
04:51of many models
04:52that are really good
04:53at each piece of what they do
04:54So there's a model around color
04:56and a model around
04:58fit
04:59and a model around fabric
05:00and so we've sort of understand
05:04the dimensions of the products really well
05:06Every product that comes in
05:08from the catalog of a brand
05:10we enrich with lots of data around it
05:13and so the combination of understanding the product
05:16and then understanding the user
05:18we get sort of better as we go
05:19so we get smarter about the user
05:21and understand the types of brands they like
05:24and their style they like
05:25and so we're able to have our sort of understanding
05:28of the request be pretty refined
05:31and then there's the element of you know salt burn
05:34and fortunately that actually works bizarrely well
05:37and so you know that vibe is understood
05:40and we're able to replicate it
05:41so
05:42Great
05:43Yeah
05:43Well thank you
05:44I want to go to Annalise here
05:46you know
05:47to tell me about ProRata
05:49I used your search engine gist
05:52you know to learn a little about it
05:53but you could share this
05:54with the rest of the audience
05:55Thank you so much
05:56and thank you Julie
05:57I mean I'm listening to your story
05:59and I'm thinking this is what
06:00initially when I started to look into AI
06:03and the business opportunities
06:04there's one word that word that stood out
06:07which is decentralized new opportunities
06:09and I think your story about daydreaming
06:12is about how do I get more precise
06:14in a particular category
06:16but over to ProRata
06:18so ProRata did not start as a search company
06:21it actually started as a company
06:23to look at how we can help media companies
06:26not just survive
06:27but also thrive in the age of Gen AI
06:30and to use a popular or not so popular phrase
06:33which is how can we stop the steal
06:35as a team we came together
06:37with deep media tech
06:39and entrepreneurial experience
06:41connected through a mission
06:42that is how do we stop
06:44or at least highlight
06:45the fact that content is taken every day
06:49from every publisher
06:50every content creator
06:52without even knowing it
06:53I was at a conference a couple of weeks ago
06:55and I would say roughly
06:5880% of publishers
07:00rose their hand and said
07:02I actually don't even check
07:04whether it's taken away
07:05daily, monthly, hourly
07:09so we decided to build an attribution model
07:12and the attribution model is there
07:13to understand for every Gen AI answer
07:17the original content that was used to create
07:20to generate the answer
07:22I'm looking at the professor
07:23who's obviously marking my homework here
07:26the original content
07:27so we fairly credit
07:29and fairly attribute pro rata
07:31the content that was used
07:33to generate an answer
07:34and I can see and hear you think
07:37how do you do that on ChatGPT?
07:39well we don't do it on ChatGPT yet
07:41because you can imagine
07:44that some large tech platforms
07:47do not necessarily like to share revenue
07:50nor would like the transparency
07:52and the accuracy of attribution
07:54and therefore let's just continue
07:56to take content
07:58and pretend attribution is not possible
08:00to prove that attribution is possible
08:03we license content only
08:06so we have partnerships with
08:09by now almost 100 media companies globally
08:13we license their content
08:15we use that content to build
08:17small language model
08:18we actually dissected small language model
08:22create a knowledge graph
08:23and we use that to generate answers
08:26all answers come with
08:27independent attribution
08:29links back to the original content source
08:32and a unique new ad product
08:35that doesn't use cookies
08:37distribution of the product
08:38is with and through our partners
08:42I can talk about this for hours
08:44I want to make sure there's time for others
08:45but what I'm hearing is
08:48you started off with
08:49I think it's a very laudable mission
08:52to stop this deal as you say
08:55as someone who puts out a little content myself
08:58and certainly my employer does
09:01we'd love to have some share of that
09:05of the potential trillion dollars
09:07that some of these entrepreneurs
09:09like OpenAI, like Sam Altman
09:11say they're going to generate
09:14away from that
09:15but it sounds to me like you say
09:16okay we want to do that
09:17but gee I guess we can't do it
09:19unless we like build our own chatbot
09:21is that what happened?
09:22yeah we had to prove attribution works
09:24and whilst building the proof
09:27the attribution
09:28and the ethical answer engine
09:30we realized there's a real need
09:32within the industry
09:33to make money today
09:35and that's why we've built this
09:38GIST powered answer product
09:39to bring AI engagement
09:41to any and every media channel
09:43whilst working on that mission
09:47which is to influence regulators
09:50governments
09:52content ambassadors
09:54to get eventually to that point
09:56where every Gen AI answer
09:58will come with independent attribution
10:00because to make the comparison
10:02I don't think any of us
10:04would buy packaged goods
10:06a packaged good
10:07without a nutrition label
10:09so well you know
10:11a proof of concept
10:12maybe it's not the best
10:13business model
10:14eventually
10:15you know
10:15as your search engine told me
10:17maybe you're looking for
10:19to have some partnerships
10:20with some of the bigger companies
10:22and then have them adopt
10:23your
10:24you know
10:24stop the steal
10:28methodology
10:30there are a number
10:31of partnerships opportunities
10:32which is we are pursuing
10:33beyond direct agreements
10:36with media companies
10:38there are consortium ideas
10:41there are sovereign
10:42decentralized ideas
10:44there are category specific
10:46industry ideas
10:48where the need for a chatbot
10:50within academic publishing
10:52for example
10:54the company's been around
10:56for nine months
10:58there's a lot of interest
10:59to understand
11:00how to build
11:02Gen AI answers
11:04in an ethical way
11:05with licensed content only
11:07great
11:07well
11:08I do have to say
11:09you didn't mention it
11:09but your company was
11:10you know
11:11conceived and founded
11:12by Bill Gross
11:13who is one of the
11:14legendary entrepreneurs
11:15of the digital age
11:17so
11:17that's going for you
11:19he invented pay-per-click
11:20and we're now inventing
11:21pay-per-use
11:23great
11:24Olivier
11:26you know
11:27so quant
11:28I understand
11:28the business itself
11:30on two pillars
11:31a privacy-based
11:33search engine
11:34and the fact
11:35that you are
11:36French
11:37you're European
11:37two things
11:38so
11:38tell me
11:39tell us about
11:41what you've got
11:41so true
11:43quant is a search engine
11:44and as you were
11:46explaining in your
11:47introduction
11:48why should we
11:49sit in a room
11:50where there is
11:51someone that gets
11:5290% of the market
11:53and get the best product
11:55the reason why
11:57we're there
11:57is mainly
11:59because we
11:59believe that
12:01for a user
12:02getting no choice
12:04when it
12:05when it gets
12:06to access
12:06to search
12:07which is
12:08what we do
12:08on the daily life
12:0910-20 times a day
12:11is not an option
12:13and secondly
12:16that you could have
12:18good reason
12:19to search differently
12:20and there are
12:20two different reasons
12:21first is to consider
