Startup Battlefield Competition - Flight #4 | Disrupt SF (Day 2)
3:16AM Sep 7, 2018
Emeric de Waziers
Ladies and gentlemen, please welcome back TechCrunch, senior writer and your battlefield host Anthony Ha.
Hello again. We are so close to being done with day two of disrupt. All we have is a final session of the startup battlefield. These are the last startups you're going to see before we choose the finalists. They're gonna be five companies and then a wild card company. Once again they're gonna get six minutes to present six minutes of Q&A and the finalists will be announced tonight and then come back tomorrow and you will get to see the finals competition.
That's all you need to know. Let's bring in our judges. First up we have john Borthwick founder and CEO at betaworks where he's led company building and investment process since 2008. Previously, he was a senior executive at Time Warner photolog, AOL and elsewhere. Next we have Julie grant partner at Canaan partners where she leads investments in biopharma and digital health. She also serves on the board of the Biden cancer initiative.
Next, we have Daniel gross. He's the head of artificial intelligence at Y Combinator. He was the founder of Q, which was acquired by Apple in 2013. This year, he founded pioneer, a search engine for a search engine for creative people. Next, we have Susan Lyne. She's president and partner at bbg ventures. She was the CEO of Martha Stewart Living Omni media, also the CEO, then Chair of gilt.com and she led AOL brand group which includes TechCrunch.
Next we have Rebecca Lynn. She's co founder and general partner at Canvas ventures, where she focuses on early stage investments in AI, FinTech, digital health, SaaS, and mobile. And last but not least, we have Lo Toney founding managing partner at plexo capital, which is incubated and spun out from gv. Before plexo and gv he was a partner at Comcast ventures. So let's give it up for our judges.
I'm not clapping for
Okay, let's bring out our first startup that startup is unbound. Presenting for unbound are Polly Rodriguez and Julia Lopez.
Sex is hard to talk about. At 21. I found myself with a Velcro fanny pack full of chemotherapy drugs standing in a seedy sex shop next to the airport in St. Louis, Missouri. my ovaries were nuked from radiation treatment following a stage three colon cancer diagnosis.
My doctors told me that I would never have children but failed to mention that I was also going through menopause. I googled my symptoms to find out I was facing a lifelong struggle with arousal and sexual discomfort. So I took it upon myself to buy a vibrator and some lubricant and I can so vividly remember how awkward I felt to be meandering through the rubber penis aisle of that seedy sex shop wondering what the hell am I doing here.
This is not an experience that is unique to young woman battling cancer. Buying sexual wellness products as a woman is the worst. Most products are designed by men from the male perspective of what they think women want.
There's also a remarkable void in educational content when it comes to female sexual pleasure. In fact, most people learn about sex through porn. This impacts not only the product design, but also the shopping experience. If you go in store, you feel intimidated and overwhelmed. It's hard to know what product is right for you. And it's embarrassing to talk about your sexual needs with a random sales associate.
Conversely, if you brave the online search of sex toy, you're still faced with the hyper sexualization of women and most likely a website that looks like it was designed in 1994.
Despite the terrible shopping experience, the industry is booming. Turns out female sexual wellness is just an overlooked market because women are still shamed for enjoying sex. To put this in perspective, the vibrator market in the United States alone is they billion dollar market annually. That's more than twice the size of the condom market.
And yet, there's still no dominant brand name that's owning this space
Until now, meet unbound the rebellious brand bringing sexual wellness mainstream by designing next generation products at half of the price.
We've designed and launched over 50, vibrators, lubricants and accessories. And we've written over 300 educational articles with more than 12 million page views. Today, we are so excited to share a product that we've been working on for over two years.
Introducing Palma a wearable fashion forward vibrator. This ring is reframing how we think about female sexuality. It's discreet yet beautiful design allows a woman to wear her sexuality on her sleeve without the feeling of shame. Let's show you how it works. Move to demo please.
You see 40% of women report chronic difficulty achieving orgasm which makes sense because studies show that 70% of women need clitoral stimulation in order to orgasm. This doesn't happen during regular penetrative sex.
This ring is closing that orgasm gap. As you can see here Julia is cycling through the three intensity settings but what makes Palma truly special is the patent pending built in accelerometer which allows the user to customize the vibration pattern simply by tapping the surface of the ring.
Back to presentation. Please
Palma is the first product to market with this patent pending built in accelerometer which is what allows for the customized experience with the ring.
When we take a step back and look at the company behind this product unbound revenues accelerated from 550,000 to 2.4 million last year, and this year, we're on track to do 4 million in sales. Almost all of that growth has been organic with a customer acquisition costs below $3, over 75,000 email subscribers and more than 50,000 Instagram followers, none of which were paid for
when we look at the competition. unbound stands out in several key ways. competitors, such as Amazon, and Adam and Eve sell 10s of thousands of products in this category, but almost all of them contain carcinogens. Because vibrators are not regulated by the FDA.
Smaller shops such as Good Vibrations not only don't make their own products but also offer an outdated online shopping experience. vertically integrated brands such as Lilo, sell these products but at twice the price
Unbound is the only affordable online destination that offers educational content to guide a woman to finding the right body safe product for her.
unbound direct to consumer business model not only allows a woman to shop from the comfort of her home, but it also cuts out distributors who take 40% of the margin today unbound requires its customers through PR partnerships, Instagram and opportunistic wholesales such as group and Urban Outfitters
But what truly differentiates unbound is the world class talent of our team. These eight badass women are an anomaly to the adult industry. They come from some of the top corporations, startups, universities and media companies in the world. We're also the first female sexual wellness company to garner the backing of top tier VCs and advisors.
Female sexuality should be just as accepted as male sexuality. But in order for that to happen, we must elevate the products, education and conversation.
Palma is the catalyst to that conversation and unbound is the revolution. Thank you.
talk a little bit about how you market
and then converting them into customers also word of mouth
Can you can you tell us a little bit about the first of all congrats always love seeing awesome entrepreneurs can you tell us a little bit about the economics of your business like what it actually costs at retail what the bomb looks like?
Absolutely. So we found that sweet spot for the retail price is anywhere from 35 to $99. The Palma ring retails for $98 today, on average, we have margins of 65 to 70%. But those are only trending upwards. As with economies of scale, and we grow, we're able to have even better unit economics.
And it's loud. How many competitors are there at right now? Because I have seen a couple others that have like AI, vibrators and things like that at UC Berkeley. So the competitive list I think the markets huge, but what else are you seeing?
