Startup Battlefield At Disrupt: Day One, Session One

Do@, RexSee, also Skyline and Deja. I have no idea what any of those companies do, let's find out. Please welcome to the stage from Do@ Abby Bendavid, the co founder and Kathy Brooks who's in charge of evangelism and strategic relationships. Big round of applause for them.

Hello, everyone. So, today I want to share with you our vision for the ultimate mobile search experience. It's a search engine built from the ground up just for mobile. Mobile search today is similar to web search only smaller. Search engines crawl the web and generate web links. Now, what if web links are not the best results for mobile?

That would make the whole infrastructure of crawling redundant. What are the best results for mobile? Mobile is all about apps. If you want a movie, you go to Flickster or IMDB. If you want a restaurant, you go through Yelp or Zigat. so what if instead of crawling publishers, we should simply let them answer users themselves with their apps.

That's exactly what we've built, so let me show you. I'm going to demo from the iPad because of the projection, but this is actually the iPhone version.

Okay, so I'm going to start with a search, a simple search, so I type "social" and the system understood that what I want is a movie. That got my intent. I'm going to click that, and I'm going to get my answers. Now, my answers are not links, there's apps, I'm being answered by the best apps. This is flixter with all the reviews.

I'm sliding to IMDB with the different information, Wikipedia. Now this is not a link to Wikipedia, this is Wikipedia. I tap in, I can go in, read, and when I'm done, I'm just going out and I can continue, NetFlix, where I can get the DVD or click a button and launch the native app to actually see the movie.

Movie clips, downloads, it can do a lot of, let's see if I have any movie clips here. getting more, Fandango is here, Amazon, YouTube, I can in, watch the trailer. Now, none of these apps actually has to be on my iPhone. These are all HTML5 apps created by either the websites themselves when they create mobile optimized website or by app developers, a team of five app developers creating apps with APIs from the different publishers out there.

Let me just stop this. So, do it as a completely open platform. Any publisher, anybody can submit an app that competes in movies, and if they're good, they 're going to be at the top, if they're not good, they're going to be at the bottom. Let me give you another example. Let's say I want to find, finished with my movie and I want to eat something.

So, let's start with my favorite. I am going to type in sushi. So, the system, again, understands my intent and gives me two options. I can go to cooking and get all the apps about how to make cookies and videos and so on, but I'm going to choose restaurants. Now, because we're in a mobile environment, the system already knows my location, so the first thing I'm getting is Yelp, with all the restaurants around me, or I can go in, and just choose what I want to eat, based on pictures.

And that's again, around here, around this area. Or I can go to Foursquare, and the advantage of Foursquare obviously is that Foursquare knows me. So it knows who my friends are, and it can give me tips for my friends ahead of other tips. And then I can see slides of the others I got where even Google places, and I can do this, and go back to the one that I want.

And, that's basically my search experience.

Now, people ask me, so how do you decide, which apps comes on top, and which app is at the bottom? So, we're doing this with a ranking engine called Social Radar. What this Social Radar does is I can, as a user I cansay, "I really like this one. I can put a heart on it." Okay? So from now on, the radar will save...you know, at the heart of the radar, this app...I'm gonna get this app first for restaurants.

But then I'm going to get the apps that my friends on Facebook like. And then I'm going to get the apps that their friends like. Social Radar is looking for signals of what are the apps that are trustworthy, that me and my friends trust, and bring them to the top and pushing to the bottom all the things that we don't like.

Now, it's not only for restaurants and movies. We actually covered - I'm going to show you - we actually covered pretty much all the top things, all the popular things that people do with their mobiles. And for each one of them, basically we're open for all the publishers. We have hundreds of publishers, over 500 for launch, that are covering all these things that people do with their mobiles.

So it's a very functional system.

Let's look at another example. Let's say I wanted to listen some music. So, put in Rihanna, at music, and I'm being answered again by the app. So first thing I'm getting is YouTube, using the old video clips of Rihanna and then because this is an iPad or iPhone, the most logical thing to do, which usually you don't find on search engines, actually the ability to download from iTunes the songs.

Pandora is here. I can read anything, everything about Rihanna, but I could also click a button and go and launch the radio, start it immediately. Slideshow, some pictures of Rihanna. Or I can go to apps like SoundCloud, which basically streams the music straight in. Down a bit. Now, while I explain I can go on and see maybe there's a, maybe she's around here and actually she's live and in New York on twenty first.

But I can also see that here is in mash up done by an app developer where telling me, ok, from last FM also Britney Spears, that if I'm a fan of Riahhna, I might want to see her as well. So, three things that we are changing with, let me just...so three things that we are changing about search that are very fundamental.

The first thing is mobile is all about apps. So we're doing the first mobile search that answers users with apps instead of with links. The second thing is also very deep is that we're not crawling anymore. A lot of you here are publishers and app developers. We're not saying we're going to crawl you and put your information on a search engine.

The opposite. We're allowing you to come up and answer users yourselves with your own app, with your own content, with your own business model and your own brand. And the third thing that we are doing is we are bringing actually a social radar that builds what we call a trust graph. Who are the brains that are trusted by you and your friends?

And then somebody else in another place might get completely different results based on who this person and their friends actually trust.

I 'm giving you about ten more seconds so...

Okay, so...

...ten more seconds to talk. So round of applause then for Do@.

So thank you very much.

Otherwise ...otherwise people tell me off that they're...other people have longer than others. But hopefully the judges will ask questions for anything you didn't cover. So who's to start with, let's start, Bonnie Pell, from Bing Search, it's mobile search, this is the future of mobile search we're seeing here today.

Yeah.

You must be excited.

Definitely, great job guys. I really like the user experience.

Thank you.

And, I like the fact that you're launching, you know, you're actually launching deep into the web pages and the scrolling interface is nice, I haven't seen that before. The app's already live, so I really like that. So, I guess one question from the search perspective and ultimately comes down to search quality.

OK.

Search quality Yes.

So out of all the things you could show how do you make sure your showing users the right ones? And you touched on that, but I'm curious how you 're measuring your results so that you can know, as you're launching this thing, that that's actually good for users in the task they're trying to accomplish.

So, the Social Radar is built on the users themselves deciding what are the best results. If I, for example, go to news. If I put a heart on CNN and I put a heart on I Live Alone, so I put a heart on the Guardian, these two will be first for me. That's the quality for me. Now, my friends will see those two things a little bit higher and friends will see a little bit higher again.

It's a social system that measures quality.

