Minna Song is the co-founder and CEO of EliseAI, a company using advanced natural language processing to automate communication across the multifamily industry—and beyond. Originally launched as MeetElise, the company powers AI leasing agents for over 350 customers, including 70% of the 50 largest rental housing operators in the U.S., but Song’s vision goes far beyond leasing.
In this episode, she sits down with Brad Hargreaves to unpack the evolution of EliseAI, from a chatbot into an enterprise-grade operating system for housing. They talk about the gap between software teams and site teams, how operators misjudge AI timelines, and what it really takes to earn the trust of institutional clients. Song also shares her thoughts on VC fundraising, breaking into healthcare, and why she thinks real estate is finally waking up to first-principles product design.
If you care about the future of operations, automation, or AI-native infrastructure, this episode is essential listening.
You can see the full video here on Substack or watch and listen on any of the following platforms:
The Thesis Driven Leader Series is made possible with the support of Neutral.
Neutral is redefining multifamily real estate with a focus on sustainability, resident health and well-being. For example, Neutral is building the tallest mass timber and Passive House residential building in the U.S with a state-of-art wellness club in Milwaukee. Beyond environmental impact, Neutral offers investors access to substantial sustainable tax credits and deductions. Accredited investors can explore available opportunities at invest.neutral.us or connect directly with their team to learn more.
The following transcript is automatically generated. Please forgive us for any errors or misspellings.
Brad Hargreaves: [00:00:00] Hello and welcome to the Thesis Driven Leader Series. I'm Brad Hargraves, the founder and editor-in-chief of Thesis Driven, and your host here for the Leader series today. Thank you all so much for joining us today. We're joined by the founder of one of the most interesting PropTech companies out there, and certainly one of the most successful of the past few years.
Minna Song is the co-founder and CEO of Elise AI. You may have seen, we did a deep dive on Elise AI a few weeks back where we really focused on the interplay between what Elise AI is doing from a technology standpoint and this broader trend of centralized operations in multifamily.
That is more multifamily operators are taking tasks that have historically been completed by team members on-site, moving them off-site to centralized offices, whether here or offshore, and that both unlocks new technology as well as requires new automation and tech to manage. Elise AI, partly [00:01:00] because of what's happening with centralization, partly because of what's happening with AI, is one of the rising stars of PropTech.
They're widely seen as one of, if not the best, AI tools, particularly from multifamily, out there. And they raised a $75 million round of financing last year, which is really notable given how difficult it is to raise venture growth rounds for PropTech companies right now. So it's really saying something that they were able to go out and land a round like that.
But one thing to understand about Elise AI and Minna Song is that they bet on AI well before it was cool and every company had .AI in their domain, and they're seeing dividends from that today. Go back 10 years ago, a lot of PropTech companies were building chatbots. These were basically leasing or customer service assistants that were programmed in a very traditional manner. If this, then that, they would see the words pet [00:02:00] and rent in a message and assume that the customer was asking about pet rent and give an answer accordingly. Yeah, they're a little bit more complicated than that, but not much more complicated. Elise AI didn't go that route. They decided to go with AI from the beginning, even when AI was not really that much of a thing back in 2016, 2017, 2018.
And so, while those kind of programmed interfaces did well in the early days, the chatbots have had an advantage. They could be controlled, they were predictable. LLMs have advanced over the past, three or four years to a point where they're incontrovertibly better than the programmed version. They can produce human-like responses, they can deal with unstructured requests, data, et cetera. So Elise was early to that, and they've since expanded into a wide variety of use cases outside of AI, from customer support to collections, to even reporting. And interestingly enough, they just launched a new [00:03:00] vertical in healthcare that I'm excited to discuss with Minna. So we're gonna talk about tech, we're gonna talk about where multifamily is headed, state of PropTech, as well as Minna's journey as a founder.
But before we welcome Minna to the show, I do want to thank our sponsor for just a moment. This would not be possible, the Thesis Driven Leader Series without the support of Neutral. They're multifamily developer redefining sustainable real estate with that focus on resident health, well-being, they are building the tallest mass timber and passive house residential building in the U.S., with state-of-the-art wellness club in Milwaukee, they operate in other cities like Madison. They've done some cool projects, but beyond environmental impact, Neutral offers investors access to sustainable tax credits and deductions. Definitely chat with them learn about what they have at invest.neutral.us.
Now, really excited to invite Minna Song, co-founder and CEO of Elise AI to the show.
Minna, welcome to the show. So excited to have you on today.
Minna Song: Thank you for having me. I'm excited.