I think that there are a lot of people that are making strong individual products. But unbound is really focused on I mean, today we have over 50 products that we've made, and within the next year, that'll probably be closer to 75. So really focused on being the one stop shop for both education and content as well as a full products that that customers can choose from.
A lot of those companies have one to two products so their LTV tends to be a bit higher than ours, which we see as $185 within the First 12 to 16 months.
Can you speak a little bit about the decision to have Palma as a wearable that you would wear in public versus your product line that could be used in private?
Yeah, I think that's a great question. So for us, we made a fashion forward line of accessories. So like bangle bracelets that transform into handcuffs, a ring that you can also use for massage, and it was overwhelmingly popular. We also have our best selling vibrator squish squish, which has haptic technology built into it. So the harder you squeeze it, the harder it vibrates. So we took the customer feedback which was these two variables and married them into this product. And we found that millennial women who our target demographic really want to wear their values on their sleeve. And so it's kind of a wink and a nod for women who know what it is but it's also just very symbolic for our customer.
Is it sold right now
is not solid it just launched today and it's on pre Sell but it'll hit markets in early 2019
and the just so I understand the business to 50 or so products that you have at unbound and the content the Ukrainian there it's basically the 2.4 million is content and commerce or how's that balance between on the revenue side
yes so our content is really focused on educating the consumer today porn has become the de facto sex ed and so what we find is that there's a massive gap in terms of women just wanting to understand how their bodies work and trying to really D stigmatize specifically female masturbation so the content is really a gateway for people to become informed and then the products are how we make money but today we have a community of over 150,000 women that follow us on social media subscribe to our content and buy from us directly online.
wasn't Can I Can we see the product
and While you're doing that, can you talk a little bit about where you'd like to see the company 10 years from now,
I believe that there should be a household brand name. And the same way that if you were to ask anybody in this audience that condom brand, they could probably name Trojan or Magnum, and there should be the equivalent of that for women. We really want unbound to be the household brand names so that when women are exploring their bodies and becoming informed about them, every woman in America owns an unbound product.
You tell us a little bit about the persona of your target user and which side is more important for them? Is it the ability to be able to learn about sexual wellness and better understand their bodies? Or is it more around the ability to be able to go and buy the product without the stigma of having to go to your experience?
I don't see them as mutually exclusive. I think that we often capture customers who are turning to Google to answer their questions and then they see the products in that they're not phallic or they're not intimidating and we don't have women in four inch high heels and lingerie and a full face of makeup on our website because women don't find that relatable and so I think then they realize that this is a product category that maybe previously they had shut themselves off to I mean I know I did but now feel that it's an approachable product category for them as well.
Alright we are at a time so one more round of applause for unbound
to the left
okay let's bring out our next startup that is Crossing minds presenting for crossing minds or Alexandra, Robicquet, and Emile Contal coming out.
Hi, we're all familiar with the screen right? Did you know that in average user spend 19 minutes every single time is trying to find something to watch.
19 minutes and we're 300 million viewers. That means that every year we are always seeing 5 billion hours, 5 billion hours. And I'm not even talking about music video game restaurant, just something to watch. So we are using recommender platform, right? But what is wrong with them?
Well, first, there's the problem of choice overload, winning all those platforms project way too many content, your brain freezes
to their approach or bias Facebook and Google advertising platform. They are bias Netflix, move away from five stars two thumbs up, thumbs down, because it doesn't really matter if you love or like something, what matter what is the most popular item in the database.
So of course we can relate that form six in average per person, good read for books, Netflix for movies, Spotify, for music, and none of those communicate with one another.
You don't think that you're tasting book might impact the tasting movie your music and vice versa.
But even if we solve all those problems, there is still out there not a single platforms that can give you a group recommendation.
I bet that everyone in this room face that question of what do you want to eat? Or what do you want to watch tonight, right? How many hours have you spent on that which many measures should be used to help the users in those cases, not only the company's making profit,
and this is why we created Hi,
my name is Ruby K. Dr. Emil mercantile Professor Sebastian Thrun in myself, co founded crossing mines, the tech startup that he's here to build the best recommender system and today TechCrunch we launching our beta high for pure cross domain recommendation engine
cross domain because it delivers recommendation for movie music book TV shows but also cross platform
Because what you're using matters if you're having a ps4, Spotify, Netflix, Hulu, you name it, you sync it to that is completely private and secure. This is powered by the best algorithms in declining and we are not an advertising platform Nora content provider. So our incentives are aligned with our users. And finally, we're very, very proud today to present the first group recommendation engines ever deployed.
But let's move to the demo please.
So as you can see, we're going to enter my height so my home for all my taste, likes and dislikes in give me recommendations, and we're going to ask for a recommendation for movies today. So I'm here and I'm with me. Oh, so we're going to sink or taste and receive a recommendation together. As you can see, all the movies are fine tuned for the two of us. But now let's say that I'm using Netflix and HBO Go
when we're away all the movies are filter for those platforms. If I click on one item wash, now it opens on my Netflix simply and that works for anything. Video games, music books, all the cultural taste in one single place.
Let's move back to the slide.
So how does that work? Well, crossing man's algorithms are based on three main pillars, deep collaborative filtering, to understand the depth of the user across many domains, 70 graph embedding to make sense of the tags, the actors, all the metadata and find the hidden correlation between items and deep content extraction to understand really, why was it
so the first time which I would disagree them, we brought the state of the art of recommendation from 75% of accuracy to 92% of accuracy in preventing someone states across 80,000 people.
So now the next question will be how do you make money right? Because Hi, it's free.
Free ad free. well as our algorithm serves and understand the user taste, the also finding the correlation between different items, which is then fed into an API integration service that's provided a solution for enterprise. If you want to increase retention, understand better user. So the core SOC problem were here
last year. And if he said that the recommendation engine make them save $1 billion,
how much you think we can make you save if we doing that much better, and across six domains.
So of course, many people have tried to do so or Facebook and Google or advertising platform. But the other is was interesting to see how the tackle the problem either with a to b to c to train the algorithms or a b2b to a blog them to go to advertising
here at crossing man's or algorithms are trained on the users with the users and help in the feedback loop in the crossing my new business guarantee to our partners that we provide the best algorithms trained on dynamic data sets.
And if we're so proud about the tech is because of our team.
As you can see, we're all from academia six years, each of us writing papers and machine learning, deep learning and human behavior prediction.
And if we're here today is because we believe that those algorithm should not be kept in the depth of academia normally for company, which should be used to build the best product for the users.
So if you're a company who want to increase retention, and you sell or just reach out,
and if you're tired or wasting 19 minutes, every single time you want to watch a movie Well, just sign up on a God I thank you.
