EliseAI founder Minna Song on automating property management with AI — from MIT CS to enterprise housing platform

are you doing? So sorry to keep you waiting. Sorry to keep you waiting on your big day. I was tuning in. I was learning stuff. The previous guest was talking about cars and we went down a whole rabbit hole. It happens. Um, but thank you so much for taking the time.

Would you mind kicking us off with an introduction on yourself and the company? Yeah, absolutely. Um, I'm Mina Song and I'm the co-found one of the co-founders and the CEO of a company called Elise AI. And Elise AI is built is a company in based in New York City.

We build AI agents that automate the entire customer journey for two verticals, the housing industry and the healthcare industry. And um, we've we just raised our series E today. So that's excited. Uh, yeah. How much? Yeah, we raised a 250 $250 million. There we go. Congratulations.

Can you give us uh can you give us a quick history of the company? We heard from a friend uh what uh he gave us a general read on your revenue ramp and it was pretty mind-blowing. So I I would love to kind kind of get the the quick history of the company since this is the first time overnight success. Yeah. Yes.

Just started yesterday. Vibe coded too.

say um no we uh yeah we've we've we started 2017 started with my co-founder Tony um who's the CTO we met in college studied um computer science both software engineering backgrounds um I went to MIT he went to University of Cambridge but we after college shortly after college we wanted to um build a company build technology for industries that serve fundamental human needs and that gave us this idea that we wanted to um we wanted to go into housing and housing is our largest expense as humans and we thought um it seemed like it had limited technology as an industry.

We didn't really know too much about it but felt like a good place for two technologists to start investigating. So I when we got started I took a job working at a real estate firm in Manhattan and um that was just sort of a research opportunity. I got to I worked at the front desk. I answered all the phone calls.

I greeted everyone.

answered all the emails and um was just really listening for what is the biggest bottleneck that exists that we can start chipping away at and it was very clear that was communication and so we were actually a conversational AI startup right from the very beginning versus I think a lot of the companies that you know are adding on AI right now um it's been pretty obvious if you've been doing it for a long time that there's a lot of value and um a couple years ago how long did you last at that job by the way your research your paid research opportunity.

It was like it was like a good three months. So, um it was it was quite fun. I enjoyed it a lot. That is a wild start. Uh how how important is it to the company to be seen as like an AI agent pure play?

Uh I like it feels like a lot of companies that are doing that are really really successful or really adding a lot of value are actually not just picking the hottest tools off the shelf but they also have a lot of traditional software and like SAS.

Um some are bolting on AI tooling but others are kind of offering elements of both whether it's traditional dashboarding or just uh traditional you know uh business logic and workflows like how has the business grown? What's the shape of it? What's demand like?

Um is it is it are our are customers receptive to something that is purely non-deterministic?

Yeah, I mean a big part of what we work on is to make sure that it's providing enterprisegrade reliability to our customers in housing and healthcare to you know highly regulated industries to just really consequential decisions for if we make mistakes it's a big deal for our customers right and for their customers their their consumers.

So um we we focus a lot on making sure it is um you know it is it is stable and it is reliable for for them. Yeah. Um but yeah it's a mix of I mean you have to you know AI is a is a really powerful tool.

It's a new type of computer really and you can solve problems with that but also traditional software and other things that you you need are are also important in solving a problem. I think what's sort of different about what we build, we're not a we're a vertical AI company, right? We're not a horizontal AI.

Horizontal being, you know, you can um solve one sliver of a problem for a bunch of different industries, but our approach is we we sort of go 10 layers deep into every problem and um that way we can actually solve that problem and end. So it's very different.

So let's say in in housing um you are um you are saying that you have you're a broken sync right?

It's very different when an AI might be able you know a large language model might be able to say hey thanks for letting me know I'll submit a ticket for you and maybe it does that for us um our AI is actually doing a bunch of things.

It's understanding how that resolution should uh how how that issue should be resolved.

it's coordinating all the the workforce and all the resources, the systems, the compliance to get that ultimate issue resolved or um there's things that we do that are not just pure sort of conversational AI pieces, but we work in the physical world.

