Andrew is Co-Founder and President of RealScout, a startup tackling residential real estate’s core problem: matching homebuyers and homes. LinkedIn revolutionized hiring by bringing job-seekers (demand), employers (supply) and recruiters (brokers) to one place - and providing transparency. RealScout strives to unlock the same revolutionary dynamic to a market that’s twice the size of recruiting.As Co-Founder and President of RealScout, Andrew drives company vision, the recruiting of its world-class team and strong relationships with the industry. Andrew is consistently featured as one of the 100 Most Influential Real Estate Leaders by Inman News and was recently honored as Forbes 30 Under 30 and Silicon Valley Business Journal's Top 40 Under 40.Recently, Andrew launched a popular video interview series in collaboration with Inman News, which enables brokerage executives to relay their questions to industry play-makers including the CEOs of RE/MAX, Redfin, Opendoor, and the National Association of REALTORS®.View the profile
About the talk
Search is changing exponentially, the days of Google rankings are old news. What's new is visual and driven by artificial intelligence.
Thank you so much for being here. Some of you may remember me from last year. My name is Andrew Flachner the CEO and Co-founder of a company called RealScout. And some of you remember me from last year. I spoke about the differences between generating leads and generating offers this year while this year. Brad made me promise not to make a mess of his stage, so I gotta be careful. Today I want to talk about a new technology called Computer Vision. It'll forever change the way you think about home search and to make my point first. I'd like to make an analogy around candy. So growing up I had a sweet tooth.
Okay, perfect. So you're in luck because under your seat, you're going to find a small tin of jelly beans. 500 them out there. So go ahead and find that now. And if you are not one of the lucky winners, then I'm sure your neighbor will help help you out. All right, so again you're not one of the lucky winners that's okay. Sharing is caring so feel free to open up that container and not shake it like a rattle that's going to be really distracting. And when you popped a container open, feel free to enjoy again share with her neighbor. So
not only I had a sweet tooth but I also knew that not all candy was created equally, so for me Jelly Bellies, were always my go-to and I would try to find my favorite flavor in a big jar of jelly beans just like this one and I'd be combining strawberry jam with peanut butter to make an astronaut version of peanut butter and jelly sandwich every now and then I would make myself a banana split if I'm feeling really ambitious. I would make myself Jelly Belly cocktail hour and part of the fun for me with carefully identifying the right flavor because there is
nothing worse than mixing up a butter popcorn for a pina colada am I my right> All right, so trust me I did it twice and a synonym for a red apple a black licorice for a Blackberry truly unwelcome surprises and let's face it through this photos of jelly beans good for you to identify that crushed pineapple coconut how many of these beans have freckles or spots textures? How many of them were deformed or as Jelly Belly called them belly flops? It's not easy right in order to store. His Jelly Bellies. You have to compare them one by one. Maybe you have to take a big jar of jelly like this and
dump them out on a table. And even then would you remember the texture that delineates a jalapeno from watermelon? Probably not. So what does it have to do with real estate and I'm sure you're wondering well fast forward 10 years to when I began selling homes and granted I still look like a child. But it was at this stage in my career that I began flipping through photos in the MLS day-in-and-day-out hundred thousand of photos, it felt a lot like sifting through jelly beans just a lot less fun. My clients were searching on Realtor.com and Zillow and they're trying to eliminate all these properties by
flipping through all these photos very quickly. And it's not easy. Think about the photos that are shared with you to that first photo have stainless steel appliances. Go to the second photo it was kind of molding. And how many of these photos contain French doors or skylights. Again. It's not easy And just like a bowl of jelly beans consists of different colors and patterns, houses consist of different rooms those rooms consist of different features hundreds of designs,
thousands of style, tens of thousands of finished appliances and so on and each one of this little differences could be a deal-maker or a deal-breaker for when your clients just like they could be for a Jelly Belly cocktail hour. And so today I would like to talk about computer vision imagine that these photos could talk and it said of images looking like this. They look like this how easy would it be to find that tangerine jelly bean how simple would it be to find a kitchen that has an island or pendant lighting or a wine cellar in the back? If you have a client that is looking for a home
with an open floor plan. No problem. There it is. And you might have missed it for the built-in speakers in the ceiling here in the kitchen technology that can effortlessly and quickly make sense of the visual world right by my favorite Jelly Bean flavor within seconds that can also identify all the homes on the market that have open floor plan and more importantly to make you stand out as a professional with access to this account is already helping keep agents in the center of the transaction and I'll show her how to do that for you to
the first it is so computer vision. It is simply the science of enabling computer to make sense of the visual world of technology the processes photos. And these Innovations have been driven by something called Deep learning as a technique where computers can learn to solve problems and more importantly they become smarter over time and Spas with any deep learning learning rather be poor Eddie Murphy first pot. Right? And so just I got a parent would teach their child at this is a house. Well programmers are teaching computers by using examples. Every single image must first be viewed
by a human Shack catalog hundreds of thousands even millions of photos must be required before the models can identify people and faces and even ruined like bedrooms and bathrooms in living room and kitchen. And so this is a pretty ambitious call The Vision or the the idea of the goal behind computer vision is to model or replicate human Vision using software and Hardware different levels a while seems ambitious actually think that these computers have a pretty good chance and I'll performing us because ultimately humans are incredibly slow Computing machines, especially when it comes
to monotonous and mechanical tasks like talking photos. And so this is something that you know, we were not meant to do. In fact, this is a historical Moment by any measure because by teaching is deep learning algorithms to identify photos were essentially giving robots. The ability to Speedway humans do is not exciting building machines that are Guided by cameras, aren't we going to send Ripple effects and how we live our lives and how we run our business is What do you notice it or not computer vision is already a part of your