Yeah, how long does it take to train high on my own personal taste?
So we build the backend infrastructure from scratch so that everything is optimized. Actually, we can retrain. Be using model on real time. It's actually once again, even less
I in terms of getting the data like in terms of understanding, you're watching my patterns and getting it right. Because I think people consumers are a little fickle and it has to work quickly, right? And you have to get the data to render the model and understand what I want. So how long does it take.
So when we just the more than 1080 people, it takes less than two minutes because we have a team of designers that worked at Ubisoft DNA and Sony and we give me five their experience at such a way that in less than a I think it was two to five minutes, we have hundred ratings. And that's, you know,
how do you get the feet of what I've watched from Netflix and Spotify.
So we don't look at that because there's no API integration between Netflix and that we are mostly allowing you to train your algorithms and
you asked me what types of things I like
there's an explicit feedback in addition to different platform nursing.
So for specific for example, you can login with Spotify and our back end of fetch all your previous history, if you will know that.
But Spotify, I don't think gives you play count for every song. So you don't know that I repeatedly play The Smashing Pumpkins as opposed to like, blink 182 that I played once, right? So I'm more broadly wondering, how do you handle this problem of kind of getting my current activity data from everyone?
So Spotify actually does that on the short term of the past. Yeah, we won't have the full history. Yeah, but on the last month, we know what you're listening.
Right. Okay. So
I guess another question I have is, what do you guys use to model your taxonomy because every one of these services has their own kind of modeling in you know, a song in Spotify could be modeled different than a song in iTunes. So what are you doing internally
so that's why we use deep learning and we mentioned a deep content extraction so meaning that every single item has An embedding which is completely learn by the deep learning architecture it could be a music it could be a song couldn't knock door from the movie we learn that from scratch
and then how do you figure out there the same like how do you figure out that IMD bs actor is the same actor as Netflix is actor
This is where we spend the last I think immune and I'd been working in this category them for years this company for a year and a half and the backend itself for two years and a half three years but that was a massive challenge that we had to when we started building this company we have to build as Damian said on the back end from scratch but also rewrite some libraries of the no and if you're familiar with the different things so that was like a tremendous amount of work to especially tackle this problem of matching data.
How do you avoid ultimately limiting my field of view so what you see with a lot of of like recommendation engines is that Over time, you start getting more and more things that are very like what you've seen before. And your world gets limited. So how do you avoid that? How do you keep showing me things that will open my mind will excite me because they're new.
So there is to answer for that one for the product and one for the take for the product we're all doing, allowing the user to define if they're more expressive mood or exploitation would, which is to aspect of the algorithms, this is something that we can fine tune and that we're actually working also on a different component that we're having on the take.
And on the technical side, that's actually very close to my PhD, which is optimization of a time. So you actually have those two components exploration expedition and you want to find one that that the user always have something that they like, but you don't fall into one what we call local minima. Yeah,
an additional feature that we're developing currently is the context. Meaning that regarding the time of the day, or the person that you're with, or even other element that define your life, we might assess in what kind of mood you you are, if you're having a terrible day, we're definitely not going to explore something like we're going to give you something that you're comfortable with, or something that might answer those needs at this
a couple of questions. One is, how do you get to the point where you've amassed enough data to be able to actually have some meaningful results at the enterprise scale level? Because you mentioned helping enterprises get over the cold start problem. So how much data is required for that? And then second question is talked a little bit about the experience for the consumer once they get to you, right, the gamification that will make it fun and engaging for them to input their data. But how do you get over the cold start problem of actually getting folks to the top of the funnel so you can even start that process.
So I will answer the first question that technical one and the cost our problem we solve it using the meta data. So for movies, it's actors, directors, we also look at the actual posture we look at the cinema using natural language processing all of that you don't need any explicit or implicit feedback to be able to do any recommendation because you know that the user like the movie where there is actor or, or when the poster looks like that anything like that is very accessible.
In addition to that when we talked to different businesses were discussing entry three men big pilots I mean be corporate they usually have a huge amount of data themselves that they don't really know what to do with its a mix of explicit and implicit feedback meaning either user say that it really love something or that you define the love something because they stuck on it for five minutes. So most of the work here is actually to start to understanding what is relevant for the neural networks or the difference algorithms and fields for that to then not even using order address like helping them with their different complexities.
All right, give it up for crossing minds and high
bring out our next startup Mira presenting from mirror or Sylvia Kang and Zheng Yang
we miss house is complicated what are the six couples fishy fertility issue was pregnant we worry about miscarriage at menopause hormone imbalances cause extreme discomfort
all these complications are heavily rooted in hormone levels but tracking hormone as hard
introducing mirror the first FDA NC registered comprehensive women's health monitoring platform Mira track cycles measures ovarian reserve predicts ovulation measures fetal house tracks your manual post progress and hormone imbalances at home.
With only a palm sized device mirror has reached the 99% of accuracy in 400 patient clinical trials plus automatic data interpretation, personalized AI learning and telemedicine. So, let's look at the demo.
Switch to the overhead camera. Please
simply pee on this mirror test one just as usual, and reverse the cap of this you're in perhaps the test one, so there is no contamination. Insert the want into the mirror analyzer. The analyzer will read your core ml concentrations and the data will be transferred to the mirror app wirelessly.
If some woman is trying to conceive the charts her cycle pattern she can share her house status with her partner or doctor and the AI learns her personal variability in the lifestyle and the tells her exactly when to try for maybe
our mirror blog advocates her on health related behaviors and the she consults with Dr. Browsers in seeking answers when in the online
if the woman is already pregnant, our mirror pregnancy test one tells her fetal status such as any sign of miscarriage
and our mirror ovarian reserve test. One tells her when to plan for baby or monitors her menopause progress.
So as you can see as a mirror analyze your screen right now mirror matters her actual formal concentration instead of a positive or negative estimation.
So let's go back to the presentation.
With 18 IPS mirror has high accuracy, low cost, and is smart with a decision making is easy to use. And the portable one compared with traditional lab equipment.
Mirra uses immuno for us is technology which is the golden standard using the hospital today. So it's just like shrinking the lab past into the comfort of your home.
So competition is limited for fertility. So ovulation prediction kids failed to add that to personal very abilities leading to miss the fertility peaks
for pregnancy. There is no commercially available portfolio monitor that tracks were Fredo status during the first trimester at home
and for ovarian reserve. In the menopause. You will have to do the lab testing which is inconvenient expensive, and it doesn't provide you continuous data. So mirror is the only diagnostic platform at home that covers every stage of women's health.
