Startup Battlefield Competition - Flight #1 | Disrupt SF (Day 1)
11:30PM Sep 5, 2018
Ladies and gentlemen, please welcome to the stage TechCrunch editor and your battlefield. Host. Anthony Ha.
First of all, thanks to whoever wrote that voiceover and gave me a self promotion. I appreciate it. So how many of you here have been to the startup battlefield before? Raise your hand.
Alright, so mostly first timers? Well, I'm going to tell you everything you need to know. It's very, very simple. Over the next hour or so you're going to see five startups take the stage, they are going to present for six minutes each. And then they're going to have six minutes of q&a with our expert judges. Judges are basically gonna be allowed to ask whatever they want. At the end of those five companies. You're going to get a bathroom break, we're going to go backstage, we're going to choose our favorite companies. And that is going to be used to select the finalists who will present again on Friday. So make sure you're all back here for that what they're competing for is the disrupt cup and also $100,000, which I believe is the largest prize we've ever given out at disrupt. And those are all the rules which means it is time for me to bring out our judges. First up, we have Jim Adler, founding Managing Director at Toyota AI ventures. He's also an executive advisor at the Toyota Research Institute. And he serves on the Department of Homeland Security data privacy and integrity Advisory Committee. That is the longest thing in my note. Next up. We have Sarah go a partner at Greylock. She spends her time thinking about b2b applications and infrastructure. Cyber Security, artificial intelligence, augmented reality and healthcare. Next up, we have Jeremy Liew, he's a partner at lightspeed Venture Partners. He was the firm's first consumer specialist before that, in the early 90s and 2000s. He works for Netscape, AOL and other companies. Next up, we have Sam O'Keefe, she managed the startup battlefield at TechCrunch for many years. We missed her. She is now the head of startup programs at Google Cloud. And last but not least, we have Hans Tung managing partner at ggV capital, where he's focused on consumer internet, e commerce, internet of things across China and the United States. He's invested in 11 unicorn. So let's give it up for our judges.
Okay, and let's actually start the battlefield by bringing out our first company. Karma car resenting for karma car as Arias, Rita and Patrick, men coming out
91% of American households own a car. That means almost all of you know the pains of owning a car, it starts with the car dealership, you have to fight to get a good deal, it's exhausting, then you're locked into three or five years of car payments, you're stuck, and then you're responsible for everything about the car registration and insurance oil changes. It's just too much to drive. So why do we still have the hassles of negotiation, commitment and maintenance after two decades of Zipcar and almost a decade of Uber and Lyft
because the average American commute is over 10 miles each way an Uber and Lyft would cost thousands of dollars a month for this commutes. So right sharing has not replaced car ownership because it's too expensive for daily use for most Americans. So instead, car ownership is the highest has ever been replacing car ownership requires a simple affordable option for daily driving me karma monster month commitment free and affordable car subscriptions. Our platform offers an all inclusive subscription to a vehicle you keep at home. Karma is simple. Everything happens in one app, you select your car, you select your monthly mileage plan and you're done. Insurance maintenance in our site assistants are all included in every plan. Karma is convenient. When you book a car we delivered to you and you keep it full time. And best of all karma is flexible is entirely month to month. So we don't lock into years of payments for a single car. So if you want a different car, maybe a truck for the winter, just change your plan. If you're traveling it composite subscription will be the simplest car you ever drove.
Our team's technical background is rooted in the auto industry. I have my PhD in computer science was applied research and fleet management. And I've worked in data science for predictive maintenance and fleet operations. Our co founder previously managed car value projections at true car or backed by TechStars mobility. And together we're building the future of auto retail
karma has built a complete technology platform for car subscriptions. We partner with fleet operators and rental companies and car dealership that owned the vehicle so we don't own any vehicles. We offer our platform as a white label solution to our fleet partners. Commerce technology has three components. The first is a connected telematics device that's installed on each vehicle. It just takes 30 seconds install and it's used for fleet management per mile billing and monitoring driving behavior like speeding and acceleration. With this telematics platform, we have integrated a unique insurance policy that is tailor made for car subscriptions through a partnership with commercial insurers. Second, will build an operations dashboard that allows our partners to manage everything about the car subscription program, from inventory management to CRM and logistics dashboard. And third, we built a mobile app that makes it incredibly simple for consumers subscribe to vehicles. Let me show you how that works. Let's move to the demo, please.
Let's say you're looking for your next car. You start out by downloading the app on Apple, Android, or iOS. Super simple. Let's move the demo please.
And you can search you can learn about the subscription program and search for availability in your neighborhood. To do so you just enter your zip code, and the app will show you available cars. By selecting a car you can learn more about the vehicle and the various subscription plans that are available for the vehicle. Looking at the plans, you might choose the 700 miles per month plan, which includes insurance and maintenance. If you drive a bus 700 miles during the month, there is a per mile fee. And if you drive under there, you can use the next month. So simple and transparent. Now let's log into the app if you don't have an account creating one is simply can do so by providing your basic information and driving license. Once you're logged in, you can manage your subscription and keep track of your driving, for example, because how many miles have driven for the month and how that compares with your subscription plan. And finally, when you're ready to return, the car is simply scheduled right in the app. And that's one of the best parts of his subscription, you have month to month flight ability. Let's go back to the deck. So to summarize, karma is a complete technology platform for car subscriptions with integrated insurance. And we offer our platform as a white label solution to our fleet partners. We have a simple enterprise software business model and we charge our partners a monthly fee based on the number of vehicles on the platform subscriptions offer just the right balance between ride sharing which is too expensive for daily use and buying or leasing which is a multi year commitment with many hassles. And that is a big business opportunity. In the United States alone. Consumers spent over a trillion dollars last year buying and leasing cars subscriptions offer a compelling alternative to leasing which by itself was over $150 billion last year.