12:23that you want to keep
12:24the control
12:26of your data
12:26you don't want
12:27that data
12:27to be sold
12:28you don't want
12:29that data
12:29to be used
12:30for any other
12:31services
12:31that you don't know
12:32you're focused
12:33on your data
12:34and they want
12:34to control them
12:35and that was
12:35the privacy
12:36reason why
12:37Quant was created
12:3810 years ago
12:39and when we bought
12:40the second reason
12:41is for some
12:42companies that
12:43consider that
12:44part of their
12:45revenue
12:45should serve
12:47something else
12:48and this is
12:49what Ecosia
12:50which is one
12:52of our partners
12:52I will explain it
12:53is doing
12:54and the revenue
12:55they do
12:56is used
12:57to plant trees
12:58or we have also
12:59a second search engine
13:01in our company
13:02which is Lilo
13:02which puts some
13:04money to NGOs
13:04so we believe
13:07that search
13:07is so important
13:08in our daily lives
13:09that it can serve
13:10your purpose
13:12which is to
13:12protect your data
13:14or to serve
13:17a benefit
13:18could it be
13:20for NGOs
13:21or could it be
13:21for the world
13:24this is the reason
13:26why it exists
13:26it started
13:27almost 10 years ago
13:29as a marketing
13:31company
13:31that use
13:34US tech
13:34most of the
13:36search engines
13:36that are not
13:37Google or Bing
13:38were using
13:39Google or Bing
13:40and then get
13:41that marketing
13:41purposes
13:42what is now
13:44different
13:44and why
13:45we are also
13:46back in the game
13:47is
13:48thanks to AI
13:49we have now
13:50the opportunity
13:51to build by ourselves
13:52a search technology
13:53and we believe
13:54that building
13:55a search technology
13:56is key
13:56for two reasons
13:57first
13:58we have to
13:59drive
14:00our business
14:01by controlling
14:02the technology
14:03and it is important
14:04because Microsoft
14:05just decided
14:06to withdraw
14:06from that market
14:08and not to
14:09promote anymore
14:10their technology
14:11to third party
14:12so it means
14:13that if we don't
14:13have solutions
14:14there is no
14:15possibility
14:16for proxy
14:18search engine
14:19to live
14:19but secondly
14:20we also believe
14:21that access
14:22to web content
14:23is more necessary
14:25and essential
14:26than ever
14:27because this is
14:28what it fuels
14:29Gen AI
14:30Gen AI
14:31exists
14:32because you have
14:32the capacity
14:33to access
14:34to web data
14:35if you don't
14:36access to web data
14:36there is no
14:37Gen AI
14:37and for that reason
14:38which is
14:39as I see
14:40a pluralism
14:43purpose
14:43we have to make sure
14:45that the global
14:45understanding
14:46of the web
14:46is not coming
14:48from a single company
14:49and that is the reason
14:50why we thought
14:51and we did it
14:52with Ecosia
14:53in a Gen Venture
14:54to get our own
14:55search technology
14:56that will get
14:57the overview
14:58of the web content
14:59and the capacity
15:00then to feed
15:01and to fuel
15:02the Gen AI
15:03so in fact
15:04it is true
15:05that we came
15:05from privacy
15:06we believe
15:07that sovereignty
15:08and that capacity
15:10to be independent
15:11by managing
15:12our own technology
15:13and also
15:14by getting
15:15a view
15:16of the web content
15:17which is not
15:18a single view
15:18is key
15:19and then
15:20it means
15:21that also
15:21the business
15:22is evolving
15:23because we were
15:23on the B2C side
15:25and now
15:25we believe
15:29that providing
15:30that technology
15:31to third party
15:31is going to be
15:32the key
15:32for the future
15:33Do you feel
15:34here in Europe
15:36there is like
15:36a hunger
15:37to search
15:39in particular
15:41with
15:42you know
15:42an engine
15:44a company
15:45which is not
15:46one of the US giants?
15:48I think so
15:49I don't think
15:50that we probably
15:51have the capacity
15:52to challenge
15:53the big ones
15:53but I think
15:54that there are
15:55enough room
15:56for other companies
15:57that can develop
15:59services
16:00with
16:01you know
16:01new features
16:02new
16:02new
16:03behaviors
16:04new capacity
16:05so
16:05and once again
16:06what is different
16:07is AI
16:08make it simpler
16:09it is faster
16:11it is faster
16:11and simpler
16:11than ever
16:12to build
16:12services
16:13as long
16:14as you can
16:15access to the data
16:16and that's why
16:17I really focus on that
16:18is if you don't
16:19access to data
16:20you will not be able
16:21to do anything
16:22this has to be
16:24very efficient
16:26and very simple
16:27and in that case
16:29we're going to see
16:30many AI
16:31case
16:32use case
16:32in the future
16:33that will challenge
16:34the search
16:35so where are you
16:36getting the data
16:37you know
16:38I mean
16:38how can you
16:39compete
16:40with the
16:41companies
16:42making the
16:43large language
16:44models
16:44and the chatbots
16:45you know
16:45to come up
16:47with something
16:48as equivalent
16:48or you know
16:50for your users
16:50even better
16:52so
16:53I think
16:54that
16:54the way
16:55to do it
16:56is definitely
16:56on you know
16:58the search
16:59industry
16:59did not change
17:00for the last
17:0120 years
17:01of course
17:02algorithms
17:02get better
17:03but the business
17:04model was the
17:05same
17:05the players
17:06was the same
17:06and there was
17:07no opportunity
17:08now
17:09you have the
17:10capacity
17:10to make
17:11search
17:11totally
17:12different
17:12search
17:13is not
17:14now
17:14sending
17:15traffic
17:15to
17:16third-party
17:17website
17:17you have
17:18the capacity
17:19to keep
17:20the user
17:20on the
17:20search
17:21engine
17:21by providing
17:23quality
17:25answers
17:26to queries
17:26which is
17:27totally
17:27different
17:28so if
17:29you are
17:29able to
17:30provide
17:31good quality
17:32results
17:33and
17:34answers
17:34you will
17:35keep
17:35the users
17:36I'm sure
17:37you heard
17:38about that
17:38but the BBC
17:39did an
17:40analysis
17:40a few
17:41months ago
17:42where they
17:42said that
17:43Gen AI
17:44results
17:45were
17:45false
17:46for
17:4750%
17:48of
17:48the queries
17:50it means
17:51that there
17:51is a huge
17:52work to do
17:53in order
17:53to get
17:54a reliable
17:55solution
17:55to make
17:56sure
17:56that
17:56we don't
17:57answer
17:58and say
17:58we did
17:59the job
17:59we answer
18:00what makes
18:01sense
18:01for the
18:01user
18:01and on
18:02this
18:02I think
18:03there are
18:03huge
18:04capacity
18:04for
18:05companies
18:05like us
18:06to build
18:07not a
18:08global
18:08solution
18:09I think
18:09it's
18:09exactly
18:10what
18:10Julie
18:10was
18:11explaining
18:11we have
18:12to go
18:13vertical
18:13from vertical
18:14and to
18:15deliver
18:15a good
18:16experience
18:17for each
18:18of it
18:18so it's
18:19not about
18:20being
18:20global
18:21it's
18:21about
18:22getting
18:23great
18:24performance
18:24and excellence
18:25on
18:26verticals
18:27okay
18:28thank you
18:29so
18:29Chirag
18:30you've
18:30been
18:30sitting
18:31patiently
18:31hearing
18:32these
18:33entrepreneurs
18:34talk about
18:35what they're
18:36doing at
18:37this moment
18:37in
18:39web history
18:40using search
18:41you've been
18:42tracking
18:43this world
18:44for quite
18:45some time
18:46what do
18:47you think
18:47is this
18:49the inflection
18:50point
18:50and the
18:52opportunity
18:53to break
18:54through
18:54that
18:55overwhelming
18:56search
18:56share
18:57of
18:57Larry
18:58and
18:58Sergey
19:01so
19:01first off
19:02search
19:02is not
19:03solved
