Generative AI is building the foundation of proptech’s next wave

For artificial intelligence, 2022 was a year of breakthroughs. Image generation models such as DALL-E, MidJourney and StableDiffusion came in early in the year, garnering much attention, and ChatGPT went viral near the end.

Riding on the euphoria generated by these technological developments, about $49 billion in venture capital was invested in AI in 2022 — 40% more than a year earlier, per CB Insights.

Yet, there has been little conversation about how AI will play a growing role in real estate, a more than $50 trillion asset class, and one of the key drivers of the global economy. We believe this represents a significant opportunity for real estate tech entrepreneurs.

AI’s emergence will cut through material use cases in real estate tech, from search and listings to mortgages, construction and sustainability.

Notably, some of the most valuable companies in the early years of the real estate tech cycle have created significant value across the subsectors listed below. All of that will be in play with AI in the future.

Residential search and listings

Google’s first real threat to its Search product could come through Bing’s integration with ChatGPT.

That said, both Search and Bing are not tailored for real estate, which partly explains why Zillow, Redfin and StreetEasy have become valuable businesses. There’s a significant opportunity for an ML-enabled search and listings engine that leverages large language models, integrates with MLS providers and provides more robust results for buyers and renters.

Real estate brokerages

We believe real estate will always need the consultative hand of brokers. They are invaluable and cannot be replaced when an individual or family is making the largest financial decision of their lives.

Yet, a number of services provided by brokers and brokerages can be automated in a similarly personalized and consultative manner. AI-powered chatbots that power real estate brokerages have great potential to disrupt this marketplace.

Mortgage marketplaces and underwriting

The single-family mortgage market is estimated to be worth more than $13 trillion in the United States alone.

Mortgage search and underwriting have gotten better over the years, but there’s room for improvement. For one, the industry stands out for its abject lack of personalization.

AI has the ability to create and work off infinite customer personas, providing more robust search and underwriting solutions.

Renters’ and homeowners’ insurance

Landlords and mortgage lenders typically mandate renters/buyers get an insurance policy before they move in.

Unlike real estate brokerages, where the agent’s role is critical, it is our belief that AI can completely automate the insurance layer, especially as it relates to renters’ and homeowners’ insurance policies.

These products are relatively cheaper and not as complex, and ML-tooled bots can improve the customer’s journey from acquisition and underwriting to policy administration and claims management.

Companies like Lemonade have given us a glimpse of what’s possible with Maya AI, but we have only gotten started.

Construction estimation, bids and materials

The world is going to add 2 trillion square feet of real estate by 2060 — about one New York City — every month for the next 37 years!

Take a moment to think about the amount of data the construction industry will generate over the next few years. Then, consider the existing BIM and BOM models and current paper/spreadsheet-based estimation and bidding tools, and their technical sophistication.

We are not going to replace general contractors at the job site, but general contractors that don’t partner with AI companies to leverage their own data will be at a competitive disadvantage in the years to come.

Sustainable construction

The built world accounts for about 40% of global greenhouse emissions, and with 2 trillion square feet of additional real estate coming up, things aren’t looking good.

Part of the problem in solving emissions from the built world is that there’s only as much we can do with existing real estate — emissions that have been already operationalized in the environment. A more effective solution is to embed sustainability at the point of inception of the project, when a building is still in its design stages.

Layering AI into an architect’s workflow to determine emissions outcomes across scenarios and subsequently make recommendations for triaging cost, zoning and sustainability is going to be critical to how the built world deals with climate change.

A moment in time

Considering the significant opportunity for real estate and AI today, we believe startups are better positioned compared to legacy real estate technology companies looking to add AI to their existing product mix.

Entrepreneur and author Elad Gil has written extensively on the nature of companies the AI revolution will birth, and he draws a distinction between two categories:

1. De novo applications built on top of large language models by startups that don’t exist today but will thrive in the years to come — for example, an AI-enabled new real estate search platform with a distinct UI/UX.
2. Incumbent products that add AI/ML tooling to remain competitive in the market and retain distribution — for instance, Zillow injects AI into its search feed but largely retains its product functionality.

When it comes to real estate tech, it is crucial to juxtapose Gil’s distinction with how 2022 panned out for incumbents in the industry. Layoffs abounded in real estate tech last year as companies sought to preserve burn and refocus on their core offerings. An index of 17 listed real estate technology companies was down more than 80% from their peak valuation, many of which went public via SPACs in the recent past.

At a time when incumbents in the space are grappling with challenging conditions, it is tough to envision existing players effectively adapting AI in a meaningful fashion this year. Our analysis indicates that mature companies are looking to play defense and preserve their core offerings, which rules out the possibility of them embracing AI for their existing products.

This creates a unique and urgent window for startups to build de novo applications for real estate with AI at the core. The technology is not perfect, but it is developing at breakneck speed. We have entered an era when programming is moving from imperative to declarative code, expediting product cycles and feedback loops in unprecedented fashion.

In all of this, the opportunity for entrepreneurs in real estate tech across search, listings, mortgage, insurance, construction and sustainability is the kind that shows up once a generation.