In our first pilots, the launched in Columbus in Chicago. We signed up hundreds of customers in a matter of weeks in partnership with a rental company and our waitlist has been growing 40% each month and is currently over 1900 people. And here at disrupt. We're publicly launching our white label platform for car subscriptions. And we've already signed up partnerships in four cities, including a leasing company with over a billion dollars in originations. Today, we ask you this now that you know about karma Why would you ever lease another car three years ever again when you can subscribe to one and if you are a fleet owner that wants to deploy your underutilized assets for subscriptions partner with us a karma car.com thanks so much
great idea I'm gonna cut to the chase
What are your renters terms with a decent companies? What what rental terms that you Lisa for a year, at least four quarter? How does it work,
so we don't own any of the vehicles we only provide the technology so the fleet operators are typically a rental company or a car dealership company. So we provide the platform that allows them to manage their fleet, manage their operations, do the billing everything they need to operationalize a subscription program. So we're only a technology vendor.
Okay. So if you look at the unit economics, from the perspective of the leasing company work with what would be sort of the utilization rate they would need to make their partnership with you break even for each of their car on average?
Yes, that's a great company. That's a great question. And part of the answer is in the fact that subscriptions are a month to month access. So that allows them to manage their utilization. So you know, achieving something like 80, 90% utilization is very possible, even with the pilots that we have seen, compared that to a rental operation, which is mostly around days, and has a lot less utilization. So that's about about 80, 90% of the realization that makes sense for the kind of Fleet that we have. We also have, you know, Pre Owned vehicles, mostly on the platform that also helps towards the unit economics
Make sense? Thanks.
So rental cars are still a small portion of the cars being used in time dealers are where it's at, why would they want to do this versus leasing.
So this is a couple things have happened in the industry that that make now the right time for subscriptions. On one hand, you have the consumer demand, which has been shifting thanks to ride sharing, car sharing where people consider vehicles as a utility from rather than, you know, something that they own and maintain for the most part. On the other hand, leasing actually has been increasing very highly over the last three years.
Today, it accounts for a third of all the new cars sold. But the problem with that has been, there is now a glut of lightly used inventory that's coming back from leasing. So where subscriptions make a really great sense is deploying those those inventory those assets other than taking him to the auction and losing money on them. So both the demand and supply and because people want a flexible ownership model, like Uber and left, but without, you know, the extra cost and the hassle steps where subscription makes sense.
And what data Have you seen to suggest that people are more willing to subscribe to a lightly used vehicle, then lease a lightly used vehicle.
So in our operation so far, as well, as, you know, market research, the number one reason that people give for subscribing to a vehicle is flexibility. So it's not that they want a newer vehicle or the latest features. But the fact that they can get out of a subscription when they don't want even if, even if that's not something that they know just yet flexibility, the number one reason so a lightly used vehicle, we're still talking about, you know, two, three year old vehicle, which in today's
manufacturers products are really great. So that is the reason that we believe consumers want flexibility more than anything else.
What's the trade off flexibility for price?
Yes, so we like to say subscriptions are 30 times more flexible than leasing, right, as a monthly basis, it depends on what the operator wants to do. But that's what you're paying for. And you pay extra are depending on the kind of vehicle you're driving and sort of term that you're agreeing to. But you're paying for the flexibility that you have over a 36 month lease, which is a typical arrangement for most people.
It could you just talk about what that premium is,
it depends from car to car. So like on a on a basic sedan, it could be about $100, $150
on an on a on a base of what so yeah, so like if you're looking at a curl our camera whereas you know, on on STV on a on a Cadillac might be different. So it's really according to the operations.
So $150 more than 100, $500,000 a month some
so like if you look at the overall cost of owning a car, not only the car payment, but the insurance maintenance everything you do, what about a $400 car you probably paying about 550, 500, 550,
which is which today is the margin that the operator has, which will probably go down over time. But this is what people are willing to have been willing to pay. This is this window between the leasing and you can also go rent for a car for $1,000 a month. But that's that's sort of the balance that we have found in terms of consumer willingness to pay.
So it's about a 30 35% premium for the flexibility
about 20, about 20, 25% is what we find out typically based on
50 on 400 is 100 over 100, 400 years. So
one question you, um, you don't own the customer because it's white labeled, right. And you don't have any assets because you don't have cars. So what's to keep you from getting really your margins squeezed and not having leverage going forward as the business scales.
So if you think about this business and operating subscriptions effectively, at scale, it really comes down to the following cars are high value asset on wheels, they depreciate quickly, they need upkeep and custom need to be qualified to drive and need to treat their cars responsibly. And that's where technology plays a key role. And it's really boils down to fleet management on one end, so things like utilization, maintenance and overall efficiency, as well as understanding the value cycle of the vehicle depreciation residual values, and that's exactly the background that we bring in our building into the subscription platform.