So, that's how you determine what to show. My question is, how do you, do you have a way of objectively testing. So, you can run it in the [xx] and say, yeah, this is actually good for our users.

We've tried to stay away from making these decisions for the users. Actually, we think it's an open, we think the best way create great results is by open competition. If you have a better way to answer a music question, you submit your app. It's just a URL that you give us, we put it on, if it's great it's going to be on the top.

We have a few examples of those, of apps that are completely unknown, we put them in the system, and two days later they're at the top.

Hilary, you wanted to jump in with something.

Actually, this is starting to address my question, which is, you know, I have a smartphone, I have a Nexus One and I never think to myself, boy, like, if only I had a meta-app to help me decide which app to run, I'd be much happier. I'd just run the app I want to run. So, how are you going to convince people to even start using it?

First of all, we are concentrating on the product We want a great product. We want something that you can find anything there and it's always the greatest app and you're always discovering new apps. And you can share there's a share button there where you can share and submit to the world what is your preferences.

So that's the product is the first thing. In addition to this, we are very publisher-friendly because we allow publishers to not be crawled, to actually answer users with their own business model. So there's a lot of potential distribution deals that we're kind of working on at that level. And the third thing to say is just the team.

We've done this before. We have a great team that has a lot of experience bringing kind of mass-market products to market.

Bradley.

I share Barney's concerned about search quality. One of things about the mobile interface, as opposed to the web interface, is much harder to scan and say, "Oh, it's the fourth result that I like." It requires a lot more flipping and intention. So it's sort of a rich get richer form factor, where the people that do make it to the first screen are going to continue to sort of get you know, so I would advise you to sort of look at ways of doing AB testing or somehow salting the results so that you can actually validate per Barney's comment.

But my question is are you going to have a business model or do you consider that passe.

Well, this is a search engine. So people tell us their intent. This is very valuable. They're biz models are built into that.

Are you going to sell placement to the apps? Are you going to prefer, you know, a sponsored app.

We have it built into the system. We're not going to turn it on for choice. We want, first of all, to bring the best result and to send free traffic to publishers, basically. You know, just to comment on your on your quality answer, these are the same exact answers, the same exact publishers that give you the apps that are so popular.

So, Flixter is answering my movie questions. They are pretty good at that.

I 'm gonna just jump in on that. So the real issue is that every single publisher thinks that they are the best app and should come up number one, but I as a user I don't necessarily agree. OK, and so I want to make sure that I am getting the actual best objective quality. And so our real questions here are how are you going to, when you've got all these publishers contributing things, thousands of them for a given answer?

How are you going to decide what's right? And how are you going to know that out of all of the algorithms you could have, many of them might be bad, that the one that you are delivering to users is actually good.

So, what I would saying is look at the user experience perspective. Okay, from user experience, if you get to a Bing result or Google, you get a list of links. OK, and then you go in. You decide on one of them based on a piece of text. You click in, you go in, and it's not what you wanted. It's a big page.

It's slow, it's not even mobile. You go back, you find another one, you click on that that's not.

On our system, swiping is not, swiping is not a commitment, you swipe and here's another one and you actually see the results at the swiping level, so you can make a conscious decision if you want to get in or not. What we're seeing from user questions is that users are exploring much more content through this interface than through any other interface.

It's because it is so easy, so quick, it's the fastest way to results, it's not the fastest way to a link page, it's the fastest way to actual results.

Micheal are you I guess my question is, on one side, I mean I think the user interface looks, just, it's so interesting because one of the challenges that are in the sort of apps world, versus what we can see on the web. It's very hard for apps discovery. But, how do you get past the question of, the cynical user who says, okay, they're directing me to an app that I haven't downloaded, that i haven't bought.

I mean, how does it go beyond somebody filling that app that the person who's created that app is just spamming you, versus, in the web world, where I can choose among everything.

I think the punishment for spamming in a system like that is very quick. Users are going to say you are a spammer, you're going to go straight to the bottom.

After I bought the app?

You don't have to buy, in this system you download duet only. Everything else is HTML 5. You don't have to download any of the apps there. But some of them, if you go to Pandora, for example, wow that's a great one. There is a button there that says launch the radio. Click launch the radio. If you already have Pandora installed underneath and it is installed, it will go straight into Pandora.

If it doesn't have it will take it to the app store to download it. So, it's actually a great tool for discovering native apps because it's connected to them. The Netflix app will go straight to the movies. The iTunes app will go straight to iTunes. This is one interface across all the native apps that is in one language, HTML 5.

Okay, so we're out of time so another round of applause for Do@.

Thank you very much.

Okay, so that was sort of number one. Do you guys want water by the way? There's a big thing of water here. I'm going to hand out water to the judges.

Thank you.

One for you. One for you.

It's deceptive, its in a hard to carry I don't know why I don't just carry it to. I thought there was ice in there. I'm afraid it's room temperature, but, you know.

So, it's time to move on to our second exciting start up. I think are you guys ready?

Yeah Please welcome to the stage Joe Restnikoff. Welcome to Josh. Welcome back from Rexley.

Hello world. My name is Joel Reznickall. I am the CEO of Rexly. My co-founder is Brad Lodenback and Kyle Flemming are here with me today. At Rexley, we are building a service that delivers superior quality recommendations and true social discovery for media content. We started this company because discovery is still a big unsolved problem.

Think about it. There's millions of pieces of premium content that are delivered through a multitude of devices . Finding what you want, where you want, is an enormous headache. In fact, it's more than that. It's a pharmaceutical grade pain.inpoint . When you go to the front page of Itunes, you see a tiny, miniscule sliver of their catalog, and some people love Beiber; some don't.

Brad does? But we think there's a better way to find what you want in iTunes and in other premium content channels as well. So, like any good entrepreneurs who went out and studied the problem, we talked to the top hundred in the Netflix prize, we read every paper on recommender systems and the answer is pretty clear to us, it said the discovery is fundamentally a data problem and not an algorithm problem.

In other words, getting good data is far more important than tuning models, and don't take my word for it. Those words come from the best algorithm tweaker in the world. So, what kind of data are we talking about, there's 2 kinds. The first is behavioral. It's when you actually spend your time and money on, on the services you actually use, which is measurable by us on the account level.

it's social data. We all know that social is important, but the really critical thing is getting the signal from the noise. So, for any product category, there are four to six people that I truly trust and that list is steroids for collaborative filtering. And all this fuels what we believe to be the ultimate recommendation.

Let's take a look at that. The social actionism, implicit recommendation, instant gratification, it's object level similarity based on what you actually listen to. It's social consensus from a wider circle, and the opinion for critics you explicitly trust. We give you the most important dimensions and then you can decide where you wanna buy it based on pricing and accessibility.