So, we have to actually unlock these things called doors and allow people to go and get into the building. So, we have to do things like hardware integrations and um you you can't avoid doing that if you actually want to solve the whole problem. Yeah. On the integration side, uh what's working on the marketing side?

Is are there channel partners or platforms that you can plug into to accelerate growth? Is it just a ton of SDRs? Is there a viral component to this? Like what's accelerating the growth other than just the product delivering value? Yeah. Um, yeah, we work in particularly non-technical industries.

So, it's not really there's not a great um market for PE people that are just naturally searching for the latest. It's not like an app store that you can go viral on and then all of a sudden you're the number one enterprise app in in uh property management. Yes. I I wish it is blood, sweat, and tears.

We have outbound sales. Sure. Um but we do have a lot of partners. Um AI is actually a great vehicle for delivering a lot of the other products that exist in the industry.

So historically if you needed um if you let's say you were a marketing software for housing providers or you were um or you have some other tech uh you have insurance or something like that.

you needed to train a human agent, a leasing agent or a property manager to make sure that got that that marketing content got distributed. And that's another manual step that people had to have.

And I think a lot of people in the industry have realized, hey, I can just partner with Elise AI and my now my marketing materials or my my uh insurance will be, you know, will be um yeah, communicated to everybody else. Sense, Jordy. Uh nothing for me. This is very exciting. Congratulations on the fundraiser.

Thanks so much. Let us know when you add podcasting as a vertical. We'll happily uh happily try it out. We're definitely in the market. Uh well, have a great rest of your day. Congratulations on the raise. We will talk to you soon. Thanks for joining. Bye. Bye. Thanks.

Uh I wanted to finish on this post from Lucas Buyer now at Meta AI. He says, "Ring on the floor laughing. The solution it draws is amazing. " Uh this is from Tor Olman. Uh so the question for chat GPT for GPT5 is let's go back to this and it's a picture of what uh you can go to the next slide.

Uh is this the head of a rabbit or a duck? And the answer that chatt gives is this image shows the duck rabbit illusion. It can be seen in two ways as the head of a duck with the beak pointing to the left or the or the head of a rabbit the ears pointing to the left. So the answer is it is both a duck or or a rabbit.

And then uh and then it says, "Would you like me to mark the outlines for each interpretation directly on the image to make the two views clearer? " And it does. Ah, that's so clear now. And it draws it on there, which is a fascinating ability to even trace the outline.

This was unthinkable in machine learning just a few years ago. What does it see that we don't? Yes. Maybe it's found a deeper truth. It's possible it's found a deeper truth. No, this is this is this is hilarious because clearly it's been RLHF to know that people play pranks on LLMs all the time.

They hit them with riddles. They hit them with unsolvable questions and trick questions.

And that one of the things people demand of their AI chat bots is that they don't mess up when they're hit with a one of the classic examples is uh a boy is injured, goes to the hospital, the doctor says I can't I can't operate on this uh child. Uh, and then the question is like why?

And the answer should be because the doctor is the is the is the patient's father or something. But you can throw all these subtle changes where it clearly indicates who is who and it'll still think that it's being hit with the with the test with the trick question.

Well, one more post from Luke Metro since we haven't covered him today. There was a a a young man, very young man, a teenager joined SpaceX at 14. At 16, uh Chiran is joining Citadel Securities. Let's hear it. Luke says the hard techy actually money is cooler. We've seen it before.

If he's really good, he'll get poached by a lab within the next Maybe. Maybe. And then one final post from you can't pull Jordy Hayes out of the timeline. Uh Sophie Netcap Girl says, "August is the Sunday of months and so here's to hoping that we just have the Sunday scaries. Fun hanging.

" Oh yeah, September is going to come around and we'll be extremely back. But thank you for tuning in today. Hope you enjoyed the guests. Hope you enjoyed the show. We will see you tomorrow. Leave us five stars on Apple Podcast and Spotify. Subscribe to our Substack VPN.