Brad Hargreaves: So, I'd really [00:04:00] love to hear the story of Elise AI getting off the ground because you have a technical background. You and your co-founder both have technical backgrounds, but you're not from the real estate industry.
Often, you look at startups out there, people from tech who come into real estate, it doesn't end well, but you've managed to figure it out. So tell me a little bit about some of the things you did in the early days and what you attribute that ' figuring it out' to?
Minna Song: Yeah. Certainly didn't come from the industry.
We started in multifamily because, when I met my co-founder, we decided we wanted to start a company and thought from first principles, how can we maximize our impact in this world? And we both come from software engineering backgrounds, met studying computer science in college, actually, so we decided we would build technology for industries that serve fundamental human needs. That was the concept and what are our thinking about the hierarchy of needs, right? It's food, water, and shelter. And so we picked shelter, and that is the residential real estate industry. So that's how we picked the [00:05:00] industry.
We didn't wanna build products that didn't really solve a problem, and that's probably pretty likely if you start building products for something you know nothing about. How we got started was, I took a job working at a real estate firm in Manhattan, actually, where our office is based, our headquarters is based, and I was just there to listen and understand what the pain points were.
It was really research into the people, the problems, the bottlenecks in the process. And honestly, in the first couple weeks, it was pretty obvious that the biggest thing that people were complaining about all the time was not getting responses. Back and forth communication being very slow, asynchronous, and just falling off a lot of the time.
So we thought, hey, AI is getting good in research. It wasn't really a thing like it is today. We started initial research about 2016. We got very lucky on timing quite honestly. 2017 was the year the transformer was invented. Transformer is the T in GPT, so that's really the architecture that unlocked a lot of the amazing inventions that have come out since then.
Brad Hargreaves: So that was the [00:06:00] big change that happened. And you were going into this not necessarily understanding that AI was going to have the moment it has had. What got you into AI? Not many people were talking about AI outside of research and academic and technical circles back in 2016. What did you see?
Minna Song: Yeah ,old techniques worked, not nearly as well, but they could produce commercially viable results, and so we knew that we could build something.
I think what we ended up building is much greater because of the acceleration of the advancement of AI. We knew that it was getting funded pretty heavily and a lot of great research was being done because we had ties to the academic circles through the universities that we attended, so, had a good friend working with professors actually working on some of these newer models and telling us the results were outperforming anything that existed. It really did come from early indication from academia.
Brad Hargreaves: So these were text-based models, you were at university. Where were you in school again?
Minna Song: I went to MIT.
Brad Hargreaves: So you're at MIT, which is where a lot of this stuff is happening. And when you say funding, [00:07:00] you're not necessarily talking about venture checks getting written, a little bit of that was happening, but not much. This was academic grants.
Minna Song: Exactly.
Brad Hargreaves: So you took it straight from academic research at MIT to working at a real estate firm in the middle of New York City and applying that theory into practice.
Minna Song: Yeah.
Brad Hargreaves: And so at the time, just to understand the context, this was when there was a boom in these chatbots, and you always have been careful to distinguish between Elise AI, which I think a lot of people say at the time where like lumping in with that trend, versus what other people were doing creating these programmatic things.
So maybe can you explain for the audience a little bit of that distinction and how you draw that?
Minna Song: Yeah, I think chatbots really refer to the old age of how people interacted through those very decision-making, choose-your-own-adventure, keyword-matching chatbots, not really true natural language input.
And we made sure that people understood that [00:08:00] what they were thinking of when they thought chatbot was not the same product that we would be delivering them, not the same customer experience. Those older chatbots have essentially a lot of limitations, can only handle a subset of questions, have predetermined set of answers, and just generally frustrating, like we've all experienced.
Brad Hargreaves: Just so our audience is clear, when we talk about these, they're fundamentally different technologies. So a chat bot would see the words pet and rent and assume that a prospective tenant is asking about pet rent, which is different than how Elise AI works.
Minna Song: Completely different. If you spell something incorrectly, it is not confused by that or not finding the response. It's really much more like your experience with ChatGPT. It's taking in the meaning even if you're not communicating very clearly, and it's giving you a comprehensive response, right?
It's not giving you a predetermined response. So you can ask it multiple questions, it can speak many different languages. You don't have to ask it to speak Spanish. [00:09:00] It will just pick up that you're speaking Spanish, so a bunch of benefits that you get from being really a true natural language conversation.
Brad Hargreaves: Obviously, a lot has changed since 2017 when you started the company and started building Elise AI and from this AI base. Tell me a little bit about, have those changes and how you've ridden the wave of AI going from something that is in academia to the most popular real estate tech trend out there. How has that impacted your company?