So on that note, let's check out the demo. What you're seeing right now is a streamlined feed follow system for iTunes. Our goal is to show you the most important items from the most important people. Any action you take in iTunes creates a story item that your followers see. Every story item has a play button, which streams samples directly from the iTunes store And also one click to buy directly from the store.

You can socialize around any item and send comments out to Facebook and Twitter for your non-Rexly user friends to enjoy. So, we weave actions in our algorithms, so think about a single song listen versus a repeat album listen. We also rate friends by their importance. Not all friends are created equal.

So we allow users to create a short VIP list that we call "Super-trust." It's limited to 6 people, it's completely secret, so it's not biased by friendship, and you can promote friends in and out of this list, and those people eventually dominate the user experience. You try to add a seventh person, sorry, denied, it's limited to six people.

So we did a lot of need finding in our product development. We talked to more users than you can possibly imagine. What we found is different people want different privacy settings which is crucial to sharing data with us. On one hand, there are people who want to share everything with everyone and we love Love those influencers, they are great on our site right now.

There are certain people who only want to share certain things with certain people, and that's okay, we tailor a setting to their needs. But for the most part, we're in the age of sharing, and people are okay with sharing most of what they do with all of their Facebook friends, and this setting is really popular and reduces a lot of friction for social discovery.

Another thing we heard from users is that they wanted to be able to curate their feed. This is the Britney Spears effect. Everyone listens to her, but no one admits to it. Who here listens to Britney Spears? 98% of you do. So there's a hide button, so you can hide any item on the song, artist, or album level and it won't appear on the site or in social media until you un-hide it.

On the other end of the spectrum, users told us they want to scream discoveries from the mountaintops. Oh my god, I just found this great- Hi mom. That's not what I was gonna write in that comment. But you can scream from the mountain tops, it's published out to social media, and what's really important to know is that when an affiliate link and we are effectively monetizing product that users are going gaga over.

And it's just the beginning, though. We're gonna expand to more mediums, more channels, so we're gonna blow out music through the next most ubiquitous channels,we're gonna add But not for long. We want to be wherever you consume and purchase content - whether it's on a tablet, a mobile device, or a living room device.

And we've seen Don Cauldwell's speech about why not to do a music so affiliate sales is our core revenue stream with modest conversions across all the product categories we can cover. We can get a $1.75 monthly ARPU, that's monthly, and we're already surpassing our ARPU contribution from music alone.

Advertising is another important revenue stream. We're already talking to advertisers, our data is solid gold for them, but, for us, the holy grail is licensing. Thinking about bringing your recommendations with you wherever you go We're creating a deep personalized taste graph which will be extremely valuable to media and brand marketers.

Once again, we are Rexly, this is just the beginning. Thank you very much.

That's Rexly, every one. Nice high five there at the end.

He did a good job.

Anyone want to jump in? Straight away Michael.

I think in the same way that the judges are expected to be more like the judges on American Idol and have those same comments. When I see this I'm reminded of Susan Boyle of Britain's Got Talent, which is I dreamed the dream. And I am wondering...

Are you saying our site's ugly?

Well, no.

But it can sing, that's the main thing.

And it does.

I think that the question is, if you really can solve this in a way that iTune's genius and Ping hasn't, I mean Ping, not Bing, then you've done something extraordinary, so but, how are you going to get past, is this an experience that users are really going to have to decide, how does it compare to what they're getting somewhere else?

Not necessarily. We don't want you to start using new media platforms. You don't have to stop using Netflix or stop paying attention to the recommendations. It's very easy to set up our site and we're going to engage deeply with you through email and social media, and eventually, we want to help out Netflix.

We want to make you a better Netflix customer. We want to be the best iTunes affiliate, so in a sense we can do something as a third party that no other party can do themselves which is cross platform data and we want to serve users just as much as we serve the industry because we come from the industry and we care deeply about content.

Bradley? Or Hillary? Either way.

I actually have two questions; one is technical and one is frivolous. So, I'll start with the technical question.

We take frivolous questions, too.

So, I've spent a fair amount of time thinking about recommendation systems, and you have what I would call classic 'cold start problem', and that it's very hard to give quality tune recommendations when you don't have a lot of data, so how are you addressing that? My second question is, where does your name come from and what does it mean?

Okay, so we have thought long and hard about 'cold start' problems, which is why our first feature is a feed-follow system. You don't need ten thousand users on the site to get utility, you can get utility from day one with 3 friends on the site; to us it's a Trojan horse to get at data and from there that's when we start layering in the true algorithmic horse power.

And you know what, recommendations in themselves are not a value proposition to most people. It's a see it to believe it. Maximum value with minimal effort and then wow them even more once we layer in those features to get the data. And the name is, it 's another way of saying recommendations without saying it and it's a five dollar domain that was available for ten dollars.

Bradley.

In the event that Apple launched a music oriented social network that was deeply integrated with iTunes itself, do you think this idea would stand up?

I think they already did.

Oh yeah I forgot.

So let me tell you a quick story, 2 weeks into our development, iTunes launched Ping and we thought, shit, we're dead. But, we took a close look at the product. We realized we were fundamentally different. We're web based and we're also cross-platform and also we learned a lot from their launch.

We went on to Core and every other forum on the internet and we actually tabulated what people wanted and those are the features you're seeing today, that was a big part of our need finding. It was all or nothing publishing the social media. We think it should be on a case by case basis. it's not all or nothing with who you can share, it's the Facebook friends.

Quite honestly we don't think there's gonna be a joint venture between Apple and Facebook. We love both companies. You throw Amazon and Hulu, and other folks into the mix - this is something we know can't be done, and we've talked to a lot of those companies already. Like I said we're from media, and we care about media.

Bonnie.

Yeah, so I agree with Michael's comment about the level of vision here. So I think, that's great, you're really dreaming big. Ah, I'm stuck by 2 questions. One is still trying to figure out: When would I, as a user, use Rexly instead of using Netflix of iTunes, sort of is there really a cross media value proposition here of using TV and the rest.

And then, related to that, is to look one of your key innovations is that there is like 6 sort of super trust people, and you can't have any more. So I have trouble believing that you know even if I wanted these recommendations, I would actually want just 6 people. Much less 6 people across music and TV and video and everything else like that.

So we ask for 6 but we also pay attention Everyone else you interact with on the site in they are weighted in our algorithms as well, and if you super trusting someone, you actually never try their stuff e're probably gonna ignore that. Remind me of your first question.