As long as the data that comes out
as a company's on the data, we have access to the data for doing things like modeling demand, because we have a very unique access to full time mobility. So those are the things that enable us to build a really great platform that becomes the fabric for subscriptions. And if you think about the sort of organizations we need to work with, for this to be the third pillar of car ownership next to buying and leasing is typically large companies that have relationship with customers by providing the technology and by providing the fabric for subscriptions is where we see the biggest opportunity to gain market share.
All right, give it up for karma. Let's bring out our second startup of this battlefield. That startup is stealthy, presenting for stealthier, probably birdwatch and outs Carrera
you don't own your data. Every time use one of these free tools to send a message or like a photo or even get directions you handing over your data and privacy.
Google makes an astounding 200 plus free tools to track, collect and monetize your data every single day.
People are starting to see behind the smoke screen that these companies value profits. And now privacy.
With the advent of decentralization, we can finally take back control of our data and our privacy
just this year, over 2000 new decentralized apps for creating and the need the sharing and communication technology we're building.
Introducing. stealthy, secure Private communication between people and applications start these a decentralized messenger with built in depth integrations think we chat except decentralized, secure and without censorship from big governments or corporations
starting give you true data ownership. You can share your data with applications you trust and revoke it from bad actors. There's no middleman. So people pay for services they value people moved to demo please
when you open starting on the phone. There's three tabs profile dap integrations and messages in messages, you can chat with other users in a public channel, for example, the TechCrunch channel where you can talk about disrupt, let's switch to one of our contacts. Here you can send secure Private encrypted messages, and even dap content from our partners.
One of our partners, graphite docs is a decentralized Google Docs alternative.
When Alex sends the graphite document to his editor, she can open the document, edit it and make changes all in real time.
So you'll see the changes Alex will make on the left phone appearing on the right hand side momentarily. And this is without leaving stop the inside
another app integration we have is travel stack a social media platform for sharing photos with travels that you can send your photos your friends and family and they can view comments and likes all from within stealthy without having to switch applications.
The travel stock photos stopped the messages. graphite documents all have separate encryption keys, and work seamlessly together inside stealthy.
This is way beyond a simple data browser. This is the power of the multi protocol to bridge multiple applications in a unified messaging experience. presentation, please.
Our protocol only touches the blockchain for verifying identity. Everything else is off chain. So this gives us near identical performance as centralized services with the benefit of decentralization
that developers can use our protocol to bring their web experience to a mobile environment with an existing network of users.
centralized competitors like we chat I'm notorious for censorship in this image. The message on the left never gets the recipient, right.
decentralized. Competitors are either hard to use, don't fully optimized for the blockchain, or still in beta stealthy is built on a scalable blockchain where people and applications are connected in one app. Next,
Alex and I are two person team. In February, we won the global communication hackathon, and in June, we started integrating depth on mobile. Today we're launching stealthy for iOS and Android under the end or the first be centralized messenger available on both platforms. We also have dap integrations in these following data security and privacy verticals,
we will make money using App rewards mining afterwards. mining is a mechanism set up by block stack to reward the most popular applications
similar to have Bitcoin paste miners to process transactions blocks that pays applications to add value to the ecosystem.
Our protocol is the earliest and already integrated into other popular depths.
We could bootstrap on these revenues alone. While longer term we plan on charging fees for premium services.
blockchain technology is entering a period of massive growth and potential think internet in 1994
our technology does it simply proclaim Don't be evil, but it's built on paradigms that can't be evil,
centralized companies will continue to destroy our privacy and infiltrate every aspect of our lives. See the 2016 elections for example, that's their business model.
The choice is ours end up in a surveillance nightmare that's gripping other parts of the world today. or download stealthy for iOS or Android and take back control of your data and privacy.
I'll kick this off as a recovering Data Broker. I'm I'm a fan, I want you guys to win. The challenge I have is what do you think these privacy concerns that you highlight and censorship concerns which are spot on can really overcome the lock in that we all have to the services that we use every day? And how do you a road that it sure so you'd be surprised. For example, in Iran, when telegram has been they were looking for other options, so flipping countries or flipping big regions, because censorship is such a hard problem to solve is not that far fetched. Another example is AWS has gone
down, slack goes down signal has gone down. So there's central services that go down. And since we're decentralized, it's much harder to take down our services than is to centralize competitors. So when people require applications to be up all the time, they will need services that we offer.
So as the go to market then not us, is it more oppressive regimes around the world its global, it could be sure, but yeah, boil the ocean is not a it's not a go to market? So is there certain areas where you want to go first in order to bootstrap this thing and get momentum? Have you thought about where to go first? Yeah, so
places where privacy and data security matters. So journalism healthcare, so we've had talks with people that want to use study where doctors and patients communicate securely, so they don't have to store patient data and be worried about HIPAA violations. That's one very interesting use case. Also legal people that don't want to use like a centralized store for storing all their legal documents. Can you start the for example, with a blocky sign integration to be able to sign documents and hold it in their own storage?
How do you think about competitors that already have some of these attributes? Right, so I use telegram for people in the crypto community I use signal to talk to security people in journalists and there are large followings behind some of these already. So signal
is just messaging and we're way beyond just messaging. messaging is a starting point. So we look at ourselves as adapt workflow with messaging integrated and telegram is not decentralized and they want to be decentralized. We believe that we add value just beyond messaging, where you can do your entire day's work through stoping. Whether it be writing documents, or paying for services or signing documents. You can do that through stealthy, so messaging is the core attribute. But there's depths that work around that Do you
have a user growth goal as a team in those sort of coming years? Because one of the challenges as you point out, I think your charts at 20 or 30 million crypto users. But in my view, there's a there's a significant difference between somebody who's bought Bitcoin or Ethereum and somebody who's going to transition to a decentralized app ecosystem.