So women's house in the Chronicle disease monitoring market is $24 billion
in the past three years, we have achieved FDA NC registration, we build the 11,000 square feet of manufacturing facility by ourself, which is ISO certified.
In the past 45 days. We have acquired more than 2500s of signup users with 300% month over month grocery. We're shipping mirror study next month in the US and the launching in China and the Europe by early next year
mirror is going directly to consumer at a price of $199 plus app subscription and the Madison service we're working with medical community on research and distribution and we will be on the retail and insurance platforms. So big data we collected will be huge value to insurance to targeted ads and the research
mirrors. Future product lines include Chronicle disease testing, such as Cyrus hormone monitoring and the kidney disease as well as general wellness, such as weight loss monitoring by the same analyzer, but different has to once
My name is Sylvia. I have an MBA from Cornell and a master's in biomedical engineering from Columbia. I served as a business director in the fortune 500 life science company managing $100 million of global p&l. Our investors, includes the co founder of Alibaba Group, and our advisors includes the director of ob Hiva in from a Kaiser Permanente.
So understand your hormone today and the reserve Mira as a mirror care. com with our special promotion. Thank you
great job handling the FDA process. incredibly impressive there could you walk us through a little bit more on your revenue model says 199 for the device the strips this description like how does that look and what's sort of your lifetime value that you're estimating for customer
sure so the device is $199 but we also offer rental services and the financing so average spend about $33 per month and the test one is about $2 each for fertility, they will be using about 10 per month. So the overall lifetime value will be between like 250 and $300 just for the hardware but at the same time we're building telemedicine series we're building our community so we will be recommending like IBM services egg freezing services to our vertical Targeting customer base. So those are our secondary revenue model, which we're still building right now.
really great presentation. I was gonna ask you a little bit about the reference to the 99% accuracy. Can you speak a little bit about how that compares to clinical grade diagnostics versus at home tests since you're talking about a really broad set of potential tests that you're reading back to a consumer
sure so we have done the clinical trial with for country patient real sample So the way we're doing like a we tested urine sample by mirror and also we tested by the lab equipment which is like this huge and as we measure is a correlation and the fund is 99%
and those tests are what doctors are using currently in the United States to provide those through clinical laboratories were is a different
exactly so that's exactly the same test used by the hospital right now and by the doctor right now and, as you use also use immuno fluorescence technology. Same thing we're just Using exactly same technology technology that we're able to make a really big sensitivity. The size,
urine tests for hormone levels are available at every drugstore. Right? And I believe they also claimed to be 99% effective. Can you talk about what it is about what you've built that is unique and what's defensible?
Sure. So the the urine testing the drugstore right now are qualitative test. And also they're not really you know, having the IoT device we're like any AI in the background. So what you're doing is like, as he said, a hearse, fresh whole NC once you're above the threshold you're ovulating below that you're not but that every woman is different or recycles different.
So what we're doing is a quantitative test so we actually draw the normal curve for us if there's no way you're gonna miss your peak and also we have AI to nourish our pardon. So that actually Chrissy level is very different, it's like totally different technology and at the same time, you know where platforms Is it just the Raiders. So we can do also like later pregnancy tasks like ovarian reserve test and the old PKU it won't be able to do and we have 18 IPS to protect this technology. And also the regulation is pretty you know, it takes time to do so we believe that's where you cannot create a defense ability for this product.
Could you could you talk a little bit about two things the IP and also the separate different ones and why why different ones and how those how as, as a consumer without How many do I need and so on shirt start with the IP please.
Yeah. So we have 18 iPS cells, IPS covers everything majored in history purse was a hardware design and the other one is that has the wand and the certain ones like the information management like for example, standard curve in normalization when you know when the user visa to further increase the you know the accuracy. And what's the second question
Oh, yeah. So the ones that really designed by application, we we started from like a consumer focused angle. It's not really from a technology focused angle. So for example for fertility is your duty, you know, two most important hormones like our Asian estrogen and for pregnancy, that's actually just another test and isn't for someone who is trying to avoid pregnancy care about progesterone. So which is the confirmation of our relation, so we separate them into batches of product based on customer usage.
All right. One more round of applause for Mira.
That brings us to our next startup Kinta AI was any Rick into AI or Steven Glinert and Ben Zax
Actually, it's Rob here today.
came to AI does production planning for manufacturers when Robin I began his company, we were working with manufacturers out in Asia. And we would go out there and hear the same story over and over again. And if someone says, Well, you know, I have a contract with some someone who I supply something to write, you know, I supply devices to this guy. And if I don't get my product on the dock the day I need to that contract halves and value so you start backing out what that problem means. That means he has to get that machine running the day he needs to get it out the door. It's all these sorts of complications and complexities. And so what we found was that manufacturers big and small want this it's not just the small guys it's not just Joe's House of plastic products. It's people who are as large as some of our clients and
Simply what we're doing is we're using deep reinforcement learning and AI to make factories more efficient. And the way we're doing it is via production planning. production planning is what machine should be making what product. And at what, minute, hour in day,
right now, manufacturing is a fragmented software market. You have your essay peas, and they make some great stuff. But they make a CRM really Fear Factory. And then at the bottom, you have execution systems, right? They're going to tell your machines what to do, but nothing is really holding everything together. And more and more, you're going to need something and hold it together, right? Because robots don't make this problem easier. They make it harder, they add rigidity to your factory.
So that's where we fit in. All right, our solution is simple. We have the data we need from you already. All we need to do is customize it around your needs. What do you need? What do you want? What Is what are your, what are your KPIs. As a manufacturer,
we have a four step process I like to think for setting it up in your factory. The first is we create a digital twin. A digital twin is a simulator of your factory. It captures the data we took and turns it into sort of a digital simulation. Then we train an AI
we after that we deploy the model. We can deploy it on premise. We can deploy it in the cloud, whatever you want. And then you can use this is an active intelligence product. It is not something that you use once and then you you don't set it and forget it. You use it every day.
Our technology is deep reinforcement learning. That's our secret sauce. It's the same stuff that's proud alpha both beat the number one go champion in the world. And it's the same stuff that powers a lot of self driving cars and all it is that there's an agent and that agent is learning how to do the production planning in your factory and he does it by running through your factory a million times.
Right, go to demo.
Yeah. So imagine that you're a production planner at a large factory.
You know, like any manager, you have a set of business goals that you're trying to accomplish, and you have resources, you have equipment and personnel that you can allocate towards trying to achieve those goals.
