OfficeBlocks Director Yusuf Jafry talks to Jumpstart about its property technology (proptech) and the problem it is trying to solve.
Proptech is a relatively new field that came into being around the late 1990s. But it was in 2004 that the industry really began to take off. Within a short duration, proptech became a frenzy that sent entrepreneurs and investors scurrying to capture the market.
Between 2012 and 2016, the industry started to mature. The number of proptech firms established each year declined as startups with similar offerings began to wind down. But at the same time, deal sizes in proptech increased, and the number of deals reduced. Investors became cautious about which companies they bet on, and began to fund companies with real potential.
Over the past few years, proptech has once again started to gain importance and prominence as governments began implementing initiatives for technological integration in real estate and worked toward creating smart cities.
To capitalize on the massive traction in this segment, JLL and Risk Integrated launched Singapore-headquartered proptech company OfficeBlocks last month. JLL is a commercial real estate giant, while Risk Integrated offers credit risk management for commercial property investments.
While there are several technological clusters that form the proptech ecosystem, data analytics and smart real estate are the biggest segments, with the highest number of companies and investments. According to the PropTech 2020 report by Unissu, data analytics companies snatched the most investment in proptech, raising $3.3 billion in 2019.
OfficeBlocks’ market intelligence platform and mobile app provide insights like the age and size of a building, estimated rent and value, floorspace data, availability and comparable alternatives of buildings across the Asia-Pacific (APAC).
The company is currently targeting all tier-one cities in APAC and is live across major cities including Hong Kong, Singapore, Shanghai, Seoul, Mumbai, Bangkok, Sydney, and Melbourne. The company is currently looking into global expansion as well.
Understanding how OfficeBlocks Uses AI and ML
Jumpstart spoke to the Managing Director of Risk Integrated and OfficeBlocks Director Yusuf Jafry to find out more about how OfficeBlocks works.
OfficeBlocks uses artificial intelligence (AI), machine learning (ML) and advanced analytics to provide accurate information about commercial properties. Jafry simplifies the explanation of how OfficeBlocks uses AI and ML by citing the example of weather prediction.
Weather is generally predicted using models based on the laws of physics. These prediction models are largely driven by the atmospheric processes that determine the weather in a particular region. But with AI, it is possible to forecast the weather of an area purely by analysis of past data and without the use of any physical law-based assumptions.
“The bottom line is, you don’t give [the AI] any ‘knowledge,’ if you will – you don’t make any assumptions, you just put the data in,” says Jafry. The results are, therefore, “…based on historical precedent, not laws of physics,” he adds.
This has been made possible, he says, with the power of modern computing. As technical capability has increased and more powerful machines have become more reasonably-priced, he says these data-driven models can be built on almost anything.
According to Jafry, commercial real estate giants like JLL have a vast amount of high-quality data “that is just ripe for the picking.” So, JLL and Risk Integrated worked together to apply prediction models and algorithms to the commercial real estate sector and established OfficeBlocks.
“In the OfficeBlocks product suite, just based on data and these AI models, you can get a rental or price estimate of any building anywhere in the coverage zone,” he says.
“You can use [the] mobile app, walk down the street, take a photograph of an office building in Tokyo,” he says. “And without having to call a broker, without having to do a Google search, without having to go to City Hall, without having to do anything except hit the little button on your phone, you will get a rather accurate estimate of the value or rental value of that building in today’s market.”
Interestingly, Jafry considers himself an AI skeptic who thinks of it as “artificial non-intelligence.”
“I don’t think AI is magic. I don’t think it’s particularly intelligent,” he says. “But what it is very good at is making use of very large data sets in a mathematically consistent way, to make these correlations, to make these regressions in order to make these predictions.”
What is this vast data that OfficeBlocks uses?
According to Jafry, OfficeBlocks uses two classes of data. The first class includes information that allows it to assess the effective value of a building by just looking at it through photographs.
“And the data that we capture, in order to train those AI models, is effectively just a bunch of photographs,” he says. “Basically, we just take hundreds of thousands or millions of photographs of buildings – buildings that are anywhere in the market.”
The second class of data includes the historical information or metadata associated with the buildings, from which a number of building attributes can be gleaned. These include the age and location of the building, the qualitative grade assigned to it by brokers, the number of floors and the basic size of the building.
“So we take the photographs, we take all of those attributes and we build an image classification model using exactly the same technologies that you use for image classification in AI in all fields,” he says. Image classification is used in medical imaging to detect tumors, and even in autonomous vehicles to detect road signs and visual scenes in front of the vehicle.
These two classes of data are linked together to create a deep neural network or a family of deep neural networks. Once trained, these networks are capable of offering insights on buildings whose photographs may not even be in its records.