So a good question for you is how many decentralized apps that aren't games or exchanges are on the app store right now. And we're one of the first decentralized messaging applications available. And all these people that are buying crypto and they're getting involved in the decentralized space don't really have live working application to us, and they have one with us. And the advantages, we still have the centralized performance with the decentralized built in solar hybrid about to
add to what Obama saying, I think you can't just build a copy of a centralized app and be successful, you have to give an experienced that superior and today I dive into different apps to check my messages. And it's kind of a nightmare. And if we all work together and have an experience where you can go to this app and share that data and you don't have to jump around that will be a better paradigm to operate under kind of like jabber was in the early days, but now integrating app data as well yet secure so that if one happens to get breached, it doesn't get infected across the board like a centralized service. But the
Sarah's point, even at the protocol level, there's whisper and there's matrix and so there's these other protocols that are also competitive. And so how do you stack up against them?
When you look at whisper today, and specifically, we're talking signal and possibly also what's up in the signal space, you'll see that they have outages, it's on their Twitter, they have they go down because they're centralized with WhatsApp. Again, the encryption might be searchable in a sense. So your data even though the incomplete encryption is not visible, complete conversation. Portions may be in your data may be sold. There isn't visibility into that there's other blockchain protocols. I mean,
there I Mercury is one that actually has gotten some traction. And so at the protocol level, there's, it's sort of this tower of Babel that that you all are are vying for that attention. And how do you break through
Well, it also there's a difference in the blockchain you choose right so for example, a good parallel a status who's building on theorems blockchain and sent on June 1. blockchain right and we're offering so only identity and storage look up is stored on the blockchain so our performance significantly better on other competitors adamant who has their own blockchain, but every message takes two to seven seconds to be delivered. And that's not people aren't going to switch to a decentralized application and sacrifice the conveniences that they are already used to. But with our application, you have to sacrifice that. And you can be secure in the fact that your storage and your data is not being sold. sold for ads. For example, the last question,
I had a question on the screen,
you showed us the stealthy app. And that's the way to access all of these other gaps in your ecosystem. How are those partners thinking about working with you? They're essentially giving up the experience directly with the user. And they're, they're having to go through you. So what's the appetite? Do you know? How do you bring those people on board? And how did they see the trade offs and working with you. So in
decentralized world, since the applications don't control the data, they have to differentiate themselves by giving the best user experience possible. And if you're going to be building communication on top of what is your bread and butter, you kind of fall behind. So you build upon protocols like salty, so you give the heads Head Start to yourself and focus on what makes you amazing. And if you saw in the demo, we don't actually just we use graphite inside studies. So you still get the graphite experience. You can still access graphite from a website, but they've partnered with us. So sharing and collaboration is instantaneous, and it makes sense to messenger. Alright, one
more round of applause for selfie.
All right, let's bring out our next startup. D ID presenting for D ID or Gail Perry and Masha blender.
Photos contained biometric data
using them with face recognition.
Anyone can track you hack your devices in theory identities. You know what, I'm going to snap a selfie right now
we are the ad and we developed an AI to protect organization's databases. We protect their photos and videos from face recognition while keeping them similar to the human eye. The Face Recognition market is booming. And whether you like it or not, it has made our faces are identifiers. Face Recognition is widely used in the US retailers are using it to analyze our shopping behavior. Our age, gender and ethnicity been identified for marketing purposes. And it is also used by banks, social networks, and even in our phones.
Governments use face recognition to identify people in protests, and some are even using it to rank the citizens behavior.
We've moved too fast with facial recognition. And it is now a threat to a fundamental human right to privacy as declared by the UN.
Now, privacy regulations, such as the GD PR are excellent policies. And they address facial images as the most sensitive personal information which organizations must protect, because otherwise they will be facing huge fines and lawsuits.
Now, almost all companies store photos of our faces, and they must protect them
because I like passwords,
you can change your face.
So reality is scary, right?
Well, that's about to change. Let me introduce the ID, the same photo without your face print at the ad protected photo is visually similar to the human eye. However, face recognition algorithms can't place it. Also, you cannot decrypt it or reverse engineering. It's like a one way function
proprietary algorithm combines the most advanced image processing and deep learning techniques to re synthesize any given photo to add protected version of it. Let's move to the demo please.
On the right, you can see the selfie just took and on the left my facebook profile photo not protected. Now you are using Microsoft's face recognition algorithm which is widely used by Fortune 500 companies to try and recognize who is in my selfie. And as you can see in the small letters below, Microsoft has successfully recognized that the person in my selfie is the same one I see my facebook profile photo in Microsoft cannot safely declare that the person in my selfie is me, Gil Perry. Now let's see what happens when we replace my facebook profile photo to a da dee protected version of it.
You're about to witness the future of privacy protection. Show them
now you probably notice any difference.
That's a man
who is that guy
for AI. There was a huge difference in as you can see in the smaller is below. Microsoft now thinks that these two faces belong to two different people. They didn't recognize me. And what about Amazon?
They didn't recognize me as well. The idea is the first and only available solution in the market to protect against face recognition. Let's move back to the presentation please and I'll talk about the market.
Let's move to the presentation. Please.