Minna Song: it's been a completely wild ride. When we were first getting started, AI didn't mean anything. A lot of people didn't know that AI stood for artificial intelligence and definitely really took off in late 2022. In 2017 we had customers, we really started with the smaller mid-size operators and then starting getting into larger operators and really, it wasn't about the technology, it was just about the problems that they had to solve. Avalon Bay and Equity Residential were [00:10:00] some of our early customers, and honestly, they were attracted to being able to solve this at scale for the first time. Any other technology in history had not been able to help them really solve the problem.
It changed a lot since 2017. We used to say this thing where our goal was to make AI for leasing the status quo. It sounded extremely futuristic at that point, but now it's a must-have for anyone in the property management industry. It's clear that AI is never gonna go away, and not just leasing either, but across the entire resident life cycle.
I'd say in late 2022, when ChatGPT came out, it really changed people's minds because they became familiar with AI as a consumer product, right? Before, it was very much the thought that AI is never going to do something as well as a human, but when people started interacting with it and saying, "oh, actually, it's writes emails better than I write emails, so I want it write all of my emails. Oh, maybe I can have it answer all of my leads as well," or "maybe I can have it handle all of my resident requests, complaints..." whatever it might be. And with no emotion and no mistakes. That [00:11:00] change in mindset was really powerful for us.
Brad Hargreaves: So what were some of those early customer conversations like? You mentioned to me when we were chatting a couple weeks ago, you sat in Avalon Bay's office down in the D.C. area for a little bit to just understand how they were working. That was once again, pre-pandemic 2017 - 2018 when AI was still very much an academic thing. It wasn't as hot as it is today and maybe even a little scary for a lot of people.
So what were those initial conversations like? We're gonna talk to your leads with AI.
Minna Song: You couldn't really just tell our early customers about it. Just showing them. I think it does the talking for itself, literally.
Brad Hargreaves: And by showing them, obviously it's having those one-on-one okay, cool. Play around with this and see how it does. But then you're gonna get to pilots, and roll it out at a building and see how it does. What was that initial data, those first experiments? What were those like?
Minna Song: It was pretty crazy. So our first customer was Stonehenge, based in New York City. When we launched the product with them, it was very MVP. We were on about five properties with them, and then in three [00:12:00] weeks, they said, "Get this on all of our properties as soon as you possibly can."
That's when we really knew we had something, for the first time, that when we knew we had product-market fit. We built out a lot of features with them, but when we worked with Avalon Bay and Equity Residential about the same time, we moved to Arlington, Virginia actually to work with Avalon. Worked very closely in their office with their team they dedicated to rolling this technology out. We were building features that really allowed this to scale to an enterprise level and enterprise-grade software.
We had done only pre-tour. When we were just serving a smaller market, then we had to expand it to how are we handling post-tour, getting these people all the way through lease, because just generating tours is not sufficient.
There were week after week, a ton of feature development. And at the time, it was just me and my co-founder; we were building the product . We didn't have other developers besides us. Really excited that we had this opportunity. Our whole company was three people. When Avalon and Equity said they were gonna roll out to full portfolios, we started hiring. We needed a bigger team to support this multi-billion dollars of real estate [00:13:00] asset. Then the ball started rolling, and more of these top 50 customers were looking to do what Avalon and Equity were doing.
Brad Hargreaves: Certainly, a great set of logos to start with. You have two of the more tech-forward big real estate owner-operators out there. Then you have Stonehenge, which is not as well known on a national scale, but certainly in New York. New York real estate is its own little bubble.
I also note that it is a very positive sign that you are able to break out of the New York Real Estate bubble because so many real estate companies that start in New York, build for New York, and New York is weird.
It's different than any other real estate market out there. So, getting a start with a New York real estate company and immediately adding on Avalon Bay and Equity Residential was probably a key to not just build inside the pool that is Manhattan real estate.
Minna Song: Yeah, we were very scared. A lot of people told us that, get outta New York. Don't sell to New York landlords. But from first principles, it made sense. Even outside of New York City, people have to lease apartments. They have to answer [00:14:00] questions about people interested in learning about their community.
They have to book appointments, they have to sign leases. Yes, those leases are strange in New York, that process and those questions are different. It was funny that, later on, customers asked us. Because we'd have, information they had to give us during onboarding. And there's " what is a child playroom? Do you mean playground?" So, it was like different amenities that had to change, but fundamentally, the product was the same and added the same value nationally.
Brad Hargreaves: That's funny. New York is is such a little bubble in many ways. So historically, a lot of people associated Elise with leasing, which is where it got its start, but you've expanded a lot outside of that at this point. Clearly there's a lot of little places within the customer journey in the management world of real estate that aI can be helpful. So tell me a little bit about how you discover those opportunities where you can add value, and how do you choose which ones to go after?