When would I use this, you know, when would I decide to use Rexly as opposed to going to Netflix or iTunes and relate to that is, do I really value sort of TV friends recommedation combined with music ?. Is it really an all media proposition here.

Got it. So, you're right. We have a very big vision, we dream very big. That doesn't mean we know exactly how we're gonna get there, so there's kinda 2 use cases or 2 paths we can go down. Rexley could either be the place you go to when you say "Hey, I got an hour to kill, what will, what in the world will make me the most happy?" or it can be a lean forward shopping experience, you can think of it as Netflix, Walmart for Digital media that's highly personalized, that helps and doesn't hurt the value chain.

And on your point about having multiple media, the people that dead media, we can say, you know, a lot of super trust list for music, another one for books and another one for movies. We don't believe you're gonna trust the same six for all those categories.

Mike you had another question?

For all of us in the entertainment business the challenge isn't finding the Rhiannas or the Justin Biebers and let's hope my six don't really like Justin Bieber. It's really finding, it really is that word discovery, which is so overused. It's about, in the entertainment business, it's about how do you get these new artists to break and also, I as a user, I can go to AOL music, I can go to Yahoo and I get the same.

I get Rhianna, Rhianna, Rhianna. So, am I gonna be able to find something like Odd Future or some other band that nobody else knows about or am I just gonna rely on my 6 friends?

It depends on what you want to get, our view is that your reasons for liking what you like are no better than mine and you can follow whomever you want that's gonna help you out. We're not seeing a self fulfilling prophecy on our site, we envisioned as ideal long tail for everyone is perfectly matched to what they want.

And the way that differs from the real world right now is the head's a little smaller but the tails are a lot higher and the area underneath is actually growing. So, we...

Brad.

That's about it.

Well we're out of time anyway. A round of applause for Rexley.

Thank you very much.

Thank you guys. Alright. We 're seeing some good t-shirts this year. I'm really excited by the quality of t-shirts. So it's kind of a Rexly one, maybe a Do@ one. All right let's move on to the next, let 's move onto the next startup. You guys? Are we ready? Or we're popping out laptops?

So, yeah, judges, if you want to just interrupt

By all means, do that.

Alright. I think we're about ready, so please welcome to the stage our next startup. We've got Weotta with a Grant Werneck as the CEO and founder, and Jacob Perkins is the CTO and founder. Please welcome them to the stage.

Weotta 's a lot like having your own personal consierge. It solves the problem of what to do tonight. It also takes all the hassle out of planning it. Today there are many products on the market that can help you find a restaurant, a bar, or perhaps a concert, but there's not a single product on the market today that can tell you what places and what events go together for your occasion, until Weotta.

What you see before you is a product of hundreds of interviews. Through these interviews, we found that most people, when planning any social occasion, think of two things: "What's my mood?" and "Who's coming with me?" That's the top of the Weotta page. Simple, straight forward, the way real people plan.

The second thing a lot of people told us, when they go out and do things with their friends, is that they often times have a particular something in mind like an event. So we made it very easy for them to tell us what that event is, and we'll generate plans that are not just nearby it, but relevant to their itinerary and their occasion.

The third thing that everybody told us that they wanted: review snippets of what's new and fresh and what's happening in the area. And that's the rest of this page. We even took this as far as offering you personalized itineraries based on your Facebook profile. Planning has never been easier. Let me show you how it works.

Let's say I want to go on a date tonight with my amazing girlfriend in San Francisco and I want a classy date. I say dating, classy and I press generate. And the second I press generate, Weotta goes out across the web, looking at thousands of places and events, algorithmically classifying them, ranking them, and on the fly composing them into itineraries to fit my needs.

You can see I have lots of awesome opportunities here, and lots of great things to do, and relevant filters to narrow them down. Planning's never been easier, but let's do something that's really relevant to all of us today, right now. We're all in New York, and, that means, for a lot of us, that we're out-of-towners.

What is the plan around Tracy Morgan?

Well, Tracy Morgan's in town, so right now, right now it's telling me, it's like, go to Tracy Morgan and then here's where you should go to dinner beforehand, and here's where you should eat afterwards.

Is that hand curated, like somebody made it?

Everything is algorithmically generated, not a simple thing is hand curated. So, it really is going out in real time looking at every place and event across the web, and saying "Oh, these things go together and these things go together", and then it's ranking them and classifying them on the fly.

What food goes with Tracy Morgan?

What's that?

What kind of food goes with Tracy Morgan?

Well, we know that Tracy Morgan's a comedian, right? So, what we're going to try to do is we're going to give you a place that's going to be...

You can stop. I'm just kidding.

But anyways.

I like how that totally disrupted the whole thing. I'm sorry. I...

That's what this meeting is about.

OK, I noticed my clock is different too.

Does Tracy Morgan know that Weotta is stalking him?

Tracy Morgan should. He really should.

I have no idea now what to do about the timing, so let's just carry on. We'll just pretend none of that happened and we'll just move on.

Okay, well what goes on in the TechCrunch Disrupt after-party, Barney? Well that's what we're going to learn right now. Alright, you guys, so what's relevant to all this today, we're all in New York, we're all going to the TechCrunch Disrupt after-party tonight. We all need help knowing where we should go to dinner beforehand.

We all need to know where we should party afterwards. Let's do this. We' re all friends here and we're pretty hip. That second Weotta did go back out across the web and it looked every single place reviewed and algorithmically classified again, it's that fast. So, Google maps is being really mean to us, I'm sorry.

So we're going to choose plan number three here. Jake, can you bring it up? If you notice, it's telling us to go to Bond for dinner, which is a pretty awesome place for dinner that's nearby. It's also telling us to go to Bar America for a drink afterwards and it's telling us to go to the TechCrunch Disrupt after-party.it made our evening for us.

You can see that we have everything here to make a decision. We have Zigat reviews, we have OpenTable reviews. Everything is here in review snippets with links out. You even have relevant information about each place, and you can get directions from place to place to place here, as well as booking links out to OpenTable.

If there 's events here, you have links to Stubhub, Ticketmaster, etcetera. And as I said, we're really gonna do this plan tonight, together. So I'm gonna delete this last thing 'cause I'm not into Bobby Flay right now, so Bar America's out. And the minute I delete it, this thing's mine. I can do whatever I want with it.

I can customize this however I want. But I do not know New York that well.