Our first customers are those to store you for those clouds, storage, social networks, banks, health care and biometric databases. Government's using the ID. These organizations can comply with regulations prevent crippling fines, we privacy leaders and most important they can guarantee their employees and customers privacy and security
with annual pricing ranging from 40,000 to over a million for large enterprises. A conservative analysis for market size is $6.4 billion. In our first customer is cloud scenario, an image and video management solution which stores for us have more than 350,000 companies more than 22 billion media assets. We've also signed agreements with top players in the finance and automotive industries.
I'm happy to announce today that we are publicly launching the ad in our software will be available as SaaS and on premise solution. Our business model is annual licensing and paper us
We are ycombinator alumni in our team consists of the up deep learning computer vision and image processing experts in Israel, which all hold master degrees, and PhDs like Dr. You have a cocaine over here. And while serving in the Israeli Special Forces in intelligence cope, my co founders and I will not allow to share photos online for security reasons. And today, everyone must deal with this problem. Now is the time to protect your data in da da is here to make sure it happens. So join us visit the d.com and be privacy leaders. Thank you
So as a as someone with a forgettable face, I'd like to keep it that way. But how do how do you keep up? It seems like it's it's an arms race, right? Today Microsoft can't match. But they're changing AI is changing AI is becoming more like the human mind. So
as these technologies advanced, how do you maintain the competitive edge on recognition? Good questions. So
first, what we saw is that the more advanced the face recognition is it easier for algorithm to full it second, we're constantly learning and adapting to new algorithms and implementing them, you know, systems. And third, we're doing that we have a red team that constantly forced to attack the ad protected photos. And then before our algorithm produces a new day, the protected photo, it needs to fool our own red team's face recognition algorithm, which has the most the most the photos went through through it. So it's the strongest, then that's why our algorithm constantly changes. And once we fill our own race teams, face recognition algorithm for sure, we will fall all other the physical condition algorithms which are trying to overcome or protection but you
have to maintain that it's human recognizable, so I could see over time it's going to become less human recognizable, it Do you see that trend happening? No,
actually, it's not the more physical ignition algorithms that are advanced, then the less changes we need to do it before they say it's a different person,
as a consumer.
Sorry, growing up as a
consumer, I really I love the pitch. I'm like, this sounds great. What is Who are you selling to India to enterprises? What's their incentive outside of maybe CSR and being able to say they protect their users data?
Okay. So just first, the, the beginning the vision for the idea was for consumers that the visuals, privacy for but really fast, we pivoted, we realize that the who needs it the most are the organizations whose suddenly found himself storing such a big amount of biometric data because all companies still photos a I can tell you exactly the which what each person in the organization wants. So security information and security officers a they're in charge of the sensitive data and the they need the protection product managers, they want to increase revenues make a better product marketing want to improve the branding as privacy leaders legal, they want to comply with the GDPR under the GDPR face images are now considered not just personal information, but sensitive personal information. So that's basically Do you have any enterprise customers today?
Any Sorry? enterprise customers? Yeah,
so first customer, we just completed delivery, successful delivery to cloud scenario, these two photos of 350,000 companies, more than 22 billion media assets there customers need to comply with the GDPR we're now starting a PC with one of the largest one is one of the top players in finance industry, a really big conglomerate.
And then we also have a commercial agreement with one of the top automotive players which a automotive eventually they need to protect the district's which a Thomas class would capture. And that's a really big concern for GDPR and lawsuits.
So far, all three companies since like GDPR, compliance is the number one motivation from the do this
GDPR compliance. Also, everything else nice to have, but very immediately want to be able to comply to that regulations seems to be the number one motivator, yes, actually, if one of them is really interested in our next product, which is also proprietary and disruptive, which will finish the completes our vision and really fix the face recognition problem is what we call VI, d plus, plus,
in which, which will enable organizations using the idea to still be able to authenticate their users, right, and I'll explain in 20 seconds how it works. Basically, if a front end they will use it. So when you enroll in, then we store the D ID photo with instructions, okay, then when you come to open the phone, again, the ID with instructions know how to transform your face without storing it into the particular domain, but and then doing face recognition algorithm than doing physical mission, then, if everything is stolen, no damage is done. But you can go back to the real photo that's going to disrupt the face recognition market, I'm really gonna fix the privacy issues.
So I this this is related to my question. There are legitimate uses that companies have facial recognition.
they applied the ID to all the faces, then there unable to use that internally as well as it is. And so what you're saying is, then they have to divulge your by your second product, right? Yes.
So it's a great question. Really?
That's a great question. So we thought about it,
you know, at the beginning, we will not blocking face recognition. And then we'll get many requests from many banks, especially banks, and also governmental organizations,
and also face recognition vendors at the beginning, we told them, Listen, we are doing the opposite. But then when this today, they have the most pain because they are the most aware of the risks GDPR consumers concerns and they are aware about this, since it's like a double edged sword. You don't want you want this data want face recognition, you don't want other organizations to hold it. So that's exactly what we made. We had the problem we need to enable them to use a face recognition a while using even though they use the idea. But we have two constraints. It's not we're not going to decrypt the data. We're not gonna reverse engineer it. It's not going to be a backdoor like a once you da da da da da. And we solved it in with the plus plus
all right, give it up for di D.