19:04despite
19:05Google
19:06being in
19:07existence
19:07for more
19:08than two
19:08decades
19:08actually
19:09this is the
19:09first thing
19:10I tell my
19:10students
19:10when I'm
19:11teaching a
19:12class on
19:12search
19:13because
19:14when we
19:14think about
19:14search
19:15you're
19:15really
19:15referring
19:16to
19:16a
19:16particular
19:17kind
19:17of
19:17search
19:18but
19:18if
19:18you
19:18think
19:18about
19:18there's
19:19so
19:19many
19:19other
19:19kinds
19:19of
19:19search
19:20email
19:20search
19:21still
19:21sucks
19:22desktop
19:23search
19:23enterprise
19:24search
19:25patent
19:26search
19:26legal
19:27search
19:27app
19:28search
19:28right
19:29there's
19:29all
19:29kinds
19:30of
19:30things
19:30so
19:31search
19:31even
19:32in the
19:32traditional
19:33sense
19:33is still
19:34a
19:35problem
19:36that
19:36many
19:37companies
19:37from
19:37big tech
19:38to
19:38startups
19:38are still
19:39trying to
19:40solve
19:40and of
19:41course
19:41there's
19:41a lot
19:41of
19:41research
19:42around
19:42that
19:42but
19:43moving
19:44to
19:45the
19:45point
19:46that
19:46we
19:46are
19:46now
19:46where
19:46we're
19:47talking
19:47about
19:47J&AI
19:47I
19:48think
19:49it's
19:49starting
19:50to
19:50make
19:50us
19:50realize
19:51something
19:51that
19:52many
19:52of
19:52us
19:52have
19:52been
19:53talking
19:53about
19:53for
19:53a
19:53long
19:54time
19:54that
19:54people
19:55don't
19:56search
19:56for
19:56the
19:56sake
19:56of
19:57search
19:57there
19:58is
19:58something
19:58behind
19:59that
20:00search
20:00and so
20:01can we
20:01get to
20:02that
20:02right
20:02when
20:03you
20:03see
20:04somebody
20:04asking
20:05for
20:07how
20:08much
20:10euro
20:11to a
20:12dollar
20:12conversion
20:14they have
20:15a purpose
20:15behind it
20:16maybe they're
20:16traveling to
20:17Europe
20:18you know
20:18from the
20:18US
20:18and they
20:19want to
20:19see that
20:19or they
20:20want to
20:20understand
20:20maybe they
20:21want to
20:21write a
20:21paper
20:22on that
20:22maybe they
20:23are doing
20:23some market
20:24analysis
20:25so can
20:26we get
20:26to that
20:27task
20:28address that
20:29instead of
20:29just addressing
20:30the search
20:32now we
20:33have the
20:34capabilities
20:34to do
20:35that more
20:36easily
20:36than ever
20:37before
20:37and so I
20:38think that's
20:39where the
20:40inflection
20:40point is
20:41that it
20:41allows us
20:42to think
20:43about search
20:44in a different
20:44sense
20:45and go
20:45at a higher
20:46level
20:46so the
20:47example
20:47that Julia
20:48you were
20:49giving
20:49where it's
20:50not just
20:50about finding
20:51a product
20:52but you
20:52have a
20:52specific
20:53purpose
20:54revenge
20:55dress
20:55right
20:55I mean
20:55something
20:56like that
20:56how do
20:57you express
20:57that
20:57how do
20:58you
20:58that you're
20:59not just
20:59looking for
21:00a dress
21:00you're not
21:00just searching
21:01for dress
21:01but you
21:02have a
21:02purpose
21:03and if
21:04you're able
21:05to focus
21:06and address
21:06that purpose
21:07now we're
21:08giving the
21:09customers a
21:10different
21:10language
21:11now we're
21:12giving them
21:12a different
21:13opportunity
21:14and now
21:15it's a
21:16different
21:16playing field
21:17than the
21:18one that
21:18has been
21:19for two
21:20three decades
21:21in which
21:22one single
21:23player has
21:24dominated 90%
21:25of it
21:25and no matter
21:26how many
21:27innovations
21:27have happened
21:28and I've
21:28seen many
21:28companies
21:29come and
21:30go in
21:30this time
21:31with brilliant
21:32new ideas
21:32hasn't made
21:34a dent
21:34in that
21:36share
21:36right
21:37but now
21:38we see
21:38many see
21:39that as
21:40an opportunity
21:40where
21:42you're
21:42probably
21:43I would
21:43and maybe
21:44you know
21:44you can
21:45quote me
21:45this
21:46in a
21:47controversial
21:47point
21:48that I
21:48don't
21:48think
21:49anyone
21:50including
21:51big tech
21:51is going
21:52to make
21:52a dent
21:53in that
21:5390%
21:54market share
21:55but what
21:56you could
21:57do
21:57whether
21:57you're
21:57a big
21:58tech
21:58or startup
21:58is to
22:00focus
22:00on a
22:01different
22:01market
22:02where you
22:03have an
22:03opportunity
22:03to
22:04actually
22:05compete
22:06in a
22:07more
22:07fair
22:08terms
22:09than
22:09the
22:10traditional
22:10market
22:11space
22:11so
22:11and I
22:12think
22:12that's
22:12where I
22:12see the
22:13movement
22:13happening
22:14so
22:14yes
22:15I'm
22:15in an
22:16academia
22:16which
22:18allows me
22:19to come
22:20here and
22:20talk without
22:21any agenda
22:21but I also
22:23work with
22:24industry
22:24I work
22:24with
22:24startups
22:25I'm
22:25currently
22:25working
22:26in two
22:28stealth
22:28AI
22:29startups
22:29one
22:29in the
22:30healthcare
22:31and AI
22:31and one
22:32in Asian
22:32tech AI
22:33and so I
22:34see opportunities
22:35I see
22:35challenges
22:36but I
22:37do agree
22:38that
22:38this is
22:39a time
22:40when we
22:40can
22:41not go
22:42head to
22:44head
22:44or toe
22:45to toe
22:45in the
22:46search
22:47realm
22:47but really
22:48begin to
22:49question
22:49why are we
22:50searching
22:50to begin
22:51with
22:51and if
22:52we can
22:52focus
22:52on that
22:53higher
22:53purpose
22:53we can
22:54do some
22:55innovative
22:56solutions
22:57that really
22:58generates a
22:59new kind
22:59of value
23:00and new
23:00kind of
23:00market
23:01so let
23:02me test
23:03your
23:04contention
23:04this might
23:05be
23:05controversial
23:05let me
23:06ask the
23:07other
23:07panelists
23:07do you
23:08agree
23:09with that
23:09do you
23:09think
23:10that
23:10Google
23:11will
23:11still
23:12hold
23:13its
23:1490%
23:14yet
23:15for
23:16purpose
23:17driven
23:19activities
23:21some of
23:22you might
23:22be looking
23:23for
23:23yes
23:24there is
23:25an
23:25opportunity
23:26absolutely
23:27I think
23:27there's an
23:28opportunity
23:28and I
23:29think
23:29Julie
23:29is a
23:29good
23:30example
23:30of
23:30the
23:31category
23:33specific
23:34as long
23:35as it
23:35comes
23:35with
23:35attribution
23:36but we
23:36can
23:36talk
23:36about
23:37that
23:37later
23:38I'm
23:39super
23:39excited
23:40about
23:40what's
23:40happening
23:40because
23:41you're
23:41totally
23:42right
23:42the
23:42search
23:42experience
23:43we get
23:43used
23:43to
23:44it
23:44we
23:44think
23:44that
23:44this
23:44is
23:45it
23:46until
23:46somebody
23:47comes
23:47with
23:47a
23:47new
23:48version
23:49of
23:49search
23:49and
23:50I
23:50think
23:51that
23:51opens
23:51all
23:51of
23:52our
23:52eyes
23:52in
23:52terms
23:53of
23:53the
23:53new
23:54opportunities
23:55we
23:55have
23:55to
23:55go
23:56more
23:56either
23:56decentralized
23:58new
23:58partnerships
23:59that
23:59can
24:00be
24:00created
24:01as
24:01well
24:01so
24:02yes
24:04on
24:05the
24:05verticalization
24:06I
24:06do
24:06think
24:07I
24:07have
24:08two
24:08of
24:08my
24:08co
24:08-founders are
24:09from
24:09Google
24:09I've
24:10spent a
24:10lot
24:10of time
24:10with
24:11them
24:11over
24:11the
24:11years
24:11and
24:13you
24:13know
24:13when
24:13we
24:14talk
24:14about
24:14fashion
24:15to
24:15them
24:15they
24:15don't
24:15really
24:16understand what
24:16we're
24:16talking
24:17about
24:17I'm
24:18sure
24:18some
24:18people within
24:19the
24:19organization
24:19do
24:20they're
24:20very
24:20good
24:20at
24:21ads and
24:21so
24:21their
24:21ads
24:22are
24:22relevant
24:22but
24:29of
24:30their
24:30recent
24:31I.O.
