What Is an AI Factory? With Cat Allday

All right, welcome back to Owner Occupied. I'm really excited today to introduce Kat Allday, head of AI at AppFolio. Uh, I know this is going to be just a really exciting conversation, and it's, it's very timely with all the discussion around AI, both in our industry and, and in, you know, the, in the universe as a whole. Um, so Kat, thanks again for coming on, and I'd love to have you start by just giving like 90 seconds on your, your background at AppFolio and where you were prior to that, and we can kind of go from there.

Great. Well, thank you for having me. I'm really looking forward to our conversation today. I'll tell you a little bit about myself. So I've been in technology for about 25 years. I've worked at several software companies. I've been at AppFolio for 8 and a half years, and I've worked both with the property management side of the business, but also in a previous incarnation, I worked with the legal vertical for a few years before that business was spun out. And I've been focusing on AI for about 3 and a half years now. We started really like a concerted effort in 2019, and I moved over in 2020, January 2020 to be exact., to, to really focus on what we're doing around this technology, um, for our customers.

Okay, great. And I, I was so fascinated when we were kind of chatting, uh, prior that first of all, that you've been at AppFolio for so long. So you, you've been there, you said 8 and a half years. I mean, that was long before they went public, so you've really seen the rise of that company as they've come to, to dominate our space, which is small to medium-sized property management companies. Definitely the largest market penetration there. Before we dive too far into AI, I'd love to hear you just maybe reflect a little bit about, a little bit on Appfolio as a company, working there, their position in the marketplace, and some of what, because you've had a really unique position having been there for so long and been able to observe their impact on the industry.

Yeah, so, you know, I actually joined AppFolio pre-IPO, so about 6 months actually, right before we went public. And the reason why I joined the company, you know, I've, like I said, I've been in technology for many years and worked at several different companies, was their approach to the way they think about solving problems. And they, you know, the ethos about listening to our customers is in our DNA is a really something that really stood, you know, struck, stood out to me when I was evaluating Appfolio as a company I wanted to join. And also, I think we had founders who were technologists at heart, but again, were sort of thinking about the future and kind of able to look around corners at trends in technology, but then also relate that to the industries that they were supporting. So, if you think about property management, What are things that are happening in the property management space and how can we use technology to help solve those and be understanding what those trends are before they actually happen so that we're ready to take advantage of those? So I think that is something that I really found fascinating about the company and where it was at the time I joined. And it's really proven to be true. You know, we are very close to our customers. We understand the, the pain, the challenges that they're having, and then being able to solve those problems for them, sometimes preemptively, uh, I think has been something that's been really unique about this organization.

Yeah, that, that rings true. I, I feel like AppFolio, you know, I, you know, I'm an outsider, but just the fact that they had a head of AI for would you say, 2 and a half years, 3 and a half years, um, speaks to the fact that they are looking ahead, look, trying to look around the corner of what's coming down the pike in our industry. I certainly don't know of any other property management software provider that has anyone as a head of AI, let alone going back long before ChatGPT was a household word and everything else. So very interesting, and I I'm always curious about big public companies, like how much of that original founder ethos or culture is still— you walk the halls of the HQ and do you still feel that? Is there still some of that DNA, as you mentioned, in the company? And it sounds like maybe that really is the case.

Yeah, I think one of the things that we do have in our offices here is we have our core values. On the walls. So as you walk around the building, you see those values. Those values were sort of defined by the founders, like in their probably the first year that the company was around, and they still hold very true. And I think that is something that, you know, a company that's— we're 16, 17 years old now— that is actually pretty significant to see, you know, this core set of values and people still sort of practice those values. It's one thing to have values on the walls, but to actually have the employees living those values, I think is really special. And from a technology standpoint, I think the things that the founders were able to do was really understand these big technology waves. We like to call them waves. If you think about they built the product from day one to be hosted in the cloud. I think that was a very deliberate decision. So they really took advantage of the internet and cloud as the foundation of the product. And then from there, the next sort of big wave was mobile. And they were thinking about, how do we now allow people to work from anywhere, through a mobile device, whether that's an iPad, a laptop, a phone, right? And the phone was really the thing that I think set, um, Appfolio apart. They were, you know, we were really able to get, um, you know, property managers to be able to do their work from that thing in their pocket that they carry with them everywhere. And, and I think AI is now that next wave, um, that we want to take advantage of.