Minna Song: Yeah, we listen to our customers. We think a lot about automation. How [00:15:00] can we get a really high level of automation for the workflows that our customers' teams are doing on-site, because they're overwhelmed, they wanna take different responsibilities off their plate so they can focus on all the good ones, right? Spending a lot of time on-site with teams is the strongest signal of finding product-market fit with all the new products that we've built. So we do leasing, we extended to a bunch of resident products a few years ago, handling delinquency collection, handling maintenance requests, handling lease renewals, then we've just expanded to a lot of the workflows that people are doing.
Brad Hargreaves: The delinquency collections as well?
Minna Song: Yes.
Brad Hargreaves: Is the AI pounding on the door saying, "You gotta pay?"
Minna Song: It is, which is a great task. It's a very automatable task and not something humans like to do. Then doing it really consistently and having a high-touch process for delinquency is particularly important.
And then expanded workflows to things that are also clearly automateable tasks, like scanning leases and if [00:16:00] there's discrepancies in the leases and what people are being charged, just general hygiene of the property. Then generating documents instead of having people copy and paste information to generate a document, AI can do that. These problems are solved in other industries, they just haven't been solutions brought to multifamily yet.
Brad Hargreaves: One interesting thing that I discovered when I was on the management side is the number of small outstanding balances that create delinquencies, and just " Hey, you have a $80 charge that hasn't been paid," and those misunderstandings causing delinquency.
I saw an interesting data point from one of your case studies about how a lot of people think about collections as someone who's three months late on rent. Well, a lot of it's not that.
A lot of it's someone who has a small outstanding balance and just doesn't know where it came from. AI being a really interesting way to solve that and make sure that the people who pound on the doors are pounding on the right doors.
Minna Song: Yeah a lot of people forget to pay. Our attention spans are very limited. So, the AI will reach out, and [00:17:00] if someone answers and says, "Hey, I'll do it later," the AI is gonna follow up at that time. They're like, "I'll do it tomorrow, end of day."
The AI is reaching back out and saying, "Hey, I haven't received it. Any issues, can I help?" And just staying on top of people and meeting them where they are is what drives a lot of the results. It's a lot about financial education. This was in your lease. Here's what happens when you get charged a late fee and try to prevent any of those behaviors.
It's honestly just a lot of education. People won't necessarily have the time to sit there and do it, but we can serve residents a lot better if the AI is handling all that.
Brad Hargreaves: One of the big themes of the piece I did on Elise AI, which came out a few weeks back in Thesis Driven, is this relationship between automation, AI, and centralization. They're married in a lot of ways of you're saving people time. We just gave the example of like delinquency and collections, you're saving people time on-site. If all the tasks are being done on-site, you can save someone a little bit of time, but you're probably not cutting headcount at the build.
But if that's a centralized role, where [00:18:00] they're sitting in a back office in Atlanta or Manila or wherever, there's a lot more value you can get from these automations and from this AI. At the same time, centralization makes it easier too. You can systematize, process it. So, I'd love to hear a little bit from you about how you see your customers combining centralization in your tech?
Minna Song: I think AI and centralization go hand in hand. A lot of people that I see doing centralization incorrectly just take someone off-site, give them the same job, just put them in a different location, either in a team or a remote person. Then they lose all the context, and communication between the people who are still on-site and the people who are centralized just completely breaks down. Residents also just don't know who to go to. You need AI to do a bunch of things.
One, you're not gonna get the efficiencies or the benefits of centralization unless you have tech. You can't say, Hey, one person in a different location. Do 10 buildings, 20 buildings' worth of work. You need AI that can do 95% of that work if you [00:19:00] expect someone to handle 20 buildings, and then they do the rest. 5% that really needs to be high-touch is very complex, requires a lot of human empathy or whatever that task might be. It's really important that you're adjusting the process that your centralized teams are doing.
The communication routing is really critical. When a renter inquires, the AI triages it to the right team. Is that a centralized responsibility, or is it the on-site team? Then, having the tools like a CRM that easily routed to the right person. And when someone walks into the community and talks to a person who's on-site, even if that role is handled by a centralized person, that team member needs to be able to quickly communicate.
That's where you need like a CRM platform that's built for centralization. And that's why we have a CRM that people can tag each other and tag the AI. And it all works seamlessly together, because it does become very complex organizationally if you start creating these off-site teams. Even if you're not outsourcing to other countries, just having people work from home, in the same region. [00:20:00] It becomes very complicated.