24:32conference
24:32and
24:33some
24:34of
24:34them
24:34were
24:35things
24:35in
24:35the
24:35fashion
24:36realm
24:36so
24:36one
24:36was
24:36a
24:37virtual
24:37try-on
24:37and
24:38one
24:38was
24:38automatic
24:39checkout
24:39but
24:40the
24:40underlying
24:42logic
24:42around
24:43what
24:43results
24:43show
24:44up
24:44and
24:44the
24:44messiness
24:45of
24:45their
24:45catalog
24:46isn't
24:46changing
24:47one
24:47bit
24:47so
24:47sort
24:48of
24:48felt
24:48like
24:48a
24:48bit
24:48of
24:49lipstick
24:49on
24:49a
24:49pig
24:49and
24:50I
24:50do
24:50think
24:50that
24:51Gen
24:51AI
24:51allows
24:52you
24:52to
24:53go
24:53super
24:53deep
24:53in
24:54vertical
24:54to
24:55get
24:55really
24:55good
24:55at
24:55it
24:56and
24:56then
24:56instead
24:57of
24:57needing
24:57to
24:57know
24:58the
24:58exact
24:59terminology
24:59that
25:00you're
25:00looking
25:00for
25:00so
25:00if
25:01you
25:01know
25:01the
25:02way
25:02search
25:03in
25:03fashion
25:03for
25:04example
25:04has
25:04traditionally
25:05worked
25:05you need
25:05to
25:06know
25:06the
25:07taxonomy
25:07of the
25:08website
25:08that
25:09you're
25:09shopping
25:10in
25:10so
25:10you
25:10need
25:10to
25:10know
25:11oh
25:11it's
25:12a
25:12red
25:12A-line
25:13dress
25:13oh
25:13they
25:13don't
25:14use
25:14A-line
25:14there
25:14they
25:15use
25:15another
25:15term
25:16and
25:16nothing
25:17comes
25:17up
25:17if
25:17you
25:17don't
25:18use
25:18the
25:18right
25:18term
25:18and
25:18so
25:19the
25:19idea
25:19of
25:19being
25:20able
25:20to
25:20leverage
25:20Gen
25:21AI
25:21to
25:21be
25:21able
25:22to
25:22have
25:22a
25:23conversation
25:23and
25:23let
25:24the
25:24agent
25:24go
25:24do
25:24the
25:25searching
25:25for
25:25the
25:26consumer
25:26I
25:27think
25:27is
25:27a
25:27huge
25:27opportunity
25:28and
25:28I
25:29think
25:29we
25:29all
25:29kind
25:29of
25:30want
25:30that
25:30in
25:30all
25:31the
25:31areas
25:31that
25:32especially
25:33in
25:33commerce
25:33that's
25:33just
25:34too
25:34hard
25:34it
25:34shouldn't
25:35be
25:35so
25:35hard
25:36so
25:37I
25:37think
25:37it
25:47product
25:47well
25:48potentially
25:48and
25:49I
25:49think
25:49that
25:50gets
25:50really
25:50interesting
25:51you
25:51know
25:53just
25:54a
25:54short
25:54one
25:54but
25:55I
25:55think
25:55that
25:55what
25:55you
25:55said
25:56regarding
25:56the
25:56search
25:57experience
25:57is
25:58key
25:58I
25:58think
25:59definitely
25:59as
26:00you
26:00said
26:00search
26:01experience
26:01is
26:02still
26:02crappy
26:03we
26:03can
26:04do
26:04better
26:04and
26:05I
26:05think
26:05Google
26:06showed
26:06during
26:06that
26:07last
26:0720
26:07year
26:07that
26:07they
26:08improve
26:08and
26:08keep
26:09on
26:09improving
26:09but
26:10they
26:10are
26:10not
26:11yet
26:11at
26:12a
26:12level
26:12where
26:12we
26:12should
26:13think
26:14that
26:17others
26:17in
26:18the
26:18room
26:18we
26:18still
26:19can
26:19do
26:20better
26:20how
26:20can
26:20we
26:21imagine
26:21that
26:22all
26:23the
26:23web
26:23documents
26:24could
26:25be
26:25summarized
26:26in
26:26a
26:27page
26:27of
26:2710
26:28blue
26:28links
26:28which
26:29are
26:29clicked
26:30for
26:30the
26:31first
26:31one
26:32and
26:32that
26:33could
26:33not
26:33be
26:33for
26:34any
26:34query
26:34whatever
26:35the
26:36query
26:36is
26:36how
26:36simple
26:37it
26:37is
26:37that
26:38at
26:38the
26:38end
26:38of
26:38the
26:38day
26:39there
26:39is
26:39only
26:39two
26:40web
26:41documents
26:41that
26:42you
26:42should
26:42click
26:42on
26:42it
26:43so
26:43you
26:44know
26:44we
26:45were
26:46in
26:46a
26:46position
26:46where
26:47since
26:47Google
26:48did
26:48the
26:48job
26:48and
26:48there
26:49was
26:49no
26:49one
26:49that
26:49was
26:50able
26:50to
26:50make
26:50it
26:51we
26:51took
26:51it
26:52as
26:52a
26:52reference
26:52does
26:53that
26:53mean
26:53that
26:53it's
26:54the end
26:54and
26:55we
26:55have
26:55it
26:55and
26:56this
26:56is
26:56a
26:56product
26:56that
26:57could
26:57not
26:57be
26:57better
26:58of
26:58course
26:58not
26:59you
26:59know
26:59we
26:59we
26:59just
27:00get
27:00used
27:00by
27:01what
27:01was
27:01possible
27:02what
27:03is
27:03new
27:03now
27:03is
27:04we
27:04can
27:04do
27:05different
27:06experiences
27:07with a
27:08different
27:08approach
27:09and
27:10which
27:10is
27:10much
27:10more
27:11vertical
27:11I
27:11totally
27:12agree
27:13because
27:14you
27:15will
27:15not
27:15be
27:15able
27:15to
27:16get
27:16a
27:16solution
27:17that
27:17works
27:18for
27:18all
27:19the
27:19use
27:19case
27:19but
27:19we
27:20have
27:20that
27:20opportunity
27:21and
27:21it's
27:21probably
27:22the
27:22base
27:22news
27:22to
27:23consider
27:23that
27:24you
27:24know
27:24the
27:24view
27:25of
27:25the
27:25web
27:25which
27:25is
27:25the
27:26view
27:26view
27:26of
27:26the
27:27human
27:27knowledge
27:28cannot
27:28be
27:29only
27:29two
27:29links
27:30to
27:30click
27:31you
27:32know
27:32Julie
27:33I
27:33was
27:34really
27:34fascinated
27:35by
27:35your
27:35talk
27:36about
27:36how
27:37Google
27:38you
27:38know
27:39surprise
27:39these
27:40nerds
27:40don't
27:40understand
27:40fashion
27:41but
27:42maybe
27:44AGI
27:45is
27:45smarter
27:46than
27:46Google
27:46so
27:47if
27:47you
27:47look
27:48down
27:48the
27:48road
27:48what
27:49they're
27:49building
27:50if
27:50you
27:50are
27:50to
27:51believe
27:51you
27:52know
27:52Demis
27:53Asabis
27:53and
27:54Sam
27:54Altman
27:55and
27:56the
27:56other
27:56lords
27:57of
27:57AI
27:58they're
27:59saying
27:59that
27:59they're
27:59building
28:00something
28:00that
28:01will
28:01you
28:02know
28:02in
28:02many
28:03ways
28:04be
28:04smarter
28:05than
28:06even
28:06a
28:06company
28:07that
28:07Google
28:07can
28:08be
28:08so
28:08the
28:10question
28:10is
28:10could
28:10that
28:11AI
28:11be
28:12as
28:12smart
28:13as
28:13you
28:13even
28:14in
28:14terms
28:15of
28:15fashion
28:15and
28:16figure
28:16out
28:17how
28:17to
28:17supply
28:18that
28:18revenge
28:19dress
28:19and
28:19maybe
28:20even
28:20order
28:21it
28:21in
28:213D
28:22print
28:26I think
28:27so
28:27I think