In this example, imagine that you're a chemical work in a chemical factory that produces dozens of distinct chemicals, each of which has unique requirements about a multi step production process in order to successfully produce each product.
What our software does is it helps you decide given the set of customer orders and what due dates you want to finish them by
what is the production plan that will achieve those goals most effectively. So when using our software, you click on our schedule and it generates a production plan. And this production plan tells you which machines and resources should be doing which tasks at which times in order to achieve the business goals that you've told us that you want to prioritize and these production plans have been optimized to achieve those goals while retaining the flexibility to account for a new customer orders that come in unexpected downtime in your factory and other changes that might come up in the real world that gets messy. back to slides.
today. We are really excited because we found proof of concept customers who really do align with our vision of what the digitize industrial revolution will be. The first is a large ODM in the medical device space. They are $25 billion market cap they're sizable and the second is BSF BSF is the world's largest chemical manufacturer and in them and in our other partner. We have truly found someone who wants to work with us and really wants to achieve the goals that we do. And it's just been fantastic. A lot of gratitude
manufacturers spend money and software. There's this myth that they don't they do and they love doing it. And cloud computing was invented for manufacturing I think in large part by
And chemical manufacturers spend a ton on software. But more to the point there's a humongous market here in what value we can provide. If you think about it, the average chemical manufacturer is going to provide,
you know, are going to have about 60 to 70% operational efficiency. bringing it up just 10% at a single plant and we sell plant to plant is worth millions for $5 million.
I'd love our team. I love working with these people. We have people who have experienced with AI, we have people who have experience from Facebook, Microsoft, Google, and it is a fantastic people to work with. And I think we are ready to take on the task of bringing manufacturing forward or being part of the story of the changes that are going on in manufacturing. today.
Intelligent manufacturing is going to be the name of the game and manufacturing because manufacturers become less human and more machine. More machines mean more data. More data means you have to figure out how to do process improvement.
And that's what we think the key is going to be for how manufacturing is going to move forward. Thank you.
How cookie cutter is your, your technology and other words for each of those clients? How turnkey is it? How much customization has to be done? How quickly can it be up and running?
Yes. So essentially what we have built or one half of our software is a flexible tool that can try to simulate the complexities of a real world factory right now, there is some as we factor in new factories, there's some amount of new complexity that we've had to add for each of our, you know, first two or three clients. We're hoping that eventually we get to 80% solution that covers a large set of the manufacturers that will work with
so some customization but not too much work.
Where does all this data set and how easy is it to get to it? I'm going to guess like the AARP folks are a little bit more open than like to see ya Siemens and the ABB is the PLC.
So what's interesting about the manufacturers that the big guys out there is that they probably have a lot of ownership of their software. And they can sort of give us the data. Now, you'll notice that we had chemical A, B, C, D. And that's because chemical manufacturers don't like sharing the names, their chemicals, but that's cool. That's fine by us, because we don't need that data. So we actually can ensure a lot of the privacy that, you know, secrecy that chemical manufacturers in the manufacturers might want.
But yeah, the data is due to the way that these manufacturers have such good relationships with with their earpiece, and with their ABS readily available. So that hasn't been a major problem. I imagine, as we move forward, we may need to figure out those integrations. But right now that isn't the biggest problem we face.
Tell us a little bit about your sales cycle. How did you find your first customer? What type of deal do you have with them? And in particular, when you talk to customers, how do you quantify the benefit you can provide to them over the existing machinery they already have going
Yeah so how we found our first customer let me talk about BSF for working with a large corporate as a small startup you need to figure out how to do relationship building so the key for us with large manufacturers is gaining their trust.
And what that really means is that I need to figure out a way to talk to that operations guy at that factory floor and represent to him. Here's the value of providing and you have to represent it as a you know, financial bottom line. So I'd say the key is relationship building the to the next question you asked. I think you can answer that one.
Yeah. So I think in terms of quantifying exactly how much we can expect and how much to advertise up front versus how much someone tries our software and finds out for my factory, how much will I yield,
that's part of why we're excited about doing these two large
pilots right now because that will give us a little bit better of a sense of how much to benchmark which will help us with figure out exactly what price point we want to be targeting. But I think a lot of these even if we're not Driving a huge amount of uptick in efficiency that we think we will at the very least, we're helping you not make mistakes, we're helping you make rapid choices. So a lot of factories right now they have someone manually drag around cells and XL when a new order comes in, and they need to change something. And so at the very least, just being agile and having a solution that helps you find those faster, we think will drive a lot of value as well.
So two quick questions. One, your two pilots were unpaid.
Oh, both of them will be either paid in a convertible note or paid in revenue. So there's a combo that we're doing with both of it with with Mr. Client and BSF. So we're starting those actually that's the that's the next couple months of my so we are getting revenue Yeah,
but your plan will be some kind of SAS model
Yes, yes. So we sell plant by plant so there are BSF has do not quote me on this 300 plants or so and we can sell to
Each of those individual plans for a separate SaaS contract now what's nice actually is that a lot of BSF plants and plants all over the world or copy pasted from each other. So you'll have a plan Detroit, but they press copy and paste a plant in Hamburg and a plant in Shanghai and then the same plant which means that we can deploy, you know, in one plant prove value and then get to for free.
And I think that's the, the sort of the model we haven't. But yes, it is a SAS, there is an installation fee, and then a licensing fee. And we're going to figure out that value once we do these pilots because we can say, you know, what's 10% of the value that we that we brought to you your plan?
Okay. And my second question is, How hard is it to actually integrate your software into whatever else they've got.
So right now a lot of them are cagey about having our software directly control a machine because like in a chemical manufacturing like situation, if we messed up a factory would catch on fire or something like that. So we're not actually Directly integrate acting with those interfaces. Instead, we're helping the person who currently makes those decisions make those decisions faster and more accurately.
So there's still kind of someone who's going to print out a schedule. And it's going to go up on the like board that tells like each of the people what they should be doing, rather than, like direct numerical control of the machine.
No questions, any it seems like you guys did a really good at doing resource allocation. But it also seems like you picked a very hard market to penetrate, have you considered or why not go after doing resource allocation for like software engineering for JIRA.
Um, because manufacturing is awesome. And no, I mean, no, I'm very serious. I think that when you look around Silicon Valley, not enough people are doing manufacturing and not enough people are thinking about it and there's a lot of money in it. And I think that it would be
foolish not to go after the hard thing because it's hard. You know,
one other quick note in the six seconds I have
Is that one thing that we do need is you need to be able to tell us the rules of the game. And people and software engineers, it's much harder to predict like, how many engineering hours will it take to add a new feature to slack that sometimes harder for a manager to articulate or it's much easier to say, how many hours will it take my rubber glove factory to produce blue rubber gloves? So that's kind of that clean. This is what made us target that first.