So I'm going to ask a few of friends who know it a lot better than me. So not just does Weotta do stuff, we get your friends in on the picture. So I'm going to ask my friend Ally, my friend Alice, and my brother Forrest. They all know New York a lot better than me. The minute I ask them and press send, they all get a Facebook notification saying that I've asked them for some advice on my Weotta plan.

You can see that we imported here down on the wall, and you can see that Forrest has added Employees Only. So in real time, boom, up here it is. You can see Employees Only is the bar and the Twitter, and Zagat and FourSquare views look awesome. So, we're gonna do this plan together tonight and the next step now that we have our plan is to make a Facebook event out of it.

What's awesome about making a Facebook event on Weotta, is that it's even easier than making a Facebook event on Facebook sites itself. Watch as I do this. There's our Facebook event. That happened in less than two seconds, every single place in our plan is here, and I can all of you guys to join us.

Here 's a link back to the Weotta plan and all the relevant booking links are here as well. So as I said we're really doing this tonight, and you're all invited to join us and employees only, after the TechCrunch Disrupt after party. So in summary, Weotta plans are generated on the fly. They're never curated, they're never canned and I can't wait for questions about the algorithm.

Making plans has never been easier, and we've just scratched the surface as we launch more metropolitans, pull in more content, learn more about you and your tastes, Weotta will have the right inventory to the right people at the right time. So today, I just made a plan around the Tech Geeks after-party.

It's your turn to make plans around that, and each day of the conference, as you guys make your plans around those after-parties, tweet them, and the best and creative tweets each day, we're gonna give you guys uber-cabs for your evening for five best tweets each days. Thank you.

Weotta, everyone. Yeah, I have no idea how this question's in the middle of the things in the sky.

Yeah, it got me totally off, but it's great. You guys are awesome.

Yeah, we're just putting you off in one of the most important pitches of your life, it's cool.

Ok, who wants to jump in again? Hillary, why not?

Yeah, so that's really cool. How does it work?

So, how does it work?

We collect a lot of data from across the web and use hundred of signals. We do a lot of text classification, trying to make sure all the good stuff floats to the top and we have our expert systems so they can put everything together into itineraries.

Cool.

It learns as you use it and it also learns as more people around you use it. We use your Facebook profile to some extent at this point, very little, and the more we learn about you the more we're going to be able to deal with that.

So there is a personalization factor?

There is a personalization layer. It's going to get a lot richer. At this point you saw those personalized itineraries on the home page, that's mostly like relationship status and a few things about your tastes, but that's going to change What does your relationship status say about where you like to go out?

Well, it's tells us what kind of plans to serve up to, so like, let's say you're married and you're a lady, well, you probably want some of those girls nights, and you probably want to know, like, hey, where should I go with my husband on a date? When the kids are sleeping.

How come you just showed us a web app instead of a mobile app?

That 's an awesome question, and I knew I was going to get that.

You have an answer for it?

So, we did hundreds of interviews, and we learned that most people do their planning while they're at work. So, that's why we launched this one first, so people could really start using it. We have awesome mobile plans, we have take it with you plans, and we're probably going to hire a bunch of 15 year oldbecause a lot of traditional media falls short of being able to do this for you, and yes, there are individual applications whether it's Yelp and others, but the big question from most people.

It's Thursday night, big night. It's date night, what am I going to do? So, I mean this is very exciting. I guess I have one question, which I am not sure how you get past. Which is, inevitably, unless, if I am single then this is great, but if I'm going out on a date, how does it know what my date is interested in?

Does she have to log-in herself, do we just compare our plans? So the first layer would be "we're going to give you ideas of what other people have done dates." The second layer would be if she's already a user of our products or you invite her invite her into there. There's two avenues you could take here.

She could help give some feedback or if she was one of our users and you were like "Hey, here's who I'm going on a date with" and she said it's okay. You guys can collaborate on that or you can get her tastes through our system into it. We haven't released that part of it yet.

That's romantic, that's really romantic.

Very romantic, isn't it? Dating in the computer age.

Yeah, so I really enjoy this. I think you're hitting on sort of one of the key problems in search and mobile search. "Dinner and a Movie" is one of our classic examples and you see in all the major search engines. So, with that said, this becomes an execution play. Right, and then it comes down to, sort of, how well, how good of a product can you build.

Yeah.

And, how do you get good quality early on? Because if I come to the site I'm expecting value and it gives me bad recommendations and sort of people I don't know and care about seeing, and I've got to like say, "Great. I had my chance at five, and then I'm done. I'm never coming back."
How do you get good at this up front?

So that's why right now we've only launched San Francisco and New York. We haven't launched the rest of the country. Mostly because we've been able to validate the recommendations were given to you are very high quality. How do we scale this quickly? We hire some awesome people, we raise a lot of money.

Right? And we scale it very fast. The awesome thing about this is since things are not curated we don't have to go hire a staff to hand do everything. We have to hire a few more amazing people. to work on our back end. And a few English majors to help us with that training system. And you know all about what goes into something like this, both of you.

Who decides that Tracy Morgan is classy? Like, who's the guy?

So, it's, yeah.

Sara Lacey asked me that by text, and I just thought I'd share that with the room.

I'd say the restaurant before it and the bar after it if I would have gone into it are probably more of the classy things, but the fact that you're going to a comedy show...

So it's like a sandwich, the bread is classy but the filling is...

Yeah, we wanted to give you a show that was available today. And so, what I said the right inventory at the right time--what I mean is, there's a lot of latent inventory out there, with local deals, local services, and with local events. And if we can push the right inventory to the right people at the right time, then you are seeing a play of a lot of billions here.

Will it tell me where Dave Chapelle is?

What's that?

Will it tell me where Dave Chapelle is?

and actually make sure I can get tickets. That will be the future of this.

This is really addressing like half of the socialization problem, right? Because you're normally trying to optimize over not just what you do, but who you do it with. So are you thinking of addressing that at all? So, I have all these awesome friends, but I don't know who's available on a Monday night and interested and.

So, here's my push to Facebook. Facebook, there's got to be a bunch of you guys out there working Facebook. You guys we need some killer APIs and there's a lot that we can do with Facebook Events here to make that a reality. And Facebook Events have been relatively ignored to where it could be and we can actually take care of that for you guys.

Do you have a feedback loop? Like, after I go out on one of your recommended dates, is there a, you know, this worked, this didn't work. I would change it this way or?

We do. It's not released in today's product that everybody gets to use for this mostly because we are still doing a few things with that.

You are sending people out on these disaster dates and you just aren't accountable.

Twitter will be a great way for us to be publicly accountable. Just tweet us and tell us, that thing sucked, that thing was wonderful, etc.