OfficeBlocks has “got these AI models that can take any building, whether it’s in the database or not, and give you an accurate rent estimate, once it’s been trained,” says Jafry.
“We’re almost trying to do what a human would do. If you’re walking down the street and you see a building, you just look at it, [and] we can somehow in our brains assess a lot of information about the value of that building, just from that visual scene,” he adds.
This forms the core of OfficeBlocks’ mobile application: its model attempts to recreate how humans perceive a building and its physical attributes while walking down a street.
But the value lies in offering estimates of the price or the rent of the building for investors, says Jafry. This is achieved by creating another set of models using deep neural networks and ‘forests’ which are a random combination of decision trees. The data fed into these models include physical attributes of buildings from JLL’s archives and the historical data of rents and building values.
The data from JLL archives and the information about rent and property values are updated weekly on OfficeBlocks at present, although the company is striving to achieve daily updates.
“OfficeBlocks is retrained every week and is automatically made live. And so it’s an agile set of models,” says Jafry.
The problem OfficeBlocks is trying to solve
The decision-making process for investments in commercial properties is largely manual at present. According to Jafry, any investor, who may be located anywhere in the world, usually resorts to an ad hoc Google search to identify places of interest.
Investors may sometimes also flip through portfolio magazines to find areas of interest, which they then look up on the Internet to narrow down choices. They usually then contact a regional broker, who takes over the process and sees the transaction through. The entire process involves minimum use of technology.
The main aim of OfficeBlocks, therefore, is to make the process of making investment property decisions easy and efficient through the use of technology. It allows investors to assess the potential value in the assets of interest in their regions of interest from their desktop or mobile phone, without any hassle, says Jafry.
“The time-consuming groundwork, the ad hoc nature of it has all been removed or largely been removed by OfficeBlocks,” says Jafry.
OfficeBlocks, therefore, is attempting to streamline the initial property research process, providing compare/contrast features that may be of use to investors.
“We’re not saying that we’re going to replace brokers or anything like that,” clarifies Jafry. Using the information provided by OfficeBlocks, investors can make informed choices before contacting a broker and moving ahead with the transaction.
Moreover, OfficeBlocks’ portfolio intelligence platform allows investors to assess how the assets can affect their portfolio in terms of risks and returns, says Jafry. The platform analyzes risk by measuring expected returns, historical variances, and market correlations.
The OfficeBlocks risk intelligence platform is a derivative of Risk Integrated’s flagship product, which is being integrated with other tools. It is a “detailed risk calculator” and the company says it has been used by insurance companies, banks and even government regulators to identify and understand investment risks.
However, it is early days for OfficeBlocks and the platform has been launched with just the first set of products, says Jafry. But the aim is to introduce the technologies and capabilities prevalent in other sectors to the commercial properties market, he adds. His vision is to make OfficeBlocks the go-to option for commercial property investment-related information.
COVID-19 and commercial real estate investments
The global pandemic has impacted a wide range of industries, including commercial real estate. But investors expect a full recovery by next year. According to JLL’s investor survey, 84% of respondents expect transaction volumes to recover fully by the second half of 2021.
78% of Hong Kong investors JLL spoke to said they will be investing the same amount this year as 2019. In fact, 4% said they plan to increase their investment commitment.
In Singapore, this confidence is more pronounced – 52% expect to see no change in their investments this year, and 44% actually expect to see an increase.
Moreover, work-from-home norms could potentially affect commercial property investment. However, according to a JLL survey of employed professionals, at the height of the pandemic, approximately 68% of respondents across APAC worked remotely. But 61% of them said they miss going to office and would prefer a hybrid model with flexible work arrangements in the future.
OfficeBlocks Director and JLL Chief Research Officer Roddy Allan told Jumpstart, “Although COVID-19 has impacted property investments and transactions, more than three-quarters of commercial real estate investors we have spoken to expect investment volumes to rebound by the end of the first half of 2021.”
“We’re already seeing signs of recovery. Asia Pacific’s investment volumes most recently rebounded 35% in Q3 from Q2. We also predict that investors will accelerate investment in key Asia Pacific cities driven by the compelling underlying fundamentals in this region.”
The fundamentals Allan talks about refer to socioeconomic conditions like rapid urbanization, an expanding middle class, a large youth population and improving governance – all of which boost commercial real estate growth.
In conclusion, OfficeBlocks is trying to modernize an industry that is riddled with inefficient manual processes. Just like the travel industry was revolutionized by the introduction of technology, Jafry believes the same will happen to the commercial real estate market.
“Being prepared, getting in there early, building the tools now and improving them over time – that’s our vision,” says Jafry. “That’s what we’re hoping to do with OfficeBlocks. So that we’re in front of that wave [of modernization] rather than caught behind it.”
Header image courtesy of OfficeBlocks