Yeah, great. Perfect segue. So AI, that's really what we're here to talk about. Um, that's what we're all curious about and Um, we're going to get to the big announcement, uh, that dropped a few weeks ago, uh, regarding AppFolio and, and AI. But I'd love to hear from you personally what fascinated you about AI all those years ago— not all those years ago, but a number of years ago when you took this role within the company. There was no guarantee that AI was going to be hot and big and the next buzz and trend. So this was a strategic bet on your part, slash passion, you know, I don't know how those, you know, how you would allocate it there. But something made you decide, yeah, you know what, I'm gonna go within a property management specific company, I'm going to focus on AI, which, you know, like, now sounds amazing. But go back 3 years, I'm sure you got some confused looks, like, so I'm just so interested to know, What did you see, or what within you was like, yes, this is where I want to go, this is the future, this is so cool?

Yeah. So one of the things, we had a CTO, uh, Chief Technology Officer, who painted this picture of what property management could look like in 10, 15 years. And it was such a compelling story, uh, that I said I want to be part of that, something that could really change an industry. And some of it is probably a little bit science fiction still, but it was fascinating enough that it made me want to be a part of it.

Can you share any tidbits? I'll give you a couple of examples.

OK, great. So one of the things was, let's say you are turning a unit. And you could have this completely automated experience where when the unit is getting ready to turn, it automatically sends, um, you know, requests out to various different vendors, to the painter, to, um, a gig worker who was going to go do the inspection, let's say. So let's say somebody shows up, they go do an inspection, they have their mobile phone, they're using— they're recording with video, walking around the unit, capturing images of the state of the unit, and then that gets uploaded into a system. And there could be some kind of recognition between the previous inspection and the current inspection. And through, you know, imaging analysis, you could determine where— if there was damage. And then if there were— if there was damage, it could then say, okay, well, that's damage on the floor, it's a laminate floor, so we need to send somebody to repair that, and could initiate, you know, sending out notifications to whoever your, your person is who does those repairs. So let's say it could help you get your turns done. Then you could have automation around like a move-in flow or the leasing flows for people to do self-showings, let's say, and automate that process with, you know, smart lockboxes and other things. And so he told the story of sort of this automate— all these automated points in a unit's lifecycle, right? From move out to move in, through inspection, through the leasing process. And it was really fascinating to me. And I was like, well, how much of that is possible today? And like I said, a lot of it could still be science fiction, but you know, I think what we're seeing is we're getting, the technology is going to enable a lot of that vision really soon with sort of a combination of things, whether that's AI, traditional software automation, IoT, these sort of smart devices, those types of things. If you build a solution that uses all those components, I think you can get— we're going to get closer to that vision. I don't think it's 15 years. I think it might be 10 years.

Yeah. Yeah, I think you're right. And that is so fascinating that that future that you painted a picture of in relaying that story. There's a lot of things that I do day to day within a property management company where I just think to myself, this should be automated or this should be handled by an AI or— Yeah, I mean, it's frustrating because it's in like 90% of situations within property management. It's the same old, same old, but it's always those 10% edge cases that you get tripped up and you're like, "Well, this doesn't talk to this," or "What if it's a month-to-month lease?" or "What if the owner's moving in after the tenant moves out?" There's always weird scenarios that you have to account for that seem to make progress frustratingly slow with technology and automation and AI. But I think AI could help be the unlock that gets us past some of those hurdles. Would you agree?

I agree. And I think what we're seeing with these large language models, that's what powers like ChatGPT. I think that's going to actually allow us to, when those edge cases do happen, that technology is going to be able to help you solve those edge cases maybe more elegantly than you could today. Because these tools will have the capability to do some reasoning, which is different than the software we have today where it's very rules-based. Like if this happens, then do this. Well, sometimes it's not a linear step. There has to be some additional reasoning. And I think that's where the recent evolution of this technology with things like ChatGPT is very promising to be able to solve those sort of edge cases.

Yeah, totally agree. All right. So not everyone reads my newsletter and knows about the big AppFolio announcement. Maybe they heard about it some other way, but for those who aren't news junkies or follow, you know, industry trends super tightly, AppFolio made a big announcement at the recent NAA conference. And announced a product called Realm, which from what I understand is sort of a roll-up of some existing AI-type products as well as a brand new product, which is a conversational AI tool that allows you to actually do work within the product. So I don't want this to be like a long ad for AppFolio or anything, but this is interesting, right? I'm interested in this. So I'd love to hear you talk a little bit about what's Realm, what's in Realm, and, and share some of the conversational AI that, that you actually generously demoed recently, and it was really exciting to me. So love to hear your take on it and maybe just summarize for those who haven't been following.