Another question on the coordination and my dinner and movie example.

Yeah.

So, it's not enough to sort of have a recommendation for a movie and a recommendation for dinner but they have to fit together. And so they're showing that they're kind of maybe geographically nearby, but are you going that extra level and saying that I actually have a reservation available at the restaurant at this time and movie tickets at this time.

Click here and do it all for me.

So that's why I talk of the access inventory part. We need to link in a lot deeper with a lot of these publishers and these providers of the inventory. We've already starting talking to some of the major providers, I'm surprised they talked to a start up as early as us. But they are very interested, cause the problem, they have not been able to solve themselves.

So, as we expand our team that's something that we are going to be able to do very well.

OK, well we are out of time.

We are in the right place.

So round of applause for Weotta!

Thank you.

Okay, yeah, I have no idea how this shouting out question in the middle of the thing is going to work. I didn't really think that through. But carry on, it's kind of entertaining. Alright. We have two more start ups to go. So, just slightly more than halfway through. Some exciting ones. I think if

Let's move on to our next startup. Please welcome to the stage Skylines.

It's Martin Panovice and Mikhael Frackers, founders of Skylines.

Sorry I rushed you there with that setup.

No problem.

Hi, I'm Rathan.

Behind the desk setting up our computer is Michael, and our other co-founder, Koon, is somewhere in the audience cheering loudly.

Oh, he's over there. We're from Amsterdam, and one second.

There you go.

One half million pictures are shared every day.

Such as this one.

And a lot of fast growing companies, such as Instagram, Wipro and Twitpic, are sharing these photos.

How do you find pictures you like?

That's where we come in.

Skyline organizes real time photos, so real time photos what Google did for websites. We connect to your social networks to collect your interests. We really believe interests are more than just people. They can be something, someone, or somewhere. So how do we do that? From Facebook, we collect the pages you have liked.

From Twitter, we collect your saved searches and recently used hashtags. And we're working on integrating to Foursquare to collect the places you're a mayor, but we have an unscheduled trip to New York coming up. Sorry, Dans. Then we collect over a million pictures a day from surfaces such as Instagram, Twitpic, and Yfrog.

And we're working on connecting many others, such as Color and PicPlz. Mention your personal interest to that stream of realtime photos, means everybody gets his own skyline, hence the company name Skylines. So, how does it look? This is our website. Let's say you're interested in pictures of this route.

You'll see a stream of realtime photos. Such as this one. That guy looks great. Skyline is realtime. It's like a Twitter timeline. It's a real-time stream of photos with the most recent pictures shown on top. You can also sort the pictures by our picture rank algorithm or only show photos from your friends.

We found that this is really useful for current events. At the other side of the world, currently at a movie festival in Cannes is going on and these are pictures taken there, almost live. It's also useful for breaking news, like the tragedy that happened this morning in Joplin. There's a lot of photos on that on our system.


That happens with live demos. You can It also shows pictures based on your location. So next we'll show photos taken around us. We are not only launching this as a website, but also as an app. Both on Iphone and on Androids. Next, The apps contain the same skylines with the same pictures based on your interest. For every photo we have a photo detail page, such as this one. It shows the photo, who took it, when, and where, if we know its location. It also shows the interest that we connected to it. And you can, of course, click true to the original photo, such as here on Twitpic.

And you can also share the photo on your favorite social networks, Twitter, Facebook or share it by email. Explorer helps you discover new topics. We have twitter trends, our on staff picks, or recent news in there Then there's the back channel. The back channel aggregates photos around one topic from all different picture services and shows them like a screensaver, full-screen, automatically refreshing.

You can check out skyline.es/backchannel for the photos of TechCrunch Disrupt. So how do we make money? Since users define their own interests, brands can target an interest like TechCrunch Disrupt. If they want, combined with location or time. So for example TechCrunch Disrupt might be suitable for large brands, or TechCrunch Disrupt FBR 94 will be suitable for a Deli across the street, who can offer a discount to entities.

We're also working on a widget for brands and site owners, who can include our picture stream on one topic, onto their own website. They can revenue share on our ads or pay us a monthly fee for an ad pre experience. This both increases our revenue and our reach. So to recap, today we launch the website, the iPhone app, the Android app, and back channel for events.

We've also opened up the sign up for our widgets, which will release soon. And all our services are built on our own API, for which sign up also opens today and which will also release soon. Thank you for your attention, and thanks to TechCrunch for, after 400 years, giving a couple of Dutch guys a chance to take Manhattan once again.

That 's, that's Skyline. More great t-shirts. Love, love the t-shirts. Bradley, straight in.

You guys are from Amsterdam?

Yup.

What are you smoking? No, I'm kidding. I've actually thought it was really cool, especially the moment that...

You've been sitting on that joke for the whole thing, haven't you?

Sorry, couldn 't help myself. I really like the mobile integration. I think that especially when the form factor is small, it's especially useful. both just sort of have nontextual browsing, and if you look at the Facebook film strip at the bottom of their mobile app, I found that useful and innovative and sort of an existence proof that there is a space to sort of consume my information through the lens of the visual, so i thought this was really compelling.

Thank you, appreciate it.

I have a question about how are you dealing with the licensing of images?

We actually, we don't host the photos ourselves. The photos are actually shown, like Google Images does, are still shown on the original site. We just show the thumbnails and those represent, so we're not hosting the original photo. We actually allow people to click through to see the original photo.

Hopefully, once we get really big, Twitpic gets more traffic from us.

That will be an interesting thing to resolve.

What I find exciting about this is that in a lot of ways photos themselves are becoming, it 's not just content, they're becoming something that people are going to go through and I think we're going to see a lot more of this on publisher sites. How are you going to make this industry Something that goes beyond just a search experience to something that's more like a browsing experience where I can find stuff, so if there's a disaster today in Joplin, I can go go through it and I can see different photos and see what's going on.

You want to take it?

Oh, sorry. We currently, we offer explorer so we will show you hints of what we think are interesting topics, and we're working on also sorting pictures differently, not based on recency but also who took the picture, where and when.

We actually call it...

Does that answer the question?

We call it "picturing," and it actually means that based on relevancy, let's say that a friend of you shared a picture of an interest you have. And if it's a location where you're a foursquare mayor, that picture will probably be really relevant for you and we'd have that information in our database to show you that picture really fast.

think this is interesting. I'm trying to think how this becomes big, so that's going on with Michael's question. You know, do you expect the users are going to start going to skylines.com and spending a lot of time there?

It's actually...