28:29part
28:29of it
28:30is
28:30the
28:30reality
28:31of
28:31what
28:31does
28:31it
28:32take
28:32to
28:32go
28:33deep
28:33in
28:34a
28:34vertical
28:34understand
28:35the
28:35nuances
28:35of
28:36it
28:36and
28:37will
28:38there
28:38be
28:39time
28:39spent
28:39to
28:39do
28:40that
28:41and
28:41will
28:42they
28:42do
28:42that
28:42anything
28:43is
28:44possible
28:44we
28:45all
28:45know
28:45that
28:45that's
28:45why
28:45we're
28:46here
28:46that's
28:46why
28:46we're
28:46in
28:46tech
28:47I think
28:48a lot
28:48of
28:48it
28:48is
28:49where
28:49do
28:51what
28:51do
28:51they
28:51care
28:52most
28:52about
28:52and
28:53where
28:53do
28:53they
28:53spend
28:53their
28:54time
28:54and
28:54how
28:54deep
28:54do
28:55they
28:55go
28:55to
28:55understand
28:55how
28:56to
28:56do
28:56this
28:56best
28:57at
28:57a
28:57vertical
28:57level
29:00I was
29:01going to
29:01make a
29:02comment
29:02about
29:02AGI
29:03because
29:03that
29:03always
29:05tickles
29:05a part
29:06of my
29:06brain
29:06it's
29:06like
29:06really
29:07okay
29:09why
29:10and
29:11what
29:11does
29:11it
29:11really
29:12mean
29:12so
29:12I
29:13think
29:13yeah
29:13everybody
29:14understands
29:14that
29:15we're
29:15building
29:16this
29:16AI
29:16some
29:17actually
29:17even
29:17disagree
29:18with
29:18calling
29:18it
29:19AI
29:19everything
29:19is
29:19AI
29:20but
29:20anyway
29:24I
29:24think
29:24the
29:25ultimate
29:26goal
29:26here
29:26should
29:27not
29:27be
29:27worrying
29:28about
29:28AGI
29:29because
29:29there's
29:29all
29:29these
29:29definitional
29:30issues
29:31is
29:31that
29:31super
29:32intelligence
29:32really
29:32important
29:33is
29:33that
29:34what
29:34we
29:34want
29:34to
29:34do
29:34is
29:35that
29:35really
29:35going
29:35to
29:35solve
29:36all
29:36the
29:36problems
29:36and
29:36stuff
29:37like
29:37that
29:37but
29:37I
29:38think
29:38one
29:38thing
29:38I
29:38agree
29:39with
29:39is
29:39actually
29:40Mustafa
29:40Suleman
29:41who is
29:46artificial
29:46capable
29:47intelligence
29:48and
29:49that's
29:49a
29:49more
29:49practical
29:50thing
29:51forget
29:51about
29:52AGI
29:52in the
29:53sense
29:53of
29:53the
29:54super
29:54intelligence
29:55thing
29:55but
29:57AI
29:57that
29:58is
29:58capable
29:58to
29:59do
29:59all
29:59this
29:59kind
30:00of
30:00task
30:00whether
30:01it's
30:01shopping
30:01or
30:02health
30:02care
30:02and
30:03so
30:03on
30:03so
30:03I
30:03think
30:04to
30:05your
30:05question
30:06before
30:08yes
30:09that
30:09could
30:09actually
30:09make
30:10things
30:11better
30:11and
30:11it
30:12could
30:12be
30:12smarter
30:13than
30:13what
30:14Google
30:14is
30:14the
30:15question
30:15is
30:16who
30:16is
30:16building
30:16it
30:16and
30:17right
30:17now
30:18all
30:18the
30:18big
30:18tech
30:19have
30:19their
30:20investment
30:20going
30:20into
30:21that
30:21so
30:22I
30:22think
30:22it
30:22does
30:22make
30:23a
30:24very
30:24competitive
30:25environment
30:25for the
30:25startups
30:26and other
30:26companies
30:26to be
30:27in
30:27and
30:27that's
30:28ultimately
30:28the goal
30:29because
30:29that
30:29requires
30:30deep
30:30pockets
30:31so
30:32I
30:32think
30:32but
30:32I
30:33also
30:33agree
30:33with
30:33Julie
30:34I
30:34think
30:34there
30:34are
30:34verticals
30:35there
30:35are
30:35niche
30:36that
30:37not
30:38necessarily
30:38require
30:39trillions
30:39of dollars
30:40or
30:40billions
30:40of dollars
30:41of investment
30:41but it
30:42requires
30:42that
30:42creativity
30:43that
30:45many
30:45big
30:45tech
30:46companies
30:46are not
30:46going
30:46to be
30:46able
30:47to
30:47move
30:47as
30:47fast
30:48but
30:48the
30:48startups
30:48can
30:50Annalise
30:51your company
30:52started
30:53with the
30:54idea
30:55of
30:56getting
30:57empowerment
30:58back
30:58to
30:59creators
30:59there's
31:00currently
31:00litigation
31:01going
31:01on
31:02which
31:03could
31:04conceivably
31:05say
31:06to
31:07the
31:07companies
31:07that
31:08train
31:09their
31:09models
31:10on
31:10all
31:10this
31:11copyrighted
31:11material
31:11no
31:12you
31:12can't
31:12do
31:13that
31:13and
31:13they
31:13have
31:13to
31:13roll
31:13the
31:14back
31:14let
31:14me
31:14ask
31:15you
31:15and
31:15maybe
31:15the
31:16other
31:16panelists
31:16can
31:16comment
31:17what
31:17effect
31:17would
31:18that
31:18have
31:18on
31:19the
31:19market
31:19that
31:20possibly
31:21another
31:21opportunity
31:22for
31:23entrepreneurs
31:23to
31:24jump
31:25in
31:28well
31:28we've
31:29always
31:29built
31:29our
31:29platform
31:30based
31:30on
31:30that
31:30attribution
31:31drives
31:31an
31:31economic
31:32model
31:34and
31:34not
31:35necessarily
31:35looking
31:36at
31:36legal
31:37suits
31:37and
31:37whether
31:37it's
31:38fair
31:38use
31:38or
31:38not
31:38fair
31:38use
31:39I'm
31:39not
31:39a
31:39lawyer
31:40and
31:40not
31:40trained
31:41to
31:41be
31:41a
31:41lawyer
31:41and
31:43so
31:43the
31:44economic
31:44model
31:44is
31:44based
31:45on
31:45if you
31:46can
31:46count
31:46it
31:46you
31:47can
31:47share
31:47it
31:47if
31:48that
31:48were
31:48the
31:49case
31:49and
31:49I
31:50don't
31:50look
31:50into
31:51the
31:51future
31:51in
31:51that
31:51way
31:51I'm
31:52not
31:52a
31:52futurologist
31:56whenever
31:56it
31:56involves
31:57either
31:58the
31:59court
31:59case
31:59or
32:00settlement
32:01and
32:01I
32:01talk
32:01with
32:02many
32:02partners
32:02who
32:02actually
32:03are
32:03in
32:03the
32:03middle
32:04of
32:04a
32:04suit
32:04I
32:04would
32:04say
32:05think
32:05about
32:05how
32:05attribution
32:06would
32:06help
32:06you
32:07understand
32:08or
32:08would
32:08help
32:09you
32:09bring
32:10a
32:10new
32:10requirement
32:11to
32:12the
32:12table
32:12which
32:13is
32:13because
32:14a lot
32:14of
32:14these
32:15are
32:15done
32:15not
32:15on
32:16the
32:16premise
32:16of
32:16are
32:17you
32:17counting
32:18it
32:18right
32:18no
32:19it
32:19can't
32:19be
32:19counted
32:20we
32:20actually
32:20don't
32:20know
32:20it's
32:21a
32:21black
32:21box
32:22so
32:23I
32:23think
32:23attribution
32:23really
32:24help
32:31from
32:31a
32:31legal
32:32team
32:32or