Bringing up our next startup. Pulte with Interpol, Dr. Russ more costly and Ed ciao.
tracking things is tough. Yet in today's data driven world, knowing the location of things is more important than ever take the logistics industry we're over 22.8 billion pallets are in circulation in the US at any given moment. But that 20% of those pallets go missing every single year. In addition to $30 billion dollars worth of facets going missing. Today's trackers are too expensive, batteries don't last and they're just not reliable. Tomorrow is trackers need to be super cheap, have batteries that last and work everywhere securely.
The location market is huge and growing. By 2022, there'll be 29 billion connected devices that are looking for location market estimates have the outdoor location market. Together with the indoor location market worth $124 billion growing at 30% every year,
introducing Pulte, the world's first fully cloud based location over cellular platform. We power 4g and 5g connected devices to have accurate location everywhere. While also enabling the lowest cost possible and the longest battery life. Today's location market is fragmented. Requiring multiple to provide indoor and outdoor location, Pulte is your single source for accurate location everywhere.
So how do we do it?
Unlike hardware based location capabilities like GPS or Wi Fi, Pulte is a software only solution that enables 4g and 5g connected devices to be able to provide accurate location both indoors and out.
We have a cloud based location solution that that
that because we move our computations to the cloud, we offload devices from that heavy compute cycle, allowing a battery life to be able to go to 50 times better than competitive solutions. In addition, since we're firmware only on the device, we get rid of all that extra hardware enabling 80% savings over the competition.
We have 74 Global patents and patents pending the core of it is something called Super resolution super resolution allows you allows us to get rid of the error inducing multi path that that allows us to be able to get to the best line of sight estimations to their surrounding cell towers that are around all of us today. This is this is this is the best solution over cellular Three, two times three to 10 times better than any other cellular solution, including ot Doa, which is using by us by today's e 911 systems.
In addition, since we're in the cloud, we collect information data from all of the connected devices on our network. This allows us to leverage machine learning and develop heuristics that really take care of the tail, which is the hardest part and location solutions today. And we get better as we scale. Let's take a look at how it works.
We partnered with a local logistics company to embed poultry enabled trackers in their palates.
This allows them to track their palates, whether they're in their warehouse or on the move, they can even track down pallets if they go missing. In fact, this one by Tyler, one of our engineers, he thought that this would make a great coffee table for his new apartment.
Let's go to the Security Operations Center dashboard. See, let's track him down. Since we're real time we're able to trigger when Tyler crosses a geo fence and send an alert to security team.
When we press the the geo fence we center in on that asset and track them down. If this is Wi Fi or Bluetooth, we would have completely lost visibility because when they got away from the warehouse, and since we're using cellular, we're able to track that asset indoors and out as you can see both in the warehouse and on the move in the back of the SUV in GPS, they would have completely wouldn't have been able to do this because you wouldn't be able to have any visibility to satellites, we leveraged 10 million plus LTE base stations deployed globally
to to to create the best location solution on Earth, we bring GPS down to earth, we sell our solution as in a SAS business model,
and a pay as you go model that allows you to add location to your to your application as simple as just adding our API to your code. This allows us to be able to be soft since we're software only relative the competition you have to buy the chip the GPS chip of the Wi Fi chip where software from day one allows you to grow as you use
our partners are all in this because this is a great value problem from chipset to the cloud and system integrators we are working with the following partners to be able to go to market over the by the end of this year sequins and ride on the chipset side sir calm and nimble Inc. Who are palate tracking. And as personal tracker solutions, including this one from circa, just press the button and you can find where you are.
And logistics is just is just the beginning. We're we're working with medical device companies to be able to locate their their devices, personal emergency buttons, manufacturing, and IoT, as well as getting the integrity of location for supply chain.
And today and our team is ready to go. We weave hundreds of years of experience including 12 startups and six successful exits.
And today, we are introducing our system integrator
program. We're inviting chipset partners, device manufacturers and system integrators to come join us to be able to add the best location possible to your solutions go to Pulte calm battlefield today we'd love for you to join us. Thank you.
Can you talk a little bit about the accuracy and availability trade offs versus GPS? Sure
where we're able to get down to single digit meter accuracy for in building solutions. And then using just the macro cellular network, which doesn't require any small cells or anything like such as inside the Marconi center were able to get to sub 15 meter accuracy.
You mentioned you have a lot of great partners on there. Have you deployed with any customers yet? If so, how many how, what's kind of the size of that deployment. So
we're building our product right now we've gone through two proof of concept trials with multiple tier one mobile network operators, in addition to working with each of the different vendors, chipsets device partners to validate to make sure that the software that we put on their platforms is ready to go. So by the end of the year, when we've also completed our initial beta platform for the cloud, which has secure security, as well as full redundancy capabilities. And we'll be able to be able to launch by the end of the year, a bigger team do you have? No, we have 16 people, 11 engineers and five people who are just business development all day long. So if I have,
so if it's logistics, right, that seems to be the use case, if I put one of these devices in a in a container at the factory in China, and then I want to track that
all the way to some dc in the US. Is there any point where I lose visibility? Sure, the what we're, since we're terrestrial, a terrestrial only leveraging the 4g and 5g signals that are out there, we're only where there's signal for LTE. So going over the ocean, or if you're in the plan, but what we do is we're able to hybridize with other solutions to be able to provide full visibility, and once you get from port to port, or into the airport entry points, we got you from there. So do you hybridize with GPS, then we can, but we actually really think that we, for the most part, replaced GPS for most of these cases that were that we're looking at. However, GPS, you can't beat in certain areas, if you're out in the middle of nowhere, where there aren't aren't any other cellular signal. So for those particular use cases, we can hybridize so you don't get the hardware savings if you have to rely on GPS as
well as your for solution, correct?