And if so, kind of why? And if not how are you going to get thinking about it. We're a couple of small guys from Amsterdam, obviously, so, it's not like we'll expect to have millions of users just on our own power. But we actually do think that with the release of the widget, which we plan soon, is that any publisher can put his, our widget, define their own keywords, lets say Gary Vanderchuck with the wine site, can put our widget on his site and then include wine.

Let's say the burgundy, the red one or the white, Musso wine or a sweet desert wine, whatever. You can target a keyword and it will show a picture that people post on wine. Once the revenue share in our ad program will do that, and if he doesn't want ads in there from a comparative competitor, then he can pay us a license fee.

So, we're actually expected to grow because publishers will help us grow bigger, so, we're open to talks with AOL and see if we can do something.

So then, let me follow on. You've got the challenge which is user generated content could be sort of deceptive or spammy or, whatever. So, if I'm a publisher am I taking a chance of like randomly sort of bad photos associated with my brand?

Of course, we're working on filtering that. If you have tens of thousands of followers on Twitter, you're probably not going to post fake pictures. We will show those pictures more often, and we're on top of it, and of courseusers can flag photos if they are inappropriate. And we're probably, for publishers, going to offer a fully moderated version.

So, they'll pay us something and we'll make sure old pictures are safe.

So they can either edit it before this picture is going or edit it right directly when there are the number of flags, kind of like on Youtube, the same system.

When you were showing the disrupt pictures. Many of those photos were sort of bad, like poorly lit or sort of crowd shots or, you know.

Yeah, I can tell you why. Because we have a couple of our staffers in the audience that took most of those pictures just now.


Exactly.

We gave them really cheap 3GS iPhones, we haven't gone in actually and paid them to get iPhone 4s and up. And that's why the pictures are not up there.

But I mean, my question is does that matter. I mean, if the photos are relevant to me, a bad photo of something I care about.

This, so we actually wanted to show you that we were timely. So we can actually find pictures that were posted three seconds ago, and you could do it on Instagram, Twitpic,YFrog, and soon services like Color and the other stuff that's coming out, that we really love. You heard Dave Morin this morning from Bath talking aboutStacks that we're introducing, or I'm hoping that he will open it up, so we can built on top of their API into photos as well.

For us it's more about showing you fast results here. Really quick, we took it from Google and they did pretty well with that.

And that's another thing, often pictures that are taken from a special angle, even if they're bad photos are nicer than the pictures. Like the Canon example, we have some amazing photos taken by people in the audience which aren't as professional. I was thinking about the real photographers. They were really cool.

We saw Leonardo DeCaprio.

And the guy was shooting himself; he was taking pictures of himself, and he said, "Look behind me", and it was Leonardo Decaprio sunbathing on a boat. It's not the best picture, but it's funny.

Yeah. So that actually brings up another question: Do you think that there is sort of a few use cases that likely dominate like celebrity spotting or it could be sort of news disaster coverage or something. Can I have anything from your early thoughts on that?

All we've, when we were testing it, we had a couple hundred try out the app and then we started looking at what kind of interest shows up the most photos, and it was Stars Wars. So, we can never tell in advance what people are going to do. We can't predict the use case, and we feel its and this is not being arrogant, but it is like Twitter.

You can't tell, I think, when they started Twitter, and then all this. I don't think you ever thought that Hilary Clinton would call up and say, "Please don't go down from maintenance during the Iranian elections." you know, you don't know, and in this case, we never knew, when we were up here, just two hours ago, about Joplin, because we were still coding last night.

We actually saw it as we were behind the demo table. And then somebody said, "Could you do Joplin?" and it showed up. So, you can't tell, you know, and hopefully it'll be more fun pictures than the pictures from the tornado, obviously.

One of the things that we were asked to do in our role as judges was to make suggestions, and I think that given where a lot of the world is going to go in terms of how people are going to view photos, it would be great if you had some sort of a timeline, so if I'm looking at an event I'm again coming back to what's about to happen in tornadoes, or what happened in Japan.

How do I, will I be able to have a time line which I can see as things are evolving? Basically, the website is a timeline.

Oh, great.

So, that's it.

So much happier when he actually uses a computer. The most recent photos are on top and the older ones are at the bottom. That is basically our timeline.

There is a twitter like timeline where the latest one is on top unless you click "relevant" and then it sorts you in a Google-esque way, or a Bing kind of way, what is the most relevant search.

Fantastic.

So he can tweet that.

Okay, well we're actually out of time, but you guys can demo that later. But, give a big round of applause for Skylines.

Thanks.

Alright, who I gather were just a couple of guys from Amsterdam. All right, so that leaves us with just one more, one more startup to go and then the judges will submit their their votes. And then again the judges votes will be tabulated, there will be some audience voting. The whole thing will be put together and then Wednesday you'll be seeing

some perhaps from this session, but some of the start ups you see today you'll be seeing again on Wednesday in the finals. So, yes, hopefully Eric explains how the audience voting, if there is audience voting, works. I have no clue.

Alright, so on to our final start up. This is Deja. And please welcome to the stage Avner Shiloh, the co-founder and Nimrod Ram, also a co-founder of Deja.

Hi . We're Deja, and we'd like to ask you to just lean back and watch this.

Yeah, Ma?

What are you doing?

I'm taking over a TV network.

We just saw a part of the movie Hackers where the main character, Zero Cool, gets so fed up with what's on TV that he decides to hack the local TV network in order to change the programming to his taste. Today, almost fifteen years later, TV programming still sucks. It still sees us as a demographic and not as individuals with particular tastes and interests.

We'd like to change that and empower you to discover new content that's relevant to you. It seems that all the video content is already online. So why haven't people cut the cord yet? We believe the traditional television still has one major asset: the passive, lean-back user experience. TV doesn't ask me what I want to see.

It just shows it to me. It's simple, engaging, and addictive. Online video sites, on the other hand, constantly ask me to make decisions or search for something specific. It's complicated and frustrating and just not any fun. We'd like to make online videos simple, engaging, and addictive, just like television.

And we want to do that with the power of social media.

Today, everybody has a social graph mapping out their tastes and interests. And we use this graph in order to share and discover videos. Eighty percent of Facebook users share video links online. We'd like to combine the simplicity of television with the power of social media.

Today we are launching Deja, a new iPad app for sharing and discovering video. We are now watching a movie from my Facebook channel. And you can see that it's really easy to just flip between videos, just like on a TV experience. On the top you can see who posted this video, when, and you can see a description about this video.