Great. Well, I think our, our marketing and brand team do an exceptional job of helping us think about how we deliver our products. And so AppFolio Realm is sort of a umbrella term that encapsulates all of our AI products. So we've got lots of products that are already powered by AI. So we have Lisa, which is our digital leasing assistant, which was sort of our first product that we launched that uses AI. We have Smart Maintenance, which is a maintenance coordinator, digital coordinator basically. We have Smart Bill Entry, which helps with invoice processing, and that uses like image rendering technology so it can actually understand what's on the invoice. We have Smart Insure, which is for liability insurance coverage. We have— we recently announced marketing descriptions using tools like ChatGPT to sort of power them. And those products are all part of Realm. And then we announced— the big announcement at NAA was this new conversational interface. So being able to, you know, with— through like a chat window, ask Appfolio property manager to do something for you, whether that is, um, hey, I want to know what my rent roll was at this property. And basically what it would be able to do is goes and like executes the report query and then brings back the information for you. That's one example. You could also ask it to send some broad communication to your residents. So the example that I think we were using, let's say there's maintenance being done on an elevator in a building. So you could send a note, an SMS message to everybody in that building notifying them of that elevator maintenance. And that can be done in like one command. And that's really streamlining that work that you have to do today, which you'd have to, you know, go do that individually for every resident if you wanted to send like that bulk communication. So those are just some examples of how that tool could be used.

Yeah. So when I first saw this demo, A few different things really struck me, um, and I'm going to monologue here for a second because this is why I really wanted to chat with you. So I'm curious to get your take on this. So when, when I first— we got on this demo and what I was expecting was, um, oh yeah, AppFolio AI, you can like ask it questions about how to use AppFolio. Like, how do I enter a late fee? And a little bot would come back and say, to enter a late fee in AppFolio, first click settings, you know, kind of like a glorified, you know, help chatbot or something like that. That's what I was expecting. So then during the demo, they first, they showed that and I was like, okay, whatever, you know, obviously that's going to work right. And then it was like they showed, um, and I forget, you had, you know, there was an engineer on with us who was doing like a live demo. He, he queried, um, he, he said something like, I'd love to see a rent roll, just like you were explaining. I'd love to see a rent roll for this property. And then it showed up right in the chat window. It wasn't like, to see a rent roll for this property, click here and then go there. No, it like actually pulled the data right from that specific software, like, um, like my data, right, if I was the, the user in this example, and displays it right there. And that— and I was like, whoa, that's really interesting, because you've, you've given the, the LLM access to property, unit, resident data, Specific to my account, not just like the AppFolio help docs that are generic, right? So that was super cool, really interesting, and takes it to a way more useful level. And then to cap it off, he did, as again you articulated, like add a late fee to 123 Main Street, $50, January 1st. And it was like, you know, sort of reproduced back what it thought the request was and then press yes to confirm or something. And then it would post the late fee for you. I might be getting some of those details wrong, but so now you're talking about like not just, not just conversation that's one way where you're, you're sort of read-only, but actually the ability to update data in the database, in my database for my properties and execute work on my behalf. Like that to me was mind-blowing and maybe I'm silly to be blown away by that, but I have not seen anything like this anywhere else. Think about when you're chatting with ChatGPT. I mean, you can't ask ChatGPT to add a calendar event to your calendar because it just, it's not connected in that way. So this is just really next level. And where I think this is valuable is when you talk about new team members coming online, you know, they have to get familiar with software to be productive. But this is going to allow someone to be productive almost right off the bat because they're able to just converse with the software. To either figure out how to do what they need to do or to actually do it.

Yeah, I think that's definitely something we've heard from lots of people that have seen this is this is such a game changer for onboarding a new hire. And I think, you know, when we were thinking about the types of problems that people might use it for, there is this sort of like, show me how, right? I think that's an obvious one. And then there's this sort of like, I need to— I want to discover. So I want to find out information that's in my database. And that's the idea of like, go look up this information and tell me. I don't know how many times you've gone and you need like, you go run a report, and then you get the report, and it's got 20 columns. And the data you really need is in like column 12. And like, now I have to like, sort through that information. And if I could just get the, the data I want at my fingertips, that's so much more powerful. Um, and you know, it saves you time.