All right, let's hear it for Ken today. I
have wing Lee presenting for Wingly or Emeric De Waziers and Bertrand Joab Cornu.
My guess is that most of you have been in a plane at one point in your life. Planes bring this image of long corridors where people are packed together like sardines. Now if I mention private aviation, you might be picturing those big private jets reserved to fortune 500 executive or people drinking champagne
and basically today your rights there's two options either be treated like cattle or pay $5,000.
What if there was another solution, a solution that could combine convenience with pleasure?
Well, being a pilot myself for over 10 years, I decided to build this solution with my two co founders Beth Hall and Lars
wing Lee makes private aviation accessible to everybody by connecting private pilot's with passenger so they can share the cost of the flights
beyond beyond travel what we provide is an amazing flying experience bringing to the aviation industry what Airbnb brought to the hotel one on wing Lee in only a few clicks, find your perfect flight be the sightseeing flight a day trip or simply traveled in towards fly fences. The platform takes care of everything for you.
Non payment pricing algorithm smart matching flight suggestion pilot vetting and insurance and you just have to sit back and relax but let's move to the demo
Can you please move to the demo?
Oh perfect. Thank you. So when you go on the platform today on we need at i o in Europe you can easily see all the flights that are posted by the pilots beforehand you can easily browse through them get more information and book The one you wants to book but we decided to go even further and we're happy to be introducing today the wing Li flight request the first platform ever allowing you to actually request a specific flight to a private pilot's. Let's say you wanted to go from San Francisco to Lake Tahoe four day trip, but there were no flights posted by the pilots beforehand where you just have to click on the request a flight button from there.
You just need to select the type of flight you want to do. So let's say a trip from San Francisco to Lake Tahoe you then decide on the number of seats you want to book
you enter some dates specification so as mentioned we're going on the day trip and we kind of flexible on the dates and from there you just add a small note for the pilots you then directly gets an estimation of the flight time and the price it will actually cost you for the bus for a flight if this matches your need you just need to click on the post my request button and this is smartly sent to the right pilots in your region. You just have to sit back and wait for the reply
but let's go back to the presentation willingly. Is that the right can you move back to the presentation please?
Yeah, so we need at the right moment and the right place to actually build a marketplace within the aviation and this is thanks to regulatory and insurance advances that bring with them great opportunities. We have now executed a successful proof of markets in Europe and we're not ready to expand.
Of course there's already some actors in the US but we're targeting these markets on there a total different angle. You might know for instance, Jet smarter they're doing an awesome job but they're focusing on big private jets whereas we're focusing on lighter aviation. Why? Because it's less affordable, it's more affordable planes to buy and less expensive planes to operate the perfect recipe for affordable flying
so deliberate events as I was mentioning including 2016 regulatory change on the European market that made it clear that private pilots are allowed to share the cost of a flight and advertise it beforehand. We didn't sign a first of its kind common shorter with utilities to ensure legal compliance. And the good news is that the US are following the same track. There is currently a bill in Congress to fully legalized slide sharing and when he is in the perfect position to take over this market the second it opens
But we also drove forward insurance innovation within the aviation space. We developed a tailor made insurance product, the first of its kind in this industry allowing us to offer an extra coverage of more than 1 million euro per passenger on every flight they do on wingless.
Now the market that we're looking at is the private aviation market globally it's a $35 billion markets and there is plenty to do to unlock and recapture value within it. private planes stay grounded more than 70% of the time and they fly with million seats on field annually. The opportunity is huge. And we decided to tackle this opportunity with a 300,000 private pilots in Europe. They're flying more than 4 million hours per year, averaging 2.5 empty seats per flight, which means already more than 10 million empty seats flying on field annually on in Europe. Now the us that we're looking at is a twice bigger market. It's worth $2 billion
and there's enterprise fly there more than 30 million hours per year, which means it's 30 million empty seats flying annually empty. And the good news is does number doesn't take into account the impact when he has as without the pilots to fly up to three times more on the same budget actually takes those number two 3 million
in Europe and more than 90 million in the US and our model is a clear win win for the pilots the passenger and environment let's take back or pilot flying two hours this would cost him $400. Now if he shares the cost, it takes it to $100 per person. And on our side when we take a commission of 15% plus $5 fix. So now what have we achieved what in the past 18 months, we managed to grow from a community of 30,000 members to more than 200,000 members from 30,000 flights online to more than 300 flights online and with a year to year growth of 300%. We recently managed to raise 2.5 million in seed raise, allowing us to aim for profitability by the end of q3 2019.
So if you want to experience the XRP draw your flights, if you want to discover your country from another angle, just go to wing lee.io and butcher flight today to fly fancy with wing.
So it seems like convenience and pleasures a theme today, I don't know.
And so have you check this. I mean, it's been flying lessons and stuff for a while. And I think it's a great idea. But have you checked the market checked it with the pilots at this point? Because a lot of people I know who fly actually like no one else there.
So we have the chance of actually been live in Europe from within two years now. So we have more than 10,000 pairs that are registered on the platform and that are actually conducting flights for the past 18 months, which means where he had more than 12,000 passenger in flights done by more than 1000 pilots in Europe. They're actually so 11,000 flights done by 1000 pilots, would
you do you find pilots are buying planes just for this or artists just really Hey I'm topping from here to there I'll take somebody with me
it really depends on the markets but it goes even further than this for the pilots we have different types of pilots the ones that are just going from one place to another and going to take person with them to share the cost but also their passion and then we also have plenty of passionate pilots that just want to fly as much as they can but their often restricted by the budget and so we allow them to fly up to three times cheaper mean they can fly up to three times more on the same budget so they're happy to basically do any kind of flight that people are wanting to do
your hours okay
what fraction of your passengers in Europe are repeat routes that is to say I keep on booking that towel
and so the return rate on the platform the recurring fee of passenger is 20% that do at least a second booking after doing the same one
right but how much to the same route
it's not on the same rather they doing it what we see the most what happens the most is people actually starting with a sightseeing flight to discover what's privatization. What slight division how it works, yeah. And then they decided to Use it to actually travel go from one place to another.
So that changes. Okay. And second question is, so my understanding is, if you're a recreational VFR pilot in the US you can share the cost, but you cannot profit from the trip. So how does First of all, how does your margin take work there because technically that's not fuel costs. So is that legal and second does not always mean that the ticket fair will be quite low.
