This week, we interviewed Andrew Sidhu from Foyer.
Without further ado…
Who are you and what do you do?
My name is Andrew Sidhu, and I am the Founder and Chief Executive Officer of Foyer, Inc., a PropTech development company. We just introduced Foyer Insight, the most advanced computer vision Artificial Intelligence (AI) suite which is focused solely on the real estate industry. It uses multi-label image classification and image segmentations to give real estate agents and platforms the tools to be more responsive to client needs by automatically ensuring the highest quality images and building an image-first experience at the fingertips of a new generation of home buyers. It is my belief that by doing so, we can elevate and streamline the home buying process.
Our company is headquartered in Trumbull, CT and we have an excellent team of developers who are working to push both our machine learning tools to a new level through breakthrough unsupervised learning techniques, and our image validation suite to ensure the highest quality of listings. I personally head the research and development of our ML team, as well as running the company as its CEO.
What problem does your product/service solve?
During my own house hunt a few years ago, I began to feel that finding your perfect home should be a more intuitive and streamlined process. I wanted to be able to search for my new home by its Kitchen, but ended up having to click for 30 seconds or more just to get to the kitchen pictures of each listing. Until now, property search sites usually relied on simple preferences such as location and price to display properties for sale.
That’s because the most important part of any real estate listing, the photos, were both completely disconnected from the rest of the listing and each other. In order to be able to jump to the kitchen pictures, Listings need to have the context of what is in the photos, nothing like that was available.
Foyer Inc was created to solve that problem and provide linked, detailed context to each photo and feature of the home. We have taken it a step further by cataloging the features of each room, stainless steel appliances, fireplaces, vaulted ceilings, and contextualizing it to searchable parameters. We can find your perfect home, like the one that has a view of the fireplace in your living room from the kitchen island, and allow you to get as specific in your search as you want. Not just limited to bedrooms, price, and a couple checkboxes which provide arbitrary results.
We give real estate companies the drop-in tools to enhance their own services and platforms the tools to evolve the buying process into something new. Whether it be an image first search experience, a better recommendation engine, and ensuring that each listing is of the highest quality.
By combining big data and our AI, Foyer Insight allows real estate agents to provide interfaces that allow clients to search photos and information based on room type, lighting, specific features (hardwood floors, stainless steel appliances), and more. Foyer also allows realty search sites to evaluate pricing trends by home features which are not usually available by just the listing details, such as the color of kitchen cabinets and countertop combinations.
In the past, the industry seemed reluctant to embrace technology for fear of eliminating the personal connection between broker and buyer. I wanted to help real estate professionals turn to solutions that can help them enhance that relationship. In fact, the human connection between the agent or broker and the home buyer is still one of the most important elements of the real estate transaction. This is why I believe it is so important to put the most advanced technology into the agents’ hands. Companies like Zillow have been using advanced tools to bridge the moat that a personal connection has in order to compete with local agents. I want to empower those agents with the same tools and democratize real estate information to make a better experience for every home buyer.
The use of AI and unsupervised machine learning allows us to produce actionable data which has been hiding in plain sight within every photo, and using that to help agents enhance and improve their clients’ experiences during the home buying process.
What are you most excited about right now?
I’m most excited about the potential impact that unsupervised learning will have on image classification because it cannot be underestimated, especially in the real estate industry. Real estate imagery is subjective in general, and can vary greatly by cultures, regions, or even neighborhoods. The only way to be consistent in all cases is to eliminate human biases by the people building the model. Removing the bias and combining it with Foyer’s ability to work off millions of images, means it isn’t overfitted to a certain décor. This allows it to be easily scalable to other markets and to perform at an equivalent level anywhere its deployed.
The adaptable nature of unsupervised learning means it can be used for many kinds of A.I. models, but it is especially good for real estate because the classifications are not cut and dry, or mutually exclusive in nature. In the past, it has been difficult to deal with edge cases with a supervised training dataset, even between developers we would have arguments about how much of an image has to be in the kitchen, or where the kitchen starts and ends. But now using unsupervised learning, the model grows on its own and develops its own consensus. Additionally, with unsupervised learning, new features can be scaled into production much more efficiently, without the need to hand feed large amounts of data. The models are able to accurately develop new classifications on their own with far fewer validation examples, rather than requiring our developers and client partners to walk it through every step.
What’s next for you?
My focus now is on working with top real estate company development teams across the U.S. to implement our technology into their stack and adapt it to each company’s own unique brand and specifications. We are also continuing to develop our visual suite that can be used by any agent in the United States on their own to provide an image first search solution to their clients today, and help them build and validate their own high-quality listings.
What’s a cause you’re passionate about and why?
I’m extremely passionate about right-to-repair and technology in the automotive space. Maintaining a fork of OpenPilot, an open-source aftermarket self-driving system, for Pre-Autopilot Tesla cars produced before 2015 is currently my primary hobby. Working with the community, I help maintain a fleet of over 100 cars which are all running our fork and hardware.
Many of these cars are out of warranty or salvage cars, and are usually only able to be repaired by Tesla themselves, and there’s even a question if Tesla will even repair the car at all. I have personally helped hundreds of people diagnose, repair, and get their cars running again after Tesla has refused. I don’t accept payments for this or any of the work I do on OpenPilot and redirect anyone who insists to donate to Planned Parenthood instead – My spreadsheet of donations currently has a total of $12,850.00 donated since 2017, I’m pretty proud about that.
I have recently taken an interest in 3D printing as well, but that is honestly just an extension of my OpenPilot hobby as our latest hardware requires us to build custom parts.
Thanks to Andrew for sharing his story. If you’d like to connect, find him on LinkedIn here.
We’re constantly looking for great real estate tech entrepreneurs to feature. If that’s you, please read this post — then drop us a line (Community @ geekestate dot com).