Um, yeah, I think about this like the difference between having Google Calendar and having an executive assistant. Yeah, right. Like, yeah, we have Google Calendar. I could go and add an event for next Friday to get my haircut, or I could just, hey, can you put an appointment on my calendar next Friday, haircut, and they're just gonna go do it, right? Um, it's kind of like that next level of usefulness and, and contextual awareness.

Yeah, I, I think, I think you're, you're spot on with being able to tell the application to do something for you. Um, that's a pretty unique sort of use of this type of technology, and I think where we have an advantage is, you know, we're leveraging this, the LLMs who have, there's just lots and lots of data, right, that allows us to, allows the system to understand natural language. So, you know, you might ask a question differently than I would ask the question, and the technology's got to understand that we're asking for the same thing. And I think that's where leveraging LLMs helps us do that. So we, you, you build this sort of like shared understanding regardless of what the user asks for or how they ask it. And then because we're able to tap into the data for the property management company, the information that's going to return is going to be very, very relevant. So I think that is really, I think, a very unique delivery of this technology. But I do think it's like super— I think about it as supercharging you. As an, as a, as an operator, like, um, you know, I can, I can just do things so much faster with this type of tool.

So the conversational AI is not out yet as of this recording. It's July 5th. Um, this recording probably be at least a few weeks before it gets released. Uh, I want to ask about pricing. So is this going to be included in the base AppFolio plan? Is this an add-on? Is this part of AppFolio Plus? Have any of those details been ironed out?

The way we're thinking about this today is that this is something we would give to all of our customers when it's ready for general use. Right now, we're not thinking about gating it behind the Plus tier or anything like that because we believe it's really powerful, and it is something that we think everybody should be able to take advantage of. So at this point, Like there's no decision about gating the conversational component. Now in the future, there may be some activities or things that make more sense for a certain type of property or user. And so we may restrict some things, but we haven't made any of those decisions yet.

Yeah. Okay. And timing, when can we look forward to this being generally available?

So that's a good question. There is definitely some work we want to do. One, we want to make sure that it is performant. So if you ask it a question, it doesn't take 5 minutes to return the response. So there's definitely some technology work we have to do to make sure it works as people would expect it to work. We will be bringing on a few, sort of what we call our charter customers. Very soon, and we plan to have anywhere from 5 to 15 charter customers. And those are folks who are really going to pound on this and let us— help us identify where the pain or the lagginess or if there's things that are returning responses that we wouldn't expect. We want to make sure we solve all those sort of technical challenges. I don't have an answer as to when, but like I said, we want to get it into some real customers' hands very soon. You know, we have our customer conference, our user group conference in September. We plan to be able to show sort of the next iteration of this product at that conference. So that's our next big milestone. And I think from there, we'll decide whether we could go into a more broad beta with the product.

Got it. Okay. And it seems like this product, the conversational AI, is something that it's like the more people who are using it, the better it's going to get. Because I assume there's going to be some sort of a feedback mechanism where the system's going to know if it was helpful or not, either by observing the interaction, or maybe there'll be a little thumbs up, thumbs down, was this helpful type thing. Is that right?

Yeah, we can sort of instrument the product so that we can understand what people are doing, the types of questions they're asking it, the things that they're wanting it to do. I think with any new technology like this, there is an adoption curve, right? And one of the things that I'm really— if I had to be honest, like the things that worry me, keep me up at night, are if we give this to customers and they use it once and it doesn't return, doesn't behave the way they thought it should behave, do they abandon it, right? And I think that is a challenge with some of this new technology, right? If it doesn't work exactly the way somebody expects, they're like, eh, it's no good. And these things will get smarter and better over time. And so we do have to look for those early adopters, the people who like, kind of playing with the technology to get it to a point where, you know, people who maybe are less likely to adopt new technology will be willing to use it because they'll be able to get the value out of it. So there is this sort of adoption curve we have to kind of think about as we're building a tool like this.

Yeah, it'll be fascinating to see what types of use cases come out of once this is available in the hands of property managers. Like, What does it turn out this is most useful for? You know, we talked about onboarding new employees. That's kind of an obvious one. But, you know, what I'm trying to think about, what other tasks property managers do on a day-to-day basis where this is faster and easier than clicking around the system a whole bunch? Because I feel like if you can identify those, optimize for them, and then sort of nudge new users like, hey, try asking Appfolio to blah, blah, blah, blah, blah, you know, that might help sort of, um, give them a great first experience?