Brad Hargreaves: You're adding another level of communication when it's not everyone sitting in the same room together, but you also can get more efficiency and specialization from that.
Minna Song: Yeah. Every portfolio is different. We helped 50 customers centralize their properties, and I swear it's 50 different setups because no one-size-fits-all. They have small buildings far away. They have small buildings close together, large buildings, different classes of buildings, different teams.
It's all sorts of combinations of things. If you don't centralize the same way that another company centralized, you might not have the success, which I think you hear in the industry, right? People are like, "Oh, I tried it, but it didn't work for us."
No, because you need a slightly different flavor of centralization. Centralization came in waves. It's not a new idea. Most people I hear are talking about it, and if they haven't started it, they're feeling like they're behind.
Brad Hargreaves: I think a lot of people in multifamily today, and certainly a lot of people in multifamily tech, forget that there was a time that accounting, the books were done at the property too. Prior to the mid-nineties, you didn't have the [00:21:00] accounting software that you have today.
You didn't have the PMS ecosystem that you have today. Even that was done on-site. People quickly realized that having a roving accountant going from building to building, doing the physical books, is a bad idea and certainly inefficient. So that was centralized 20- 25 years ago.
So you're right. This has been an arc.
I'm curious what you said around different flavors of centralization or different forms of centralization for different types of owners. I'd love to maybe give you an example of that would be helpful because you hear about centralization discussed very generally.
It's okay, you can take these roles, you bring 'em into a back office. Maybe it's here, maybe it's somewhere else. It does feel very one-size-fits-all. So how are you experiencing that?
Minna Song: Some people have dense clusters of buildings, sharing leasing offices, and admin roles.
I see a lot of people do create centralized or work-from-home roles to retain their best staff, and back-office tasks, delinquency and other [00:22:00] functions like renewals. Some people get efficiencies without centralization. They just get efficiencies on-site. They're not backfilling a leasing agent that turns over or retaining the best staff because they're taking off the most mundane work from them. So even that I've heard is centralization, which maybe doesn't match the word, but there's all sorts of ways that you have to dig into with a customer when they say, we wanna centralize.
Okay, but what resources do you have? What does your staffing already look like? And is this the goal? is it increasing the performance of your asset or running with a leaner staff? All those goals that people have end up with different structures and I'd say technologies are different, right?
If you're doing the delinquency or ACM sort of traditional roles, collecting rent, that's a very different centralization model. You probably can have those people fully remote. If you're doing leasing, you need an AI or self-tour product, like our AI-guided tour product, to facilitate. Especially when people have like a roaming leasing [00:23:00] team or someone sharing resources between multiple teams that you have some that are A-B. You can access through a lockbox or a smart lock without a human actually being there, or to supplement a human, maybe that human goes there, but there's two tours booked at the same time. And actually one needs to be a self-tour in that case, because renters want the first time slot that they prefer. So every flavor of centralization for every different workflow can be quite different, and harder to implement, easier to implement depending on how aggressive people's goals are.
Brad Hargreaves: When you think about a centralization and Elise AI target customer anyone in multifamily could get value from these things. But when you see a customer say, "Hey, we want to, try this out. We're interested." What makes you say, "Wow, this is gonna be a slam dunk. This group is gonna get a ton of value out of it, and gonna be super easy and straightforward." Versus someone you're like, all right, this might take a little more handholding. What are the lines there?
Minna Song: Honestly, it's just mostly customers getting in their own head. We spend the first amount of time with them [00:24:00] really understanding what they mean and if they're committed to getting these goals. We've helped customers stand up a centralization team or a person and training them to be specialized in it.
Having someone dedicated to it just shows that the customer is really interested in doing it because it's not just implementing tech, but also rethinking what are the metrics that we're looking at, are those different, particularly when it comes to what the agents are doing. Maybe you're not measuring the touch points any longer with leads because AI is handling that.
So I'd say analysis paralysis. Those that are most successful at it just need to start and need to have a deadline for themselves. It seems simple, but it's the biggest differentiator between someone who's gonna be successful and someone who's not. I think that is a big piece of what makes someone successful at centralizing is because you do need the technology. Do you have the tech in place first, or do you try to centralize roles without technology and then layer technology in?
Fundamentally that just does not work because a few different reasons. One, you're not getting [00:25:00] the efficiencies, the communication, all those breakdowns, but two then you're working with vendors and saying, " Hey, I've done my way, now fit your solution to that." Versus seeing what the suppliers have already built out and the best practices from multiple different customers.