Correct. One thing that we do do is we can make GPS better. So one of the biggest issues with GPS is startup time. And one of the reasons why the battery life goes away so quickly is that you're constantly looking for satellites. So to the extent we can provide a cellular system location, that can actually reduce the time that GPS can actually acquire signal as well, at the end of the day, we believe that are good for from our go to market strategy standpoint, we believe we can open up a whole new category of use cases because we can reduce the cost down to the single dollar range for tracker by getting rid of the GPS
from a maintenance perspective, what's the time between charges then with this lower power solution?
So depending on the use case, so that palette tracker is an example. We can go from one to two years, typically, they'll see four to six fixes a day, they want to know where where a particular tracker, particular palette, maybe or afraid might be of during the day, and they're typically sensitive triggered, right? So if there's some sort of motion detected will take over from there. So four to six times a day can get you to a year plus of battery life. So you
fell on the ocean? Like roughly what proportion of us would you have enough 4g 5g coverage to get to get coverage? So
if you see the map commercials, it's 99% of Americans and the public. And so one things one things we realized that 99% of the population population which is different than the yarder graphic coverage, indeed, and do roughly what it is, what is it of the actual topography? I don't know exactly. So at the time, when I was running the network for Metro PCs, and we added all of our roaming partners, it's roughly about 70% of the overall geography in terms of square mileage. But when you think about the use cases that we're enabling, where you're bringing containers to the edge of population areas, because you're delivering them for big box retailers, or the like, generally, that's where you're going to want to know where things are. Right. So when you're in or when you're in transit, on highways, right. So or even during the rail. Right. So a lot of when we built our networks, we also covered the railways because that people and want to have connectivity. So if you think about the coverage of where the use cases are going to be, we think it's a lot larger than that.
How much does a tracking chip cost? And how does that relate to, for example, how your customers would think of the value of a pallet or the cost of GPS and
so we've we've been working very closely with a number of different ships at partners. If you look at some of the things that are coming out that are LTE IoT based profiles, so there's something called narrowband IoT cat m, these were variants of the standards that allow the chipsets become almost 10 times cheaper than they've been in the past for your your smartphone so in China one of our the chipset partners they were talking about they're talking sub 50 cent chipsets for LTE connectivity all of us are e commerce investors that's where all the question towards you are about tracking the power from China to us enough understand understand
all right one more round of applause for multi factor
alright we have one more company in this round of the startup battlefield that company is McCarthy was ending from McCarthy are Nick White House and Richard differences. Go Come on out guys.
Our world is getting more complex for individuals like you and I protecting or even understanding our legal rights as hard this problem is massively multiplied. In business. There's no means pressure on businesses to move faster. But to do so they need to navigate layers of policy confusing laws and changing regulations. With all this confusion. You think it'd be a great time to be a lawyer right? But many see law is slow, expensive and intimidating. Court rooms are at capacity law firms are writing off work and in house legal teams are ignoring risk. Just to keep up manual legal was losing shear and relevance to growing legal tech industry that has over simplified the problem glossing over the nuance that is so important to law, McAfee ventures, changing how law works. Leading PhDs and a swarm of lawyers have built author, a highly advanced legal AI platform that is arming the armies of the industry think AWS or Azure or follow this platform combines 18 plus custom pre trained learning models that can be configured and integrated into any legal process. These algorithms allow lawyers to virtualize the expertise and automate all manner of legal services. Let's see one example of us now move to the demo
rich has got a time critical project and it's stock he's finally been sent a contract by his supplier that's a large and complex document
sending this to his legal team take weeks to get back sending externally will cost thousands if not the same amount of time. Instead, he's going to use author to do an initial QA and read line of this contract author intelligently reads contracts, creating structure from unstructured language, classifying the contract type identifying and labeling clauses and details author uses its higher level inference engine to CES this for issues and risks
in this example, author has found a problem with this contract. If it didn't, it would automatically proven in the document you can see what author is looking for and the issues it's found. It's important to note author is telling you why something is wrong and giving you alternative solutions. Zoom in just or
explain what's going on here. Yeah, sure.
Absolutely. So on the right hand side, you can see that there are closes details and issues, these clauses are the different things that author is looking for the details of the different details that are looking for such as termination periods, and details like that closes at the different different causes, the issues that author is found here, there are three issues that it's found. The first is a term is no term clause in this contract. So authors suggesting to term clauses to be inserted and I believe are appropriate at 24 month in a 36 month term clause. So for Richard and sets that you'll see that will drop into the document.
The next issue it's found is that a termination notice duration for this contract is too long. So 90 days, breaks the corporate policy that we've set refresh the author suggesting 60 or 30 days. And the last the last thing that author is notice here is it's read the liability clause, and it's determining that this liability clause is to unfavorable. So that's legal reasoning. And it's rewritten this clause in both favorable when a neutral wording and so if we just got hit with neutral wording and answer that then what you'll see is the contract has been approved. And we can now download this and send it back to the supplier or the markups so it looks like she's taken weeks long process and completed it and a matter of minutes. If we can move back to the slides
built on 10s of millions of data points, legal data points, and 10s of thousands of hours of legal expertise. Author can read, write and reason like a lawyer, the author platform has mastered nuance unlike traditional AI techniques, our patent pending technology understands tiny details and language that make all the difference in law authors. Human comparable learning allows us to learn from an incredibly small data sets as you work and most importantly, authors outputs are entirely explainable, no black box making result, defensible and trustworthy
competitors have a point solutions that tackle part of the problem such as contract analytics, due diligence and document discovery. But the AI fails to enhance the knowledge base of any specific business. Creating generic results role in contrast author enables your intelligence to be shared across the entire organization, giving you agency over your data, your training effort and your competitive expertise
authors been used to reduce the legal effect drafting by up to 70% completely automate court decision classifications help resolve legal air queries by up to 60% and give a legal Market Insight up this 99% faster.