So the private pilots as you're saying, It's totally right. They're not allowed to make a profit out of it need to hold a commercial pilot license for this and it's operated on the different certificates so they're only sharing the cost meaning that instead of paying $200 to go to tell for the weekend they'll be paying $50 because they're sharing this with passengers so for them they're not looking to make this as a living it's really just saving some money now on the legal aspects of us taking a commission we're not taking Commission on because we're conducting a flight we actually just a matchmaking platform and that's what we're making the passenger They're paying the matchmaking service and the fact that we reinvented the flight.
how do I find out whether the pilot who is now say they will take me to Xyz has flown 1000 hours or has blown three hours? Is there a rating system? Is there some sense? You know, if I want to Uber, I can find out that my driver has made, you know, 1100 safe trips?
That's a really good question
or has cardiac issues or something
good questions because it's really at the heart of the DNA of the company, it's being as transparent as possible and give all the information to passenger that don't necessarily know what is it to have a license what's a good pilot experience right or not. So first, we check all the licenses and medical of the pilots to ensure that we don't have pilots with medical problem as you were mentioning, and and then we also ask for all the logbook information so they experienced
The number of hours they don't globally and the recent experience they have because this is very important. And we display all this transparency to the potential passengers for them to actually know who their pilots
but we go even further than this because there's also the rating system of course, but we connect really early in the process the passenger and the pilots so they can have a chat together and the passenger can know more about its pilot before actually doing a flight.
Most people using this for sounds like recreational sort of versus transportation.
Yeah, so the most the the use case where the primary use case is more leisure one so people get actually they go on the website and the Discover what they can do. And it gives them ideas of trip they could take and it's what we want to do today. It's really about educating the market that private aviation does not have to be reserved to an elite but actually was like aviation everybody could use it.
And this is the two goals we have today with me it's educating the market towards the fact that private aviation can be accessible to everybody and creating the biggest and the first A community in private digital community in privatization.
And on the on the hardware side is the majority. What percentage of the hardware is is owned by the is owned by the pilot versus small little companies who are doing tours around a particular area and you are now using this in order to get more customers.
Can you repeat the question please?
Yes, are all are all it is all the airplanes owned by the pilot or other small companies who have a couple of charter
this varies on the markets, even in Europe, in France, they each presenter renting the planes we go to the UK, they're more property based. What's really important is that whatever if they're renting the pilot the plane or the owning the plane, they're allowed to share the direct cost of the flight and it's not a problem if they're renting it.
Okay. And last question. customer acquisition cost.
Yeah, so today it costs us one year 52 actually gets a passenger to become a member of the community and then we have a really strong engagement with mailing, once they're in the in the community to get actually somebody flying in a plane. It cost us 20 euro to get somebody there.
Susan has a burning question. Yeah, there are there limitations on the airports that can fly into. So could I go on and say, I'd like to find someone who's flying to Shelter Island on on Friday night or East Hampton? Or are they not allowed to fly in there because they're not commercial pilots
so they're really bigger, feels like SFO or close to general aviation. So you won't be able to learn to San Francisco Airport. But for example, Auckland you can go into really depends on the airports. But one of the biggest said that lighter planes have compared to private jets is that they can access any kind of small runway whereas private jets get limited by the length of the runway. And that's why we really believe it's the future of transportation. India is in light radiation and the bigger one
all right, give it up for wing Lee.
We are so close to being done. We have one final company. This is the wild card startup, which means they were chosen from the startup alley and the only found out they're presenting a couple of days ago. So be supportive. Be generous. This startup is Vtrus, presenting for Vtrus, or Jonathan Lenoff and, Renato Moreno
all 350,000 industrial facilities in the US are required to inspect their infrastructure for safety, reliability and profitability. These companies are spending over 100 million dollars a year on these inspections where there's equipment in process there's inspections that means you have people putting on harnesses and inspecting at heights if people putting on respirators and climbing into confined spaces or you have people walking around active equipment, there's a problem industrial equipment as our industry inspection is dangerous for workers.
It's extremely expensive. It's tedious and time consuming it's prone to human errors require specialization, and it's extremely hard to scale the solution autonomous robots introducing Abby, the autonomous drone for indoor industrial inspection,
here's how it works, inspectors teach Abby what data to collect were to collect it. And how often to do so Abby within a Tanase navigate and indoor environments, collect that data and return to the base station to recharge. It will then upload the data to the vitreous cloud for review by the inspectors.
We operate an indoor environments. As you know, indoor environments are full of different objects and moving things in order to autonomy navigate, we have to know precisely where we are and what's going on around us. We've pioneered a technique with onboard cameras to understand the world around us. We're going to go to the live demo now on the computer.
What you're seeing now is a mission that we ran in our office we're going to start off with a blank world
As the drone moves, we're going to see a live reconstruction of this environment. This is a fully immersive GUI that allows us to look around from any vantage point. For example, Renato, you might look at the base station.
The ability to look around the environment from the vantage point and perspective that you want is key for these facility operators to truly understand the interplay of their different equipment. It's key to understand what's going wrong, how to diagnose that issue, and what the solution is in front of the drone. We see the live first person view this is the live camera feed from the drone that's critical for people who were watching the reconstruction to understand what it sees.
This 3d reconstruction allows us to operate without GPS fully indoors and this mission. We told Abby to go around autonomously explorer and then go look at a pressure gauge that we stuck up on the cabinet.
As Abby's going towards the cabinet. You're going to see a bunch of red boxes up here. This is our obstacle avoidance system we give Abby drone high level
Man who say go for Go back, go left go. Right, but we let Abby drone do the heavy lifting of understanding. Don't crash into that cabinet. You know how the flight controls can actually work. We can also view the world in various different ways but give you the world with a geometry centric view so that colors and textures don't necessarily money up everything when I mean drone is done and to return to its base station to recharge. Here's what that looks like to the hardware demo.
This is a B drone. When it lands on the base station. It will tell the base station Hey, I'm running out of batteries. I need to recharge
the base station will center Abby drone and then recharge
what we've created is a safe drone for indoor use. And for around people. Give back to the presentation please.
Yes. Next slide, please.
We've completed the loop for full autonomy with our base station to allow fast, safe recharge,
we collect the right data at the right time. with built in sensors. We do 3d reconstructions, high resolution photos and videos, thermal photos and videos, 360 photos and videos for situational awareness and environmental monitoring.
All that data collected is synced to the vitreous cloud and displayed in the form of automatically generate reports and the intuitive user UI that you just saw.