Everything has to be very configurable and customizable, but it might be actually better to come in with an open mind and say, " Hey, how can we start with technology? And how can we change our organization structure to centralize and to make sense within that?" So it's just are you reinventing the wheel or are you starting where people have already figured it out.
Brad Hargreaves: I love that. That's a really important point. I was giving a talk this morning, and responding to a question someone asked me about actually about AI and technology. The question was very much in the framing of how can AI improve the workflows I already have?
The most exciting stuff of this tech and what's happening is that it allows you to rethink the workflows entirely, and how you're doing what you're doing. When I see success and failure in the real estate world of embracing new [00:26:00] technology, the people who do it the best are those who take a de novo lens to every technology and say cool, now I know this is possible, knowing what I know now, what can I do and how would I design my workflows, my process, my org chart in a way that can make use of this technology.
Do you see multifamily and anyone in the multifamily industry thinking like that?
Minna Song: Increasingly some operators still view AI as a sort of a smarter widget or an add-on rather than really a true teammate that can run an entire workflow end-to-end. A lot of people rely as well on this patchwork of point solutions that solve very isolated problems, but they never talk to one another. Not enough customers think about technology from a renter experience.
And then how do I get that renter through the entire flow that they need, whether that's the move-in process or an entire end-to-end maintenance process from when they call [00:27:00] in, before they were talking to a person, but now they're talking to an AI, how does that work order get done and what's their experience in that process? They think a lot from the property staff point of view, and they're very isolated solutions making each one of those things more efficient.
But altogether, there's still bottlenecks between those things. So a lot more people are starting to adopt this sort of agentic or like autopilot model of operations, which unlocks an entirely different set of efficiency gains and then actually you're collecting so much more data.
Then operators are really starting to think, I've never had this amount of data before. I've never had this quality of data before. How can I use this to make predictions at scale?
And that's where the next big change is going to be.
Brad Hargreaves: So it leads really well to what I wanted to ask next. We have a good sense of what AI is capable of today, where it's been rolled out, multifamily owners embracing it, people in real estate embracing it to varying degrees.
What's next? What are we sleeping on that we're gonna see over the next two to [00:28:00] three years that might not be possible today, or might be on the cutting edge today?
Minna Song: Data is the next big thing. So is that predicting who's renewing? I've seen some customers start building some of these models themselves, but very few have the resources to do that.
Brad Hargreaves: It's like a predictive model of okay, if someone does X, Y, and Z, they're likely to renew if this one, versus someone else who may be less likely to renew.
Minna Song: Yeah. What are the indicators in their experience? The obvious ones being how's their maintenance experience been? But actually it's so many different touch points from starting from the leasing process that end up telling you if someone is going to renew, which then of course impacts your pricing and knowing how much supply you're gonna have on the market and marrying that within, what's the exposure of the building?
So, data, you're gonna see a lot more predictive decision making coming online. That's a lot of what we're working on as well. Resident sentiment. People [00:29:00] are surveyed out, gathering information from residents in a natural, organic form, you're texting about something, you have much more open line of communication. How can we use every single one of those interactions to get in front of your performance?
I think people are sleeping on data in the industry, but that's gonna change really soon.
Brad Hargreaves: Yeah. I assume you're building stuff around predictive renewals and that kind of side. Do you have a predictive renewal product?
Minna Song: We don't right now.
Brad Hargreaves: You don't. I assume it's coming. One other area I'm interested in is predictive maintenance, I think is super interesting. You have all this sensor data, this maintenance data, do you have a view on that? There's gotta be something there.
I was talking to some owners at AIM in a closed-door session a few weeks ago, and that was a topic that just everyone is very interested in, but nobody's figured it out.
Minna Song: Yeah. So we launched a maintenance app for technicians, so essentially it takes in information from the residents inquiries, and then it will create work orders, triage, and [00:30:00] then the maintenance app will actually send that to the right person. So it auto assigns schedules every work order by priority, technician skill, the duration of how long we are learning that task will take, location.
So if people are centralizing, it's really built because a lot of our customers want to centralize maintenance. So we'll take their GPS coordinates and actually send the right person with the right skillset to the job first, translate between different languages, between the resident's preferred language, the technician's preferred language, and so you are actually collecting way more information, particularly about, "Hey, does it take longer to fix this washer and dryer than it did than we expected it to?" Maybe these are more complicated issues, even though if you looked at a normal traditional work order system, they might look exactly the same.
So I do think that there's a lot of data we're collecting from this that can be used to think about things like... do we need to replace, do we need to invest a lot of capital in property renovations or and things like that. I [00:31:00] really think that wasn't possible without this level of data.