Legal tech is a $30 billion industry we serve right across legal risk and compliance combined. This is a trillion dollar market. And we've partnered with Accenture and digital endure integrators such as the Lloyd to rapidly transform the space
our pricing model has three components a platform access fee, volume based API charges, and an optional app fee if you're a legal service provider, lawyer or in house legal team and you're looking to improve profitability and build velocity go to disrupt that McAfee Finch calm start building the future of law.
So when you say the liability clause in his contract is too long, you're making a judgment about a deviation from a standard. Where does that Stanford come from? So that documents do I configure a set of policies best not
a standard that is we're leaving, we're actually reading the language, we're determining the parties who's involved in that understand the context of that, and we make a judgment on that. And that threshold hair can be curated by the legal team in terms of where that right, so
that's a policy that you as a legal team have set. So that's what I'm asking. So how many of these types of judgments or I would think of them as configuration? Yeah, so
so out of the box, there'll be a this is a number of things that are just there. And what we found in the legal space is that every lawyer has a different tolerance level. And so we allow that to be configured. But a number of those things, are
we talking 10s, hundreds, thousands, 10s of thousands of configurations will not talking now we're talking about for an individual document like that. It's team No, but this is something that is going to be extensible across a lot of different document types, right? That's the whole point. So yeah,
absolutely. So so there are these both implicit and explicit kind of inference, right. So and then implicit information based on a data set based on things that we we identify. And so that takes care of the large amounts of configuration they want to do. And then there'll be 1015 things that you might want to configure for a specific agreement in that example,
is there a subset of industries or types of contract that you're starting to train your model on to get this level of flexibility?
Yeah, so sorry, I guess we're not just focused on contracts. So we're far as a platform we're reading we're writing we're problem solving with thinking like a lawyer. So what we've seen that that covers both insights generation and capture that covers contracts, we call that review. So being able to review approval fits into review drafting and also triage, which is conversational AI. And so we're, we found that is predominantly in house teams in the corporate space legal teams in the corporate space, really interested in stuff as well as can consumer base legal tech firms, and also large system integrators. How long how many engineers and data scientists have been working on this. So we have 19 people in total, across New Zealand, and America, and predominantly with PhDs for PhDs, legal engineers and developers. So
that's 19. Yep.
And then we asked to fit into that. So
19 total employees,
19 title 17 the developers.
So I don't think anyone's gonna have a problem with putting lawyers out of work, the good thing, but a lot of this data is confidential, very sensitive. So it's, it's a cloud service on prem. And how do you get in this gets to the the points raised earlier, how do you gather the data, so the integration post sale shortens over time and doesn't carbon editorially explode? Yeah. So
I think that's, that's a good question. And, and data is one of the problems that we had going into this, we saw that deep learning approaches, we're going to always cause a problem and law for a number of reasons, explain ability to be able to, to get to nuance, and the ability for people to be able to train on bespoke cases, a lot of law firms and house teams don't have 30,000, 10,000, 5000
examples to learn something. And so what we've actually built is human comparable learning, which is very much about incredibly small data set to learn. So in some cases, our models learn from two to four examples. And then they don't ensemble approaches, we can bring in data that transactions are there to to boost that over time. And so by removing the need for a huge amount of data, we actually start providing value quite quickly, without this challenge of privacy in the collection of a bunch of data. We also have a number of different tools around deity, the identification of data as well, so that you can start pulling data and that removes identifying features. So if
you're, if you're doing learning off such small data sets, then how do you prevent overfit.
So that's literally that is turning tuning. And that is the hardest thing that we deal with. So it's human. It's human oversight. Yeah, absolutely. That's why we have a team of legal engineers. So everything we do is a legal engineering engineer or a lawyer. It's both. So it's a it's a lawyer, who's actually an engineer,
our someone that has a law degree and a degree has been
has been working and take
a look. Illegal engineer is a lawyer who's worked in tech. Yes,
I understand the technology understands what our PhDs talking about talks, the language can read the code,
so there's 10 of those in the world.
And we've got you've got a handful
one of the audience have you roll out your services? And what are the initial contracts type of contracts that you're focusing on right now. And so
some of the use cases we have so that just contract some of them have been around extraction classification across large funds and large industries we've we've seen, we've done things around financial advice in monitoring the financial advice that's been given out from policy and corporate documents like that we're working with government and legislative spaces. And we're also working in sec and house where we're seeing things like sales contracts, so sales,
how do you charge them such as such a wide variety of services? Yeah,
so what we tried to do with our charging program is to try to make it as easy to acquire as as, as possible, try to make it a standards we have a platform charge, and then we have an API charge which is basically based on volume All right,
give it up for McCarthy
of the startup battlefield all of our judges are gonna follow me now backstage they're gonna deliberate help us choose our finalists. They do let's give them one final round of applause.