Yeah, I think, I think being able to provide people with some guidance on how to use a tool like this, um, I don't know, you know, how many of your listeners have tried ChatGPT, but you got to kind of have— there's a, there's a little bit of an art to how you engage with it, right? And, um, you know, we want to make sure that it's like a no-brainer. Like, I can ask this thing to do X, Y, and Z for me, and it will do it without having to, like, learn how to ask it. And I think that's some of the things you're hearing about some of these tools like ChatGPT. You got to figure out how to— like, you might ask it to do something, and then you got to refine the ask a couple of times before it gets what you want it to do. I think that is something we're thinking about as we're designing these products is how can we help the user understand what the thing can do for it. And I think that is, you know, that's a very hard thing to do from a user experience or design perspective. It takes really— it takes people who are way smarter than me to figure out.

Yeah, there's, there's whole— I mean, prompt engineering, I think, is the lingo that folks use for this. And I think the advantage here is that because the tool is built into the product and it's customized for a specific use case, it's not like you're going to have to be like, you are a property manager assistant, right? You won't have to be typing that because that'll already be ingrained into it. So you can just start right away with your questions.

As long as you're asking it to do things that you could do in the app, it should be smart enough to figure those, those actions out. And so, yeah, I agree. I'm very eager to get this into some real customers' hands because we are going to learn a tremendous amount very fast once we do.

Excellent. I've heard a lot of good things about smart bill entry and smart maintenance. I think those two specifically are the ones I hear. I'm curious, like, if you're able to take what you learned in the development of the conversational AI and sort of bring that to some of those other products, or maybe vice versa. Did any of the learnings in those products inform the direction you went with the conversational AI? Because again, AppFolio, you— AppFolio as a company has been at this for a while. This isn't your first go at an AI product. And the, the, like, um, you have such a lead on the, on the competition, to my knowledge, in terms of being first to market here with a really compelling use case. I'm just so interested in, in that as a, um, competitive advantage. And like how, how that has all like played out internally between products and, and for a future roadmap?

You know, one of the things that I think has been really exciting is because we had this really concerted effort around AI in our organization, we have built an infrastructure that allows us to swiftly build functionality. And so we have a team called the AI Factory Oh, I like this. I mean, that basically has built the platform that allows us to build these tools. And we made an investment in this team. And that team has been involved in every AI product we've built. So they're either consulting or being a team member building a product. And so what we've been able to do is take all the learnings from all the products we've built and build this platform that is now available to more of our engineering organization to start thinking about, okay, how do I harness the power of AI in solving a customer problem that, you know, I'm working on in this maybe different area of the product? And so the AI factory team does a couple of things. They built the infrastructure and the platform, And they've also done education. So not every engineer knows how to build an AI product. Software engineers have different specialties, right? And so we've built these, what we call our AI Academy courses for our internal teams to help educate people, um, on like what problems are even worth putting AI towards. Because, you know, uh, AI is not— it's, you know, it's It's not something you want to use for every, every problem. It's not always the most economical or the fastest way to solve a problem. And so helping people understand like what kind of problem is, is AI good for, that's like the sort of the baseline education and then education on the more technical, like actually how to develop AI products or how to think about the models. And so that's been an investment we've made as an organization because we really believe this technology is going to be transformative in the industry. And so we've got this sort of— I think that's really— if I'm spilling the beans, that's our competitive advantage.

You know, that is fascinating. So a platform internally within AppFolio to help engineers utilize AI or explain why they shouldn't, right? Yeah. To help solve customer problems. Very interesting. So just to get some of your expertise, because I'm curious personally, what types of problems are useful? What types of problems is AI useful for and what types of problems is it not useful for?

Yeah. So I think where AI is, is helpful is if you're doing these sort of— if the activities are very routine and you can do the same thing every single time, that's a great place where you can use AI to help automate., right? That's one example. I think the other example is when you're trying to look at a lot of data and pull insights out of that data. That's another place where AI is really powerful. And that's, you know, whether it's pulling insights or being able to make recommendation, like a recommendation engine, those types of, you know, solutions are I think where AI tends to excel. I think what we're seeing is with ChatGPT and large language models is this, the natural language processing. So anywhere where you