The real estate market is not easy to navigate. Those who buy must take into account a thousand variables (including rising interest rates), those who sell want to try to do it at the right price and under the right conditions. Fausto Jersey, the CPO of Casavo, tells us how the Italian real estate startup uses artificial intelligence to simplify the life of both those looking for a home and those who want to sell it. And to find the right match between these two needs.
Casavo and artificial intelligence for buying and selling a house
Casavo is an Italian real estate startup that aims to simplify and speed up the process of selling and buying a home. The idea behind the startup is to offer a instant buying service, i.e. the direct purchase of properties by an interested buyer or by Casavo, which then restructures them and resells them on the market. In this way, sellers can receive an offer and close the sale in a short time, without having to worry about bureaucracy and negotiations. All inhabited by an automatic evaluation process based on artificial intelligence.
On the other side, home seekers can find their dream home in no time using AI to find the right match. This speeds things up, avoiding wasting time on properties that aren’t right for you.
Our interview with Fausto Maglia, CPO of Casavo
The technology behind Casavo’s digital platform for home buying and selling interested us. It is an artificial intelligence that automates a market that is usually very “slow”. Anyone who is looking to sell or buy a home knows that it can become a very complex process. So how can Casavo’s AI evaluate the house we sell or find the one we are looking for in a few minutes? We asked Fausto Maglia, the startup’s Chief Product Officer.
Could you briefly introduce yourself and explain how Casavo was born?
“Sure, I’m Fausto Maglia, I am joined Casavo since its foundation in 2017 and since 2020 I have held the position of Chief Product Officer. Casavo was born from the intuition of the founder, George Tinacci, who, starting from a personal passion for the real estate sector, understood that they could create a new market category thanks to the use of technology. A huge and scarcely digitized market indeed, it implies great opportunities.
“Casavo was therefore born to innovate the real estate market and allow those who want to sell and buy a house to do it in a simple way. Thanks to our digital platform, we are able to match the needs of home sellers with the wishes of those who buy, guiding people throughout the buying and selling process and offering a simple and transparent experience”.
How does Casavo manage to carry out an automatic property valuation (AVM)? What data does the machine learning system need to figure out the value of a home?
“By accessing our site or our app, you can receive a Instant evaluation of your property and understand what are the key factors that determine its market value, as well as being able to view similar properties sold nearby. All this is possible thanks to EVAour proprietary algorithm (AVM) which thanks to the machine learning compares hundreds of similar properties in a given area and then refine the value estimate based on additional details related to that specific property.
Fausto Maglia, CPO Casavo
“The model comes constantly updated by our Pricing team taking into account the most recent data points from the listing portals we rely on e by government bodies such as the Revenue Agency. After being thoroughly pre-processed, cleaned and refined, this data is fed into the model for improve performance and reflect the trend of the real estate market as much as possible.
“Briefly, the approach employed is that of‘supervised learning (more commonly known as supervised learning). Where for each incorrect prediction an error is calculated with respect to the real value of the property. And with machine learning techniques yes reconfigure the parameters of the AVM until optimal performance is achieved. We monitor the performance of our models over time, following new training, by evaluating the predictions on a “test-set” of Casavo real estate properties. We use mathematical-statistical metrics such as errors in absolute value between predicted price and real price (MAE), mean absolute error and median percentage (MAPE), coefficients of determination such as R2 and others. There is no absolute indicator, we use a set of metrics to evaluate the goodness of our models over time”.
Is the analysis of videos, photos and floor plans accurate (is there any data about it)?
“Yes, it is very accurate. We use Deep Learning combined with Computer Vision techniques to provide instant feedback to sellers right when they are uploading documents and images to our app to complete their home profile, reporting immediately what is missing, for example photos of the kitchen or bathrooms.
“We use deep learning micro-services built by us, such as Breezer (for classification) e Rum (for object detection), which have a accuracy around 90% on the test-set and that includes images that the model has never seen in training. We are therefore certain that the performances are optimal. Furthermore, the training dataset is balanced, as is the test dataset. We are talking about hundreds of thousands of images from public datasets, aggregated and pre-processed”.
Can a person ask for a re-evaluation if they disagree? Is the evaluation process explained to the user?
“As far as the issue of real estate valuations is concerned, however, more than one can be requested, but a negotiation cannot be openedbecause it is precisely the market value, not a purchase offer or the price at which the seller will necessarily decide to put his house up for sale.
“When a user makes an evaluation on our site, he can see the average price of a property similar to his on the classifieds portals. However, the negotiation and the time elapsed between the publication of the ad and the actual sale lead to a difference between the asking price and the market value. Transparency is one of Casavo’s founding values, which is why we offer the user the true market value of your homeexplaining in detail what factors influence on evaluation. Our goal is to allow the seller to better understand the market dynamics, thus offering him the tools to make the right decision.
How does the matching process between buyers and sellers work? Is it just a question of delimiting the area and the possibility of spending, or are there other important variables?
“Casavo uses a proprietary algorithm, called Barberwhich through machine learning mechanisms analyzes the so-called buyer interest, i.e. the set of all search preferences of a buyer (price range, area, characteristics of the property), which are then matched to the characteristics of the properties valued on the Casavo platform, including the area and the market pricethus facilitating the direct meeting between supply and demand through a fully digital processalso made possible by the close collaboration we have with local real estate agents”.
What are the main advantages of using a tool like Casavo? Are you saving only in terms of time or also in terms of money?
“Time and money are certainly key factors, and it is the user who is in control of the process, while having full support from Casavo at every stage. All that anyone who wants to sell a house has to do is go to the Casavo website or download the app and complete a few stepsstarting from the real estate valuation up to the buyer’s choice. In fact, the seller can choose directly from the app whether to sell to one of the buyers selected by the Casavo algorithm (thereby maximizing the selling price), or directly a Casavo (thus obtaining the 100% liquidity in a few days). The entire process is transparent and in the hands of the seller thanks to our app, and each of our offers is accompanied by market data to facilitate customers’ decisions and to educate them on the dynamics of the sector.
“For those who buy, one is available on the Casavo platform wide range of verified listings, both by Casa Casavo and owned by partner agencies. Casavo ads are characterized by innovative features such as i virtual tours and online booking of the visit to the property. We also make the life of home buyers easier with a series of complementary services including Casavo Mortgagesservice that simplifies one of the most difficult aspects of a sale and delivers support in finding the most suitable loan among those offered by our partner banks. Casavo Mutui in fact takes care of all the management of the necessary documentation and liaises daily with the chosen bank up to the notarial deed.
The National Youth Card
“I also want to name a very recent novelty, we have joined the National Youth Card: by choosing to buy a Casavo Casavo, or houses purchased directly from Casavo and renovated to be “ready to live”, the under 36 will be able to benefit from an offer thanks to which…