It 's really easy to just share any video that I like, and we support Airplay, so you can use your iPad as a remote control to truly enjoy online video on your TV. When I zoom out, I can see the Deja world. In Deja, all the videos are laid out by date, with the larger one being the videos most trending on your network.

But it's really fun to just move around and immerse yourself in a world of content.

All these videos that you see here are just from today, and they're all coming from my social networks. I could just plug in Facebook, Twitter, Google Reader, or any RSS feed, and Deja will immediately start gathering videos ready for me to watch.

I can tap any video to play it. Just swipe it to move to the next video.

In Deja, each user is a channel. It's really easy to just tap each user to see what videos they've posted. And when you see all their videos in one place, you really get a sense about who that person really is.

Deja is a great tool for video discovery. And we see each video as an entry point for a new world of content. So, for example, if I just pinch-zoom a video, I immediately open a world of content related to that video, and I can do this again and again, until I find what I'm looking for. Even a simple video search becomes an instant channel of content.

So for example, if I search for Joplin tornado, I immediately get an instant channel of content ready to be consumed. Deja is a free app, and we're launching today, it's going to be live on the Apps store later today. I want to talk a little bit about modernization. We're currently focused on increasing our user base.

But, we believe that video advertising is an obvious match for a product such as ours. We also believe that, we also believe in the Flieberg[sp?] model of strategic partnership with premium content providers. And, we're here because we truly believe we can democratize television and we want to do it together.

Thank you.

So Deja, democratizing television at last. Who wants to jump in?

I'll jump in. So, so first of all, I really like the user experience, and I think that the sit back experience on TV invites a whole new world of sort of search and discovery. So I think they're kind of pushing that space. I think there will be big value and big winners there. So my question for you is, what are the sort of either innovative features you have now are your strategies going forward so that you can be one of the big winners, instead of kind of a feature that then gets consumed kind of by larger players.

I think right now our main feature is user experience as you pointed out. And we're really focused on the streamline lean back experience, sitting back and enjoying the video. Further down the line we're really interested in curation and seeing how we can make this a two way conversation between people who are viewing and people and your friends, and bringing channels of content out of stuff that you like, and not just from the web.

Aren 't you worried that as a company like Skylines extends into video, they're just going to crush you?

I think they should be worried from us.

Well, congratulations on launching your app. That's a big deal! I actually had a question about the UI. It's really beautiful, but I love that as you, you click in on a video, you see other things related to that appear. What is it spatially aware? Are things in space more related if their closer or farther away?

How does it decide right now that's true. It's actually spatially aware, we start building out particle, building it out from the center, and it's less and less related to its added to outlyers but we're always playing around with new interfaces trying to find other ways to set up the contents so it'll be cool and fun to play with.

I think that the interface really magical, and really innovative, and I know that that sort of Manga layout, where sort of both Those are really hard problems to solve. They've never taken off, sort of, as an everyday interface, you know, apart from Minority Report and things like that. It may be too far advanced.

Would you respond to that? Do you feel like this is a, the world is ready for an interface like this, maybe with tablets or.

I think so. I mean, obviously we're here and made this interface. We feel that the iPad represents a revolution in the way people interact with whatever content they want to interact. And the power's here, and even the minority report hacks, which have been coming around from Connects have started to take seed and really take off around the world.

So I think we're ready, yeah.

You spoke a lot, sorry.

No, no. Go ahead.

You spoke a lot about like the sit back, relax, watch the TV experience, and yet your URI is a really active experience, like it still looks like you have to, you watch a video, it's probably just a few minutes, you page into the next video. Is there one button you can push that says, "Just show me interesting stuff, I'm going to, you know, take a mental break now"?

I think, you want to answer that?

Yeah, sure. We see our interfaces separated into two parts, discovery mode and lean back mode. And the full-screen mode is an automatic mode that lets you just watch and just flip between the videos and the Deja World is the discovery place where you can just scroll around and find new interesting stuff.

But even in the Deja World, when a video finishes playing, it automatically plays the next one. So even if you're not active at all, it's a lean back experience.

Michael?

And what you've created is really beautiful, not just from a user interface perspective, but also from the perspective of Something that, that really could change the whole way that people watch television and the average American today is watching about 30 hours a week of television, which means they're missing a lot of sleep and they're eating in a chair.

If you can get a piece of that time, then that's going to be extremely valuable. Are you expecting that either people individually or other media companies are going to start branding their own channels, or is this totally a self-initiated experience?

Right now, we're, sorry, right now, we're totally self-initiated, but obviously as we mentioned in this clipboard model of forging the strategic relationships with premium content providers is a great way for us to monetize and give high visibility to brands and other TV networks in our application.

Have you thought about actual TV integrations? It would be cool if that were an electronic programming guide that actually showed me what my friends were watching or have watched in their living rooms, in the moment, as opposed to scrolling through 500 channels.

That's a good idea. No, we haven't thought about it yet.

So we got to hit you again on the relevance question. so you're showing lots of possible candidates and I've seen a lot of companies in the space. There's a lot of exploration, different signals, different sort of features and ways of presenting. So at the end of the day, what do you have inside your company in terms of DNA for search relevance, making sure you're bringing back the best things and collecting data and tuning your system so that I can trust I'm going to get the best things instead of going to some other service that will do a better job.

Well, we're starting out from your social feed and that's something that other companies are also doing as well. So I think it's a good starting point because it means that things that are coming in are the things that your friends have shared and are probably more relevant to you. Over that we want to perfect our algorithms and find out from the users and from the information coming in if we can make it better.

I don't have a good answer right now.

Oneyou can do is measure long clicks, like do people actually get through the video and watch it or do they click and abandon, you know, "not for me".

Passive data exhaust.

Are you actually keeping track of what users are watching as opposed to what the users are recommending?

Yeah, we're keeping track analytically of what they're watching, what they're sharing how much they've watched, and this is information that can be used to put back into the algorithms and make our search relevance higher and better for the users.

And do you have sort of core, like search engineers inside the company now?

Sorry.

Do you have search engineers, sort of quantitative data mining type of people inside the company?

Not yet.

OK.

Not yet.

OK, well we're out of time. And so, one more round of applause for Deja.

Thank you.

And let's get one more big round of applause, that's the last in this session. One more big round of applause for all of our startups in session one. That was disrupting, search and discovery. So time for the judges to go and submit their votes. So let's get one final round of applause for our amazing panel of judges, for Barney, and Hilary and Brandy and Michael.

Thank you guys.

And I think now, I think you go that way. I'm not entirely clear, Eric's going to show the way.