32:32coming
32:33from
32:33either
32:370.5%
32:38or even
32:39less than
32:390.5%
32:40of media
32:41companies
32:41have done
32:41a
32:42deal
32:44you
32:44could
32:45use
32:45attribution
32:45to
32:46understand
32:46whether
32:46you've
32:46done
32:47a
32:47good
32:47deal
32:47or
32:47a
32:47bad
32:48deal
32:50I
32:50think
32:50the
32:51good
32:51news
32:51is
32:51that
32:52we
32:54experience
32:54that
32:55there
32:55are
32:55some
32:55value
32:55in
32:56the
32:56content
32:57AI
32:57without
32:58content
32:58doesn't
32:59mean
32:59anything
33:00if
33:01you
33:01don't
33:01have
33:01good
33:01content
33:02shit
33:03in
33:03shit
33:03out
33:03we'll
33:04say
33:04any
33:04engineer
33:05in
33:05the
33:05room
33:05but
33:06that's
33:06true
33:06so
33:06you
33:07need
33:07good
33:07quality
33:07content
33:08and
33:08at
33:08the
33:08end
33:09of
33:09the
33:09day
33:09even
33:10if
33:10for
33:10the
33:10last
33:11two
33:11decades
33:12the
33:12value
33:13was
33:13on
33:13big
33:13tech
33:14and
33:14we
33:14were
33:14talking
33:14about
33:15data
33:15was
33:15the
33:16capacity
33:16to
33:16track
33:17users
33:17and
33:17to
33:18get
33:18more
33:18efficiency
33:19at
33:19the
33:20end
33:20of
33:20the
33:20day
33:20this
33:20does
33:20not
33:21give
33:21you
33:21the
33:22capacity
33:22to
33:22get
33:23AI
33:24with
33:25high
33:25level
33:25of
33:26performance
33:26and
33:27trust
33:27so
33:27we
33:28are
33:28back
33:28and
33:29say
33:29okay
33:29we
33:29need
33:30good
33:30content
33:30to
33:31train
33:31models
33:31because
33:32we
33:32have
33:32that
33:32AI
33:33does
33:33not
33:33exist
33:34and
33:34then
33:34we
33:35need
33:35good
33:35content
33:36and
33:36fresh
33:36content
33:37to
33:37deliver
33:38to
33:38the
33:38user
33:38because
33:39it
33:39is
33:39at
33:40the
33:40day
33:42the
33:42goal
33:43of
33:43any
33:43company
33:44experience
33:45that
33:45makes
33:46sense
33:46so
33:46we
33:52that
33:53we
33:53should
33:53not
33:54only
33:54focus
33:55on
33:55tech
33:55it
33:56is
33:57important
33:57but
33:57once
33:58again
33:58if
33:59tech
34:00does
34:01not
34:01rely
34:01on
34:02knowledge
34:03on
34:04real
34:04knowledge
34:04that
34:05has
34:06no
34:06value
34:07so
34:08another
34:09potential
34:10disruption
34:11on top
34:12of
34:12gen
34:13AI
34:13which
34:14people
34:15are
34:15talking
34:16about
34:16and
34:16some
34:17people
34:17presumably
34:17or
34:18allegedly
34:19are
34:19building
34:19is
34:20some
34:21sort
34:21of
34:21device
34:21that
34:22goes
34:23beyond
34:24the
34:24phone
34:24the
34:25mobile
34:25device
34:26which
34:27is
34:27the
34:28AI
34:28device
34:29I don't
34:29know
34:29whether
34:29it's
34:29a
34:29button
34:30or
34:30a
34:30necklace
34:30or
34:31whatever
34:33Sam
34:34Altman
34:34and
34:34Johnny
34:34Ivor
34:35cooking
34:35up
34:35something
34:35but
34:36it's
34:36a
34:36different
34:36way
34:37to
34:37access
34:37it
34:38and
34:38presumably
34:39I guess
34:39this
34:39would
34:40create
34:40new
34:40obstacles
34:41and
34:42opportunities
34:42have
34:43any
34:43of
34:43you
34:43given
34:43thought
34:44to
34:44if
34:45something
34:45like
34:45that
34:46takes
34:46off
34:46how
34:47you
34:48know
34:48what
34:48you
34:48were
34:48building
34:49could
34:50take
34:50advantage
34:51of
34:51that
34:51or
34:52whether
34:52it
34:52provides
34:53an
34:53obstacle
34:53to
34:53you
34:55I
34:56mean
34:56I
34:57think
34:57the
34:57devices
34:57are
34:58already
34:58here
34:58I
34:59mean
34:59I'm
34:59wearing
34:59an
34:59aura
35:00ring
35:00you
35:01would
35:01argue
35:02that's
35:02based
35:03on
35:03data
35:03which
35:03is
35:04so
35:04important
35:06provides
35:06me
35:06with
35:07insight
35:07I'm
35:07sure
35:08one
35:08could
35:08call
35:08that
35:09AI
35:09so
35:10I
35:10think
35:10it's
35:10a
35:10great
35:11opportunity
35:11and
35:12because
35:12all
35:14in all
35:14that
35:14you
35:14walk
35:15around
35:15the
35:16exhibition
35:17here
35:17and
35:17every
35:18stand
35:18has
35:19AI
35:19in
35:20its
35:20title
35:21and
35:21its
35:21word
35:21and
35:22its
35:22weareprorata.ai
35:23everybody
35:24has AI
35:25behind it
35:27personally
35:27I'm
35:28desperately
35:29in need
35:29of great
35:30use
35:30cases
35:31you
35:31know
35:31what
35:31is
35:32the
35:32actual
35:32how
35:32does
35:33it
35:33really
35:33enhance
35:33our
35:34life
35:34and
35:34now
35:34it
35:34enhances
35:35the
35:35user
35:35bringing
35:37it
35:37back
35:37to
35:37prorata
35:37the
35:38great
35:38use
35:39case
35:39for
35:39us
35:39is
35:39you
35:40can
35:40count
35:41it
35:41so
35:41you
35:41can
35:41share
35:41it
35:42and
35:42it
35:42brings
35:42transparency
35:43to
35:44the
35:44user
35:45before
35:46you
35:46answer
35:48we're
35:48going to
35:49open it
35:49up to
35:49questions
35:50if anyone
35:50has
35:51one
35:51so
35:51think
35:52of
35:52your
35:52questions
35:53while
35:53share
35:54us
35:54his
35:54thoughts
35:55I
35:55think
35:56my
35:56view
35:56of
35:56the
35:56future
35:57is
35:59devices
36:00should
36:00disappear
36:01we
36:02shouldn't
36:02even
36:02have
36:02to
36:03think
36:03about
36:04what
36:04device
36:05should
36:05I
36:05be
36:05using
36:05would
36:06I
36:06have
36:06my
36:07device
36:07connected
36:07to
36:08things
36:08or
36:08whatever
36:08I'm
36:09using
36:09this
36:10particular
36:10device
36:10I
36:11mean
36:12ideally
36:12I
36:13shouldn't
36:13even
36:13have
36:14to
36:14worry
36:14about
36:14thinking
36:14about
36:15is
36:15this
36:15on
36:15is
36:15this
36:16really
36:16focus
36:16on
36:16me
36:17I
36:17should
36:17just
36:17be
36:17able
36:18to
36:18speak
36:18and
36:19somehow
36:20somewhere
36:20in the
36:21thin air
36:22there
36:22would
36:22be
36:22just
36:22the
36:23right
36:23intelligence
36:23to
36:24focus
36:25on
36:25this
36:25person
36:26because
36:26they're
36:26speaking
36:27modulate
36:28and so
36:29on
36:29so
36:30whether
36:30it's
36:30a
36:30device
36:31that
36:31you
36:31wear
36:31on
36:31your
36:32hand
36:32or
36:32neck
36:33or
36:33whatever
36:34a
36:35user
36:36shouldn't
36:37have
36:37to
36:37think
36:37about
36:37all
36:38of
36:38these
36:38things
36:38and
36:39they
36:39should
