Navitas: Applications of AI in real estate + construction
Navitas Capital, a California-based investor in proptech, has written an exhaustive 34-page report which explores the firm’s belief that AI is on the verge of disrupting these massive technologically analogue industries.
The VC offers an investment thesis, case studies, key challenges and concerns, companies to watch, and real-world AI applications of which can be found in this edited extract.
The future potential of AI in the built world is already beginning to manifest itself in a variety of early applications and successes. Not surprisingly, some of the early wins are occurring through end-to-end vertical solutions that independently capture large, unique data sets, run Machine Learning or Deep Learning on these data sets, and create outputs that either replace or augment an existing solution on the market.
Some of the applications augment, or replace, a human activity today and some of them replace a machine-based solution that’s either more expensive, less accurate, or more cumbersome to deploy. Many of these solutions are market-ready, with the ability to create value to real estate and construction organisations immediately. Below we have highlighted a few key areas where we have seen promising AI services and solutions emerge:
Build + design
We believe AI can impact real estate from inception – namely the build and design process. From AI tools impacting digital architecture to the increase in affordable visualisation tools crafting a library of digital twins, the beginning stages of building are being more carefully documented than ever before. With the ability to learn from processes, contractors and developers are able to create standard best practices, better understand the timing of delays and bottlenecks, predict progress, and quote costs of construction with much more accuracy than before.
Moreover, we believe areas such as litigation, insurance, and lending – all immense factors / budget line items in construction – can be greatly impacted by AI, given their heavy dependence on the certainty of time and cost. Adjacent areas such as labour supply, behaviour and safety, materials procurement, blueprinting and job site layout, and space utilisation can be drastically optimised through insights generated by effective AI. Finally, documenting data on the types, aesthetics, locations, and driving criteria of desirable real estate is already beginning to alter developers’ behaviour and decisions around where and how to build.
Notable companies: OpenSpace, Matterport, Reconstruct, Versatile Natures, Agora
Buy + sell
The foreseeable end state of real estate data development parallels that of financial services, in which the influx of data leads to a process of standardisation around specific data benchmarks, helping to create institutional-grade, transparent data to support freely trading assets, or fractions of those assets, in the capital markets. While this represents the ‘holy grail’ of real estate, few have addressed head-on the more mundane and headache-inducing aspects of data integration.
Likened to the fibre optic cables laying the infrastructure for high-frequency trading, real estate data requires a process of clean up and rewiring to truly bring real estate to the markets. We see a variety of companies tackling this area, with some focused on creating the front-end interface that can generate actionable and predictive insights, others focused on the middleware and data ingestion stage, and others on training its model to deliver next-generation AVM tools, Automated Valuation Models, to automate buying and selling decisions.
Notable companies: Cherre, Skyline, HouseCanary, Reonomy, Bowery
Operate + maintain
From leasing to maintenance and operations to payment processing, the property management industry alone is worth $22bn. That being said, 70% of all owners are not using a property manager and are instead electing to self-manage. Property management fees can stack up, from basic management fees to leasing fees, eviction fees, advertising fees, maintenance fees, etc.; the list goes on, typically setting the owner back 6-12% of monthly rent. However, self-managing demands a huge time commitment, with the industry average at 47.5 hours of leasing and 46.6 hours for management per rental unit.
Therefore, while many have the impression real estate simply requires patience and long-term capital to reap rewards, the operations, repair and maintenance of real estate assets remain huge cost centres that are both labour-intensive and far from optimised. In recent years, several solutions have emerged to help automate and standardise certain processes including maintenance, inspections, leasing and showing, and workflow processes, increasing efficiency by as much as 70%. In addition to simply automating standard workflow processes, AI-enhanced software offerings can better recognise patterns and predict tenant inbounds, requests, repair issues, late payments, etc., thus increasing occupancy rates and retention while simultaneously lowering costs by reducing headcount and / or increasing the number of units per staff under management. We see huge opportunities particularly in lead generation and conversion within the leasing space, given the abundance of new data sources allowing tenant demands and qualifications to be more closely matched with identified properties and automating follow-ups to close deals at higher conversion rates.
Notable companies: MeetElise, Truss, AskPorter, Squadvoice, LeasePilot
Perhaps most compelling is the ability to optimise real estate construction, development, and utilisation in an unprecedented way. Increased data on foot traffic, space utilisation, and tenant behaviour has unlocked many opportunities for developing next-generation space. As more and more of the workforce becomes remote, the purpose of clearly delineated spaces (i.e. an office space, a coffee shop, a residential studio) is becoming blurred. On the one hand, this could mean that owners and employers are losing control. On the other, this represents a large opportunity for those that understand and execute on data to offer multi-purpose spaces with the possibility of capturing more of a person’s daily life and of drawing in more diversified populations to particular spaces.
Understanding patterns on how people interact with their built environment unlocks countless opportunities, whether in building optimisation (i.e. turning off the lights at 10pm and increasing AC at 3pm), inspection and underwriting (knowing the 3rd floor houses heavy machinery, putting pressure on certain support structures), or architecture and design (i.e. knowing where to incorporate green space and common areas for optimal usage). Moreover, as we enter the Amazon age of e-commerce and endless package delivery, advances in AI and its applications on route optimisation, supply chain, delivery, and fulfilment has led to the rise of creative uses of space for fulfilment, warehousing, and last mile logistics.
As a final note, while insights on cognitive patterns of human behaviour are key to optimising real estate, we also see the emergence of robotics (AI in mimicking human motor functions) to be an equally key driver in optimising construction, as well as maintenance & repair, inspections, appraisals, and other labour-intensive and, therefore, expensive processes. We have been excited by certain robotics solutions that offer better, faster, cheaper and safer solutions offering enhanced tools of the trade to help enable construction workers, rather than displace workers.
Notable companies: SpaceIQ, Gyana, Enodo, Siteaware, Aquicore