daily life. Many of you have seen it being used by Facebook Facebook uses computer vision to identify this wonderful woman standing next to me is my fiance Julie who sitting over here computer vision is also being used by Facebook to help vision impaired people hear their timelines and here who's interacting with our content because I can't see it. I see many of you on Snapchat and when you space swap Okay, then you probably know what's coming and for that I apologize, but I stay slot this morning with Lee Brown or NC for the week and here are the results. Terrifying something you simply
can't unsee and then we are real estate conference after all. Obviously, the nest is the maker of smart thermostat housing their outdoor home surveillance camera, what should identify for the person in front of your door in this is far more advanced than simple motion detectors. This is semantic vision is understanding the context of photo the way we do and finally for my last one example any wine drinkers in the house. Okay, I figured so it when you scan a bottle of wine, you can see what type of a y knot is how much it cost for two pairs well with you and ratings and
reviews in cases ranging from skin cancer breast cancer computer vision has identified some types of cancer are classified During certain types of cancer within 93% accuracy. Think about what With a psychology not knocked out. My computer vision is in the world of Transportation right using computer vision here are to identify pedestrians and differentiate them from motorists all in a very Dynamic environment. And so what do you think? It's helpful or creepy? Facebook designer find your photos and sooner. Completely changed the way we saw we will stay the way we start for Real Estate space at
home store in the past 10 years in every listing are completely ignored because there simply is trapped in photos there hidden and we all know the consumers want a photo center experience the beginning it means that they want to search those photos. Right? And I stayed until we can understand the images and that's going to unlock a treasure Trove of data about every listing on the market and make it through and consumable until we share with you our story in 2012. We began manually creating a types of experiences
ultimately scale using computer vision to label more than 10 million. But each of these models in the same way that Google or Facebook use mind-boggling amounts of data for their algorithm. We traded this vast data set and Taylor it around real estate first real two experimental feature called future search for the first time allows a consumer to click on a category of home or a room feature like great outdoors and see all the homes on the market with large backyards. It is really cool the same thing applies for
updated kitchens updated bathroom and even pool until many companies. Does Matt look familiar and I on the second house was compared. And so compare for the first time I've consumers to compare two or three properties that they really like side-by-side room by room. And so here you see me part weave automatically position of the kitchen photos side-by-side all the bathrooms side-by-side even the living room side by side in this is a really beautiful and it's good way to search for a home. Finally, we reinvented the photo Carousel
50 or 60 per listing which is great. But you expect your clients or you for that matter to flip through 60 photos just to find out image of the bathroom. Now, you don't have to put on bathroom see all the bathroom photos in one view same thing goes for kitchens in other rooms in the house. The clearly just makes searching a lot easier right allows consumers to quickly navigate to the rooms that are most important in their search. So this is all for me to Compass in the past few years and implementing and training with model. What does the future hold? Is it hard to imagine a world where
computer vision can predict the marketability of homes or even a sale price what are fine-grained room classifications like Game Room vs. Nurseries or Dex versus patios? Can we extract the finishes with the building materials out of these photos like a quality of the carpet for the type of hardwood using that Nursery Water? How well the Landscaping is been done in the backyard or even the type of tile using the audio? What compute computer vision of the future tell us the architectural style of home or how long it'll be before you have to replace your roof? What do I have watched? The
Landscaping is in the surrounding neighborhood. How many cars you can fit in the driveway or the garage? So these techniques can be used to extract data like square footage or construction materials in this is all going to bring a lot more transparency to you and your fine. Little potato concert that we created as an industry and that and then made ubiquitous by companies like Zillow and realtor.com. And now you've got hundreds of millions of people going to change. Once again, the question is who will this new era of home search. Will it be Brokers and agents will
associations and imma let this be a little bit slower to adopt as change and Outsiders to dominate with the technology. I think we'll be able to partner with these Outsiders and make it our own. This is our opportunity to embrace change and drive the industry forward for you while there is a spectrum between complete autonomy robots are Guided by cameras and no skill. That makes us uniquely human Self-driving cars that belongs on the far left of the spectrum. It turns out that these cars are a lot safer when we're not behind the wheel but what about radiologist turns out to
the big part of the spectrum were robots and humans intersect. This is called man-machine symbiosis the radiologist with the skills to implement and manage this software are going to be in higher demand for something as critical as cancer detection. Does Technologies going to help Radiologists perform at higher levels, but a trained professional will always be essential and interpreting the results in the context of various other factors like medical history and symptoms. The world of real estate making sense of data and more importantly making decisions through the state
and through local expertise and highly personal skills and machines can take fire into just like radiologist the Brokers and agents to adopt the technology are going to be naturally able to better serve our clients and will not really being hired a man who has expertise is that you standing in in the consumer experience. It's the only way to experience in 2020 and computer vision is still in its infancy the true facts of this technology, but one thing is for sure pioneering the use of this technology now is a rare opportunity for our industry to take advantage
of it and forever changed the nature of Home sir. Because in this Jelly Belly Bean world of real estate, we have a decision to make are going to take the technology and lean with it or are we going to sit on the sidelines while someone else stores are jelly beans? A real Scout restraining for Willy Wonka status and we hope you are too and so enjoy your jelly beans with April tonight.
Buy this talk
Access to all the recordings of the event
Buy this video
With ConferenceCast.tv, you get access to our library of the world's best conference talks.