36:39just
36:39be
36:39able
36:39to
36:40interact
36:40and
36:40get
36:41their
36:41work
36:41done
36:42so
36:42I
36:42think
36:42again
36:43my
36:43first
36:44point
36:44was
36:44go
36:45beyond
36:45search
36:46search
36:46is
36:46just
36:47a
36:47tool
36:48to
36:49the
36:49path
36:49that
36:49we
36:49want
36:49to
36:49get
36:49to
36:50if
36:50you
36:50can
36:50focus
36:51on
36:51that
36:51all
36:58practical
36:58sense
36:59I
37:00think
37:00our
37:00vision
37:01is
37:01because
37:01shopping
37:02is
37:02overwhelming
37:02and
37:03there's
37:03a
37:03million
37:03versions
37:04of
37:04everything
37:05out
37:05there
37:06we
37:07should
37:07know
37:07the
37:07customer
37:08well
37:08enough
37:08that
37:08they
37:09could
37:09say
37:09I
37:10need
37:10a
37:10red
37:11dress
37:11to
37:12go
37:12to
37:13Viva
37:13Tech
37:13and
37:14they
37:14can
37:14just
37:15speak
37:15it
37:16and
37:16we
37:16can
37:16send
37:17to
37:17their
37:17phone
37:18the
37:19five
37:19best
37:19selects
37:20for
37:21them
37:21and
37:21they
37:21trust
37:21the
37:22advice
37:22well
37:22enough
37:23because
37:23they
37:24understand
37:25and
37:25feel
37:26understood
37:28great
37:29I'm
37:30looking
37:30for
37:31a
37:31hand
37:31in
37:32the
37:32audience
37:32and
37:32there's
37:33one
37:33over
37:34there
37:34got to
37:35get a
37:35mic
37:36to
37:36this
37:36person
37:43say
37:43your
37:43name
37:43and
37:44tell
37:44us
37:44my name
37:44is
37:44Claudia
37:46I'm
37:46an
37:47ex
37:47farfetcher
37:48so I've
37:48had some
37:48experience
37:49working
37:49in
37:50e-commerce
37:51and luxury
37:51e-commerce
37:51and I
37:52have a
37:52question
37:52for
37:53Julie
37:53now
37:54with
37:54MCP
37:55protocols
37:56and agent
37:56to agents
37:57talking to
37:57each
37:58are you
37:58seeing
37:59opportunities
37:59in working
38:00with some
38:01of the
38:01luxury
38:01brands
38:02or just
38:02fashion
38:02brands
38:03in general
38:03that would
38:04have
38:04daydreaming
38:05partnering
38:05with
38:06to sort
38:06of
38:06enhance
38:07the
38:08experience
38:08for
38:09the
38:09shopper
38:10ultimately
38:11on the
38:11brand
38:12site
38:12you're
38:12saying
38:13yeah
38:13I mean
38:13are you
38:14seeing
38:14any
38:14innovation
38:15on their
38:15end
38:15as well
38:16that could
38:16potentially
38:17have
38:17that
38:17authentic
38:18sort
38:18of
38:18dialogue
38:19between
38:19what
38:20daydreaming
38:20is doing
38:21and the
38:21brands
38:21what
38:22I would
38:22say
38:22is
38:22all
38:23the
38:23retailers
38:23are
38:24interested
38:24in
38:24it
38:25and
38:25starting
38:25to look
38:26at it
38:26and
38:27probably
38:27being
38:28bombarded
38:28by many
38:29vendors
38:29who are
38:30claiming
38:30to be
38:31able to
38:31provide
38:31this
38:32for them
38:32they've
38:32also
38:33all asked
38:33us
38:33if we're
38:34doing
38:34this
38:34or
38:34providing
38:35services
38:35for them
38:36we're
38:36launching
38:37in a few
38:37weeks
38:37with a
38:38consumer
38:38facing
38:39platform
38:40but
38:40we definitely
38:41would love
38:41to figure
38:42out how
38:42to be
38:42most
38:43helpful
38:43to the
38:43brands
38:44and
38:44retailers
38:44so
38:45we'll
38:45be
38:45looking
38:45at it
38:46as
38:46we
38:47keep
38:48going
38:48so
38:48it's
38:49a good
38:49question
38:49I see
38:49that
38:50LVMH
38:50has
38:50Maya
38:51that
38:52they're
38:52building
38:53so
38:53I was
38:53trying
38:54to
38:54yeah
38:55I was
38:55wondering
38:55if
38:55I think
38:56both
38:56caring
38:56and
38:57LVMH
38:57have
38:57done
38:57some
38:58experimentation
38:59yep
38:59thank you
39:00was there
39:01another
39:01question
39:01over
39:01there
39:10I have
39:11a question
39:11about
39:12so
39:13previously
39:13in the
39:132000s
39:14the disruption
39:15used to be
39:15in hardware
39:15iPod
39:16iPhone
39:16iPad
39:17and now
39:17it seems
39:18to be
39:18in software
39:18you have
39:19the application
39:20ecosystem
39:20and AI
39:21but with all
39:23the technology
39:23being developed
39:24for example
39:25by Johnny
39:25Ivan
39:26Sam Altman
39:26it doesn't
39:28seem to be
39:28taking off
39:29at all
39:29so like
39:30what do you
39:30believe
39:30would be
39:32necessary
39:32for there
39:33to be
39:33another
39:33hardware
39:34disruption
39:34because all
39:35we see
39:35right now
39:36is a lot
39:37of startups
39:37doing incredible
39:38work but
39:38doesn't really
39:39take off
39:39and big
39:40companies
39:41doing things
39:41like Apple
39:42Vision Pro
39:42and they're
39:44very promising
39:45but people
39:46aren't really
39:46buying into
39:47VR
39:47so what do you
39:48believe would
39:48be necessary
39:49for hardware
39:49to be
39:51disrupting
39:51the industry
39:51once again
39:52like I used
39:53to
39:54okay well
39:56you know
39:57maybe a little
39:58off the
40:00focus here
40:01you know
40:01Chirag
40:01you ever
40:01thought of
40:02that
40:02yeah I
40:03think
40:03so hardware
40:05for a lot
40:06of this AI
40:07deployment
40:07often hardware
40:09is sort
40:09of the
40:11the threshold
40:12and that
40:13threshold
40:13needs to be
40:14lower
40:14so for a lot
40:15of the VR
40:16thing you know
40:16the reason
40:16it didn't
40:17take off
40:17is there
40:19was an
40:19extra step
40:20involved
40:20before you
40:21get into
40:22that world
40:22before you
40:23can experience
40:24you know
40:24the things
40:25that they
40:25were so
40:26designed for
40:27and I think
40:28the same
40:29thing I can
40:29see with
40:30AI
40:30that I
40:32don't think
40:33building
40:34and there
40:35have been
40:35attempts
40:35you know
40:36before Johnny
40:36Ivy
40:37and Sam
40:38Altman
40:38there have
40:38been other
40:39attempts
40:39of AI
40:40devices
40:40but they
40:41don't
40:42they haven't
40:43taken off
40:43because that's
40:44yet another
40:44thing
40:45that customer
40:46has to
40:47buy
40:48configure
40:49use
40:50and it's
40:51not always
40:51clear what
40:52value you
40:52derive from
40:53it
40:53so I think
40:54the right
40:54path forward
40:56is making
40:57devices disappear
40:58that it's
40:59something that
40:59just connects
41:01to works
41:02with integrates
41:03with your