There's some competitive technologies out there, for example, mainly operated robots like a DJI drone. However, we have distinct advantages because we're safe and scalable because we don't require a skilled pilot we work 24 seven, and we geotagged the data. This is critical and environments where hundreds if not thousands of pieces of equipment all look exactly the same. We also don't need GPS which means we can operate indoors
and other competitive technology is fixed sensors. However, these require massive infrastructure you guys
To put sensors all over the place, run power communication to them for us. just plop the base station in the industrial facility and you're ready to go. We're completely non disruptive to the facility. And we can move everywhere, which means we can always capture the right data at the right time.
Our key industries have high acid value and high downtown impact. downtime is time spent not producing the desired end product that's industries like continuous process like pulp and paper, food and beverage, chemical plants and data centers. The total addressable market for this is $15 billion dollars per year.
Today, we're working with to fortune 500 companies and an industry leader to form the foundation of our pilotless program and begin signing up customers. We've interest from over 130 companies for inclusion in this pile of this program.
Our go to market strategy is threefold. Today, we have beneficial partnerships with OEM to tap into their existing customer base and to get feedback and case studies.
these case studies will drive sales, but they'll also generate the content we need for conferences and publications
with the conferences and publications. And just in case studies, we're going to create the brand name that we need to partner with industrial suppliers and external salespeople.
We're offering Abby as a service for a subscription of $4,000 per unit per month.
Our team is a team of dedicated individuals with experience computer vision, industry, application, product design, real time computing and robotics.
Let's pioneer a new state of the art industrial facility operation. If you'd like to join the pilot program or learn more about interest, please visit www.vtr.us.
Yeah, I'm curious as to who are the people that are currently doing the inspections and is that their only job so you're going to replace them or do they get freed up to be more efficient and do something else?
Yeah, we definitely see ourselves as a augmented, we don't see ourselves over play, sir. Today, the inspectors are either full time inspectors or the operators, the operators of people tasked with actually physically manipulating the machines. And the problem is that you have people very skill levels, assessing the subjective analyses. And so what we want to do is collected data for them and presented in a manner where one person can quickly move through all the data collected at a facility
God. And one quick follow up and who do you actually sell to or your channel? Who do they sell to inside of the company.
So there's four main people we talked to, there's the safety manager, the operations manager, the maintenance manager, and the engineer manager for each in charge of different things. For safety. We're clear when you know, if you don't need to stick a guy 100 feet up in the air, you can just put a drop there. It's clear when they're engineering, for example, they're in charge of the long term fixes. But the problem is that if people are always putting out fires, they don't necessarily have the manpower necessary to go out and do those assessments. And that's where we're as a force multiplier able to really help those industries.
Can you talk a little bit about how you came up with your pricing? Because it does seem that at, I guess, $48,000 annually, I could just buy a DJI drone and hire an operator,
you could. But then the scalability, that solution is incredibly limited. We're talking with facilities that have, you know, facilities all over the world. And the problem is, how do you get your data center or your induction facility that you can't put near large talent pools to be the same all across the world. So we're trying to use autonomous robotics to offset the requirement for those skilled pilots.
It's also the case that many for example, data centers, they are located outside CDs were the the pool of a skill set people are unavailable and you really want to keep an eye on what's going on in your assets. That has to be keep up up to a 24 seven
and maybe as a follow up to that, what was the rationale for the pricing because I actually have the opposite impression is that for some of these industrial solutions, you could be leaving money on the table.
Sure, yeah. So, you know, to give you kind of a price point, some of these confined space inspections can be as much as $80,000 per one. And so you're right. We're very cheap alternative in some respects. But we really want to allow the possibility to scale linearly within these facilities where one facility doesn't just get one drone. We want them to have 10s of drones. And so we want to keep the bar low enough such that it's still accessible for these facilities.
And what are your unit economics? Or was it what are your unit economics? What does it cost you to create a drone and to keep it up?
Sure. So what you see today this guy is a bomb costs around $1,000, but that's because our main supplier today is Amazon and we're using off the shelf components. We feel like through conservative estimates, we can bring this down to around $4,000, which is going to allow us to give a new drone to these industry facilities at least every two years. So
our main goal is to have continuous data capture and not have these facilities necessarily worry about the piece of hardware that they own.
When facilities choose not to go with your solution. What's the reasoning? Is it cost? Is it safe? safety issues? What is it?
Sure. So we haven't had someone who's kind of just flat out said no, today so far, the advantage of being the first mover
and a drone
is that you won't want
it. They want a drone
they want to join. Exactly, yeah, yeah. So the advantage of the first mover is that no one else is doing this. But the disadvantages that we have a lot of education on our hands to teach people what an autonomous drone is, how can interact in the facility? What does it mean to have this operating around workers, you know, trying to remove the perception. This is a Big Bad Robot that you might have seen in a movie and said, this is a drone that's going to move very slowly through the space, understand there's an obstacle yield to you, those kinds of things.
The question was procession was well, so many, many customers that have the perception is that a drones are actually a tool they don't see it as a robot. They when they think of a robot, they think of something that is Some wheels.
So we are really a single location, a process for us to get the message across that this is a tool that you can set it up and it will be up and running for you. He can do the monitoring for you 24 seven, and once you teach you what to do is just going to be readily available.
Can you talk a little bit about the software because hardware wise, you spoke about the hardware a lot, but it looks like I mean, that looks like a quad copter with a little Rico camera stuck on top. Yes. So. So you're right. It's from parts from Amazon. I recognize some
talk about the software.
Yeah, so we think I was a system integrator. So obviously we have not a design like it seems like a new model a new propeller but we have grabbed the right type of components to make the system work in the best possible scenario. So for example, on the perception system we have integrated that a sexual a depth sensor that allow us to capture that measurements in injury spaces where we will be very difficult to to capture with originally stereo cameras and yet we keep the price very low.
We have also design and the right processing power that that can combine all of these sensory stream and do the state estimation with our bodies on such as Islam and creating the right representation of the of the internet space with three dimensional mapping sites that the robot can move in the space accurately without the needs of GPS.
So this is something to keep in mind and many of the other ones that operate autonomously out there rely on GPS signals that when you move that into an inner space is completely available and unreliable so we need to rely entirely on computer vision algorithms that run in real time
and now also the three maps that we create they are a creative instantaneously the the set of the purpose of knowing
building that don't worry leasing space, how to avoid obstacles and how to how to navigate a particular way that you do sitting in the space.
All right, let's hear it for vitreous.
company for today your guys. Alright, that brings an end to the startup battlefield. Thank you to all of you for sticking around late to see all the stars before you leave. Let's have one final round of applause for our awesome judges.