Artificial intelligence is reshaping how real estate investments work, and the risk of getting left behind is becoming more and more significant, says AppFolio's Mike Sebastian.
Ever since ChatGPT launched in late 2022, AI has been a key focal point for businesses of all stripes, as they try to figure out precisely how important this development will be for their operations, and where potential efficiencies might lie.
The same is certainly true of the real estate industry, and the private equity real estate market too, as operators and investors alike get to grips with AI. According to Mike Sebastian, industry principal and director at real estate platform AppFolio, it is a technology that holds revolutionary potential - and that threatens to leave late adopters behind.
Understanding the timeline for these developments will let you distinguish the "hype" from what's real and what has staying power. I am in the camp that agrees that AI will be the next revolutionary wave of technology - as significant, or even more than, the internet. Right now, it is in its infancy, like the first websites, many of which were unusable. But nowadays, most people could not conceive of doing their job without the internet, even after the 2000 dot-com bubble imploded.
AI is going to follow that same pattern. And I will tell you why: the breakthrough started when ChatGPT 3 came out, allowing ordinary consumers to say, "Oh my gosh, I can talk to my computer." The effort going into developing that technology further is unbelievable; we are getting new models weekly.
But where is this technology useful and where is it overhyped? In every circle, not just real estate, you hear about Artificial General Intelligence (AGI), or a machine learning model that can both plan and carry out tasks on a general basis, not specific to one thing. That does not exist yet. That is the most obvious example of hype.
The real revolution is language models becoming a user interface. In time, they will expand to become a part of a larger system, interacting with other models. These used to be purely probabilistic models, and that was a problem. They weren't consistent, and they would make up data. The newer techniques are fine-tuning these models to make them work with instructions, or a specific set of data, which in turn makes the model more deterministic, less probabilistic. That way, users can truly rely on the information that AI is giving you - and that is revolutionary.
We see AI as having three main applications in proptech: one, talking to your data to get exactly what you need; two, actionable items like streamlining communications; and three, building workflows, or describing what you want done and having the AI do it on a schedule.
We have a whole AI research department focused on this. They have been able to fine-tune it for project management. So, instead of combing through thousands of reports and looking for problems, you can say to the interface, "Show me all my work orders" and then have the same interface write and send mass texts all at once directly to your tenants.
In short, then, AI cleans up the busywork in real estate management, so that the team can focus on more important things.
This is one of the biggest challenges in private real estate: if you are a fund manager, and you do not do management in-house, you have firms sending you mountains of data, each in their own system. Sure, it might be the same data points, but it is all in different places with different labels. Maybe rent-4001 is rent-4003 in a different manager's account. And then you have to clean all of this by hand and normalize it - which is a huge cost and leaves the potential for manual error.
AI can take all these different spreadsheets, then clean and match them up in a normalized way. That is not a future hope; we have gotten it working. Our program matches up all these disparate reports within minutes, even with slightly different wording or new accounts. Since the asset manager's real job is not to clean the data but to look at it and find problems and ways to solve them, AI saves hours of work.
There will always need to be a human in the loop. Language models do not plan, and they do not reason. And crucially, the AI cannot tell the asset manager what the problems are or how to solve them, or take action to solve them itself.
Automation has a couple different effects. Personally, I think AI will help real estate managers do their jobs better. Will it eliminate jobs? Potentially, but think back to the 1950s and 60s: you now have bunch of jobs that do not exist any more because we have moved on from typewriters and punch-cards. Now those people do other jobs - other jobs created by the new technologies. AI will do the same thing, I believe: create more jobs, just different ones.
"If we are not managing real estate investments by hand, then in theory this can be a near-zero-margin business"
I also believe that eliminating the tedious work will make people happier in the jobs they do. If I'm a property manager, I spend an enormous amount of time talking to tenants. If I'm managing a 500-unit apartment building, I'm spending a lot of time having the same conversations over and over, looking at repair orders, answering questions that tenants could theoretically find out on their own.
AI would streamline all of that. We have a tool that allows our clients to chat to an AI agent, and it will put in a work order for them. Now I just saved five minutes, but the person still feels like they were served. Both sides get what they want. It makes for a higher level of service for the tenants, and it reduces my busywork as a manager, so I can focus on the things that add value to the property and make me happier.
Then, looking at it from the GP side, if I'm raising capital, let's say I need to market to 1,000 investors. I have to enter a ton of data to market this fund. But now I can take my program, load it in, and the system will parse and automatically fill out most fields, saving me hours or days. Then I can have the same model summarize the information and write the email to investors for me so that I can give it a review.
It is not actually changing what GPs do; it is just making their lives better. With that busywork out of the way, I can more easily get to taking the calls and building the relationships. After all, AI cannot have a relationship with the people.
Technology reduces the time needed to do tasks. It allows scaling and allows us to do more with less. And it can reduce outsourcing, too. It comes down to saving money and saving time.
Also, think about what your competitors are doing. If you can see a deal faster, have all your data in one place, email your investors immediately and raise capital in a day, you can beat out all the other bidders. You will also be able to tell a better, more thorough story to your investors around what you are doing with their money. If you can tell them a better story, with more and better-organized data, they are going to trust you more. So much comes down to trust in our world.
In the modern market, if a GP tries to resist adopting AI, I do not think they will be able to scale. They will simply not grow at the same rate as their competitors and will ultimately be relegated to small businesses. Other companies are now scaling and have new ways to raise capital faster. After all, if we are not managing real estate investments by hand, then in theory this can be a near-zero-margin business. Sometimes investments lose money - that is inevitable - but with AI we can at least cut costs.
That is especially important in the current market, as it is hard to find deals to pencil. With all the constraints on the market, like insurance and loan rates, returns are getting compressed. We are also not seeing as much distress in the market as we thought we would, so fewer deals are coming that way. That means GPs will need to focus on providing quality service to tenants and investors, so that they want to be repeat customers.
The biggest challenge is educational. Not enough investors or managers understand what it can do for them. So it is going to take broad market education.
Proptech has been overhyped, too, like AGI. A lot of people might misconstrue the hype in one field and assume that AI and proptech are overhyped in general, and thus overlook the potential revolutions that their business could be taking advantage of. So, I say we overcome hesitation by demonstration. The only way to break those misconceptions is to show off AI's genuine potential for the real estate industry in practice.