41:03existing lifestyle
41:04existing you
41:05know things
41:05as opposed
41:06to yet another
41:07device that you
41:08need to worry
41:08about powering
41:09and connecting
41:10configuring it
41:11and so on
41:11so I think
41:12that's what
41:12it should be
41:13and otherwise
41:17it creates
41:17a different
41:18level of
41:18inequality
41:20right
41:20that there
41:21are people
41:21who will be
41:21able to
41:22afford it
41:22there are
41:22people who
41:23can't
41:24and I
41:25don't
41:25that's not
41:26at least
41:26my vision
41:27of AI
41:27that it
41:28needs to
41:28really benefit
41:29everybody
41:29and not
41:30just some
41:31people who
41:31can afford
41:32to buy
41:33that device
41:34it's like
41:35you know
41:35think about
41:35the web
41:36think about
41:37the internet
41:38right
41:38it needs
41:39to be
41:39accessible
41:40and not
41:41just for
41:42people who
41:42can buy
41:43an expensive
41:44computer
41:44right
41:49let me
41:51oh yeah
41:52there's one
41:52more hand
41:52I'll take
41:53that
41:53then I'll
41:55hi
41:55my name
41:57is Sergey
41:57as well
41:58and thank
42:00you for
42:00your
42:01conference
42:02it
42:02I mean
42:03it was
42:04really cool
42:05and I'm
42:06an ECO
42:06since
42:072010
42:08and of
42:10course
42:10we are
42:11really
42:11worrying
42:11about
42:12what's
42:12happening
42:13and
42:14actually
42:14my
42:14question
42:15is
42:15to
42:16Madame
42:16Bornstein
42:17I'm
42:18sorry
42:18for
42:18the
42:18pronunciation
42:19like
42:21well
42:22your
42:23app
42:24or
42:24LLM
42:25model
42:25that you
42:26create
42:26imagine
42:27the situation
42:28I have
42:28my store
42:29with
42:29this
42:30kind
42:30of
42:31dress
42:31dress
42:32that
42:32for
42:33wedding
42:33or
42:34whatever
42:34and
42:3510
42:35other
42:36colleagues
42:36here
42:37have
42:38the
42:38same
42:38store
42:39well
42:40which
42:41store
42:42will be
42:42higher
42:43and why
42:43in your
42:44LLM
42:45model
42:45this is
42:46my
42:46question
42:47it's a
42:48great
42:48question
42:48where
42:49the
42:50rank
42:51order
42:51of what
42:52we show
42:53is tied
42:53to a
42:54combination
42:54of
42:55what's
42:55the best
42:55match
42:56to the
42:56query
42:57along
42:57with
42:58what's
42:58the best
42:58product
42:59recommendation
43:00for the
43:00person
43:01and so
43:01we use
43:02this
43:02combination
43:03of
43:03personalization
43:04based on
43:04what you
43:05have spent
43:06and what
43:06brands
43:07you like
43:07and your
43:08preferences
43:08to sort
43:10of
43:10re-rank
43:10the best
43:11results
43:11for the
43:12query
43:12so
43:12we're
43:13always
43:14sort
43:14of
43:14combining
43:14those
43:15elements
43:15to
43:16understand
43:17if
43:17your
43:18store
43:18and
43:18another
43:19store
43:19has
43:19the
43:19same
43:20product
43:20in
43:20it
43:20then
43:21we've
43:21collapsed
43:22that
43:22product
43:23into
43:23one
43:23and
43:24you
43:24can
43:24see
43:24the
43:24different
43:25places
43:25and
43:25the
43:25prices
43:26where
43:26you
43:26can
43:26buy
43:26it
43:27but
43:27we
43:28you
43:29know
43:29the
43:29rank
43:30is
43:30based
43:30on
43:30what
43:31you're
43:31most
43:32likely
43:32to
43:32like
43:33relative
43:34to
43:35what
43:35you're
43:35looking
43:35for
43:36that
43:37brings
43:37up
43:37a
43:38bigger
43:38issue
43:39when we
43:39close
43:40on
43:40this
43:40question
43:41the
43:43basic
43:44difference
43:44I guess
43:45between
43:45a response
43:47from
43:48generative
43:48AI
43:49to
43:50find
43:50something
43:51and
43:5110
43:52blue
43:52links
43:53whatever
43:53you want
43:53to call
43:54them
43:54is
43:55it
43:55gives
43:55you
43:55one
43:56answer
43:56with
43:57Gen
43:57AI
43:57and
43:59if
44:02you
44:02don't
44:03often
44:03know
44:03what
44:04goes
44:04on
44:04behind
44:05the
44:05scenes
44:05often
44:06the
44:06people
44:07making
44:07large
44:08language
44:08models
44:09don't
44:09know
44:09what's
44:10going
44:10behind
44:10the
44:11scene
44:11so
44:12how
44:12can
44:12you
44:12trust
44:13this
44:14is
44:14giving
44:14you
44:14one
44:14answer
44:14how
44:15can
44:15you
44:15trust
44:16what
44:16that
44:16is
44:17you
44:17know
44:18you're
44:18getting
44:19a
44:19sole
44:20answer
44:20maybe
44:21it
44:21is
44:21you
44:21you've
44:22got
44:22this
44:22rigid
44:23set
44:23of
44:23rules
44:23but
44:23it
44:24just
44:24could
44:24be
44:24someone
44:25who
44:25pays
44:25you
44:25the
44:25you
44:26know
44:26even
44:27if
44:27it
44:27doesn't
44:27look
44:27like
44:28a
44:28revenge
44:28dress
44:28that's
44:29what
44:29they're
44:29going to
44:29send
44:29you
44:30you
44:30know
44:30and
44:31in
44:31other
44:32kinds
44:32of
44:32things
44:32politically
44:33things
44:33could
44:34be
44:35rigged
44:35in a
44:36certain
44:36way
44:37you
44:37know
44:38I don't
44:39know
44:39whether
44:39people
44:39will
44:40stop
44:40using
44:40these
44:40things
44:41because
44:41they
44:41don't
44:41trust
44:41them
44:42but
44:42how
44:43do
44:43we
44:43get
44:43around
44:43that
44:44one
44:44problem
44:44well
44:45thank
44:46you
44:46Stephen
44:46because
44:47the
44:47answer
44:47to
44:47me
44:48at
44:48least
44:48to
44:56the
44:56instrument
44:57so
44:58it
44:58needs
44:58to
44:58be
44:58independent
44:59I
44:59pitched
44:59you
45:00a
45:00big
45:00softball
45:00didn't
45:00I
45:01thank
45:02you
45:03yeah
45:04but
45:04we
45:04just
45:05need
45:05transparency
45:06getting
45:06one
45:07answer
45:08is
45:08acceptable
45:09if
45:09you
45:09understand
45:10how
45:10it
45:10comes
45:10to
45:11you
45:11if
45:11not
45:11I
45:12was
45:12complaining
45:13about
45:1310
45:13blue links
45:14which
45:14are not
45:14enough
45:15this
45:15is
45:15for
45:16sure
45:16that
45:16getting
45:16only
45:16one
45:17is
45:17not
45:18despite
45:18if
45:18you
45:19know
45:19what
45:19was
45:19the
45:19process
45:20and
45:20if
45:20transparency
45:21which
45:21did
45:22not
45:22happen
45:22in
45:22fact
45:22in
45:23the
45:23digital
45:23industry
45:23so
45:24far
45:24is
45:24the
45:25answer
45:26I
45:26want
45:26to
45:26thank
45:27you
45:27all
45:27it's
45:27been
45:27a
45:27great
45:27discussion
45:29thank
45:29you
Commentaires