Brad Hargreaves: Yeah. It certainly seems like one of a handful of areas that feels, I don't wanna say obvious to me, because there's still some things that have to be figured out there, but if we're talking about, where is AI having an impact in property operations today in three years that it's not today, feels pretty high on my list.
Minna Song: Maintenance is an extremely exciting problem to solve. It's so complicated. There's so much opportunity. There's not that much being done in space because it's so complicated and some of these problems are solved in other industries, right? Uber knows how to optimize when a driver needs to get to a rider, right? Why can't we do that in multifamily, too?
Brad Hargreaves: I wanna pivot for a second and talk a little bit about fundraising. Elise AI raised 75 million back half of last year. That's pretty impressive. Growth rounds are hard to come by in the PropTech world right now.
There's not a lot getting done, so kudos to you for that. One thing I did want to just [00:32:00] get your sense of, you obviously raised back in, I assume you raised some capital back in 2017, 18, in the early days, correct?
Minna Song: Yeah, 2019. We were about two and a half years into the company before we raised.
Brad Hargreaves: Really. That's great. How much did you raise in 2019?
Minna Song: 1.9 million.
Brad Hargreaves: 1.9 million, 2019.
Minna Song: The seed rounds were very different than the seed rounds are today.
Brad Hargreaves: Yes. Which was easier, raising 1.9 million in 2019 before AI was cool or raising 75 million today when AI is hot, but PropTech is not. Which was easier?
Minna Song: Oh, certainly $75 million. The $1.9 was brutal. I think because people saw vertical markets as much smaller. It was just not that attractive. I think until people saw the value that AI could unlock, did PropTech look like a very big market.
So that was like number one is probably is this going to be a large enough opportunity to make it a venture-scale opportunity? And now it's quite obvious that it is. And of course, it depends on what type of product you're [00:33:00] building. Yeah, $75 was much easier.
Brad Hargreaves: Wow. This is not what I expected you to answer. Everyone talks about how hard growth rounds are now. Your metrics must be bonkers, that it was easier than the $1.9 million seed round in 2019.
Minna Song: We're working hard.
Brad Hargreaves: So tell me about this healthcare thing. So you launched Elise AI for healthcare last year, correct? Tell me about that.
Minna Song: Yeah, a lot of people say healthcare and housing seem so different. Is there any overlap? And really, there's a ton of overlap, actually. Yes, the customers are very different. But the technology is the same. You can think about our leasing product where we're qualifying inbound prospects inbound renters have questions. We need to collect information from them, and we need to schedule them for a tour. It's very similar to what we do on the healthcare side. Patients come in or they call into our into practice, and then we need to gather information, insurance name, we need to answer some questions, and then we need to schedule them for appointments.
So it very much feels like Elise AI or Meet Elise, as we used [00:34:00] to be called in 2017, 2018. And it's really exciting to be able to apply the technology to another problem that serves a fundamental human need.
It shares the same platform. So, as we make advancements in our healthcare products, it is making our housing products stronger and vice versa. Healthcare is primarily over phone calls, so that's why we had to build out our voice AI product. And of course, our customers on the housing side have a ton of calls that they need to handle, from leasing to maintenance, is a lot over voice, so we're excited that the two can benefit from each other.
Voice isn't a very hard technical problem to solve. It's been really exciting to see the traction there. We primarily focus in women's health and dermatology, so we are seeing that a lot of those practices are benefiting because they have a high volume of patients they need to handle. Similar to the housing side, you have a high volume of leads and they're missing calls, high abandonment rates. It [00:35:00] sounds very similar when you dig into the operations.
Brad Hargreaves: I love that. Two big needs. Before we wrap, I would love to move into the lightning round. I have a couple of quick hit questions and answers for you that I ask every one of our guests here on the Leader Series. Shall we jump into it?
Minna Song: Sure. Sounds good.
Brad Hargreaves: Awesome. So, first question. Tell us about one startup real estate developer or entrepreneur you're watching and why?
Minna Song: I have been very impressed with the Engrain team. We did a partnership with them recently. As people are centralizing, mapping properties becomes so important. It's a big loss if you don't have that person on-site and all that tribal knowledge. So I think that's a really exciting one in the time of centralization.
And then I'd say, maybe this isn't exactly what people think is PropTech, but I'd watch robotics. I think Tesla is gonna be very successful there, and Figure is a really big well-funded robotics company right now. As you think about the physical property and how do you service that? It's gonna be [00:36:00] robots.
Brad Hargreaves: Next question. When we are recording this podcast in the 2030s ,what is the most important real estate tech topic we'll be talking about and why?
Minna Song: Oh, am I gonna be invited back? Thank you.
Brad Hargreaves: Absolutely Minna.
Minna Song: Yeah, 2030s. We're gonna be talking about zoning laws.
Brad Hargreaves: I talk about zoning laws all the time anyway. This is one of the few shows where I've not talked about zoning laws and zoning laws have not come up.
Minna Song: It's not gonna change. I think robotics is gonna be a big one. I think in general, the world's gonna change a lot, right? Less work. There'll be more focus on life balance and community connections. I think density of communities or cities is going to increase.
People are gonna live longer, right? A lot of the AI academic research right now is happening in extending life or reversing aging. And if people live longer, that means they're here for longer, and probably they will tend to have more children, especially if they're not working as much partially for fulfillment, partially because it becomes cheaper, because generally AI will create more wealth and [00:37:00] actually that wi ll end up, creating even more density. So then I think people will actually spread out as well. So it'll be cheaper to deliver things from with autonomous vehicles. I think that the whole landscape of where people live is gonna be really interesting. And even though I said zoning laws as a joke, like it's not going to be a joke if all those things come true. And so I think it depends on if you're saying 2030 or the end of 2030. But I do think that a lot of the research in biotech is gonna change the population a lot. And I think amenities. I think the other big thing is amenities are gonna change.
I think the three technologies I really believed in were AI, and do believe in AI, CRISPR, which is like gene editing, and VR. And Apple's made some moves. I don't think Apple's going back, as many people believe that Apple Vision Pro is not gonna be a thing. I think it's gonna be even more of a thing soon. And you're gonna need VR rooms or the 360 treadmills. I think so much is gonna happen in [00:38:00] VR that I don't even know what gonna look like, but it's gonna change how people live too.
So I think all these things are really exciting. I don't think progress just happens magically. I think it takes a lot of sacrifice from groups of people to make these things happen. So it's a little unpredictable, but those are the areas I think I'm hopeful or I wish would have a lot of progress.
Brad Hargreaves: I love that. We could do another episode on Minna Song views on the future of technology. There's a lot there, and fascinating stuff too. The questions around longevity. I'm praying you're right. I wanna live forever. May not forever, but for long, longer than humans typically live.
Minna Song: There might be a line where everyone lives forever and stay healthy right now because you might be right on the..
Brad Hargreaves: Right on the cusp. You're a little younger than me, so you're you're probably further on that closer to that cusp than I am, but...
Minna Song: i'm training.
Brad Hargreaves: Hopefully my kids are under that cusp. That's what I'm hoping for. What's one city or place you would bet on?
Minna Song: Depends if it's short-term or long-term. I do think New York City is an incredible city.
I really believe in science, and technology [00:39:00] and those happen where there's high density of people. I really hope that New York can become denser, even if you can believe that or other cities become as dense as New York City, because I think that will actually accelerate the growth and innovation.
I think Chicago's gonna be really exciting. There's great universities there. I think there's a lot to be done. We just opened up in Chicago office which is a part of the reason I think there's not enough AI companies in there, but there's a ton of people who there who could be contributing to AI.
Brad Hargreaves: I love that. I hope Chicago figures it out. It's a great city. Last question.
Minna Song: A lot of the real estate companies in Chicago too.
Brad Hargreaves: There are some awesome operators and PropTech companies in Chicago too.
Minna Song: For sure.
Brad Hargreaves: Last question. What's your favorite app on your home screen?
Minna Song: Oh, I love our maintenance app. I work all of the time, and all I think about and breathe is work. It changes every single day right now. I probably spend the most time looking at the new features on our maintenance app, but it's our first app, it's actually quite exciting for us.
Brad Hargreaves: That's awesome. I love it. Minna, thank you so much for joining Thesis Driven Leader Series. [00:40:00] So great to have you on the show today.
Minna Song: Thanks so much, Brad.
Brad Hargreaves: Appreciate you all tuning in today and joining me for that conversation with Minna Song of Elise AI. You'll definitely wanna join us next week too, as we have one of the most powerful people in the housing world on the show, Sue Yannaccone, the CEO of Anywhere Brands.
Now, you may not have heard of anywhere, but they're the largest residential brokerage conglomerate in the world, and you've probably heard of some of the brands they own, like Better Homes and Gardens, Century 21, Coldwell Banker, Sotheby's, Corcoran, and more.
The residential market is going through some pretty serious transformations right now, and maybe you've heard about some of them: changes around buy-side agent compensation, clear cooperation potentially going away. We're gonna dig into that and really understand what's going on in the residential market right now, what is the impact of AI and technology?
So you will definitely want to listen in to next week's conversation with Sue Yannaccone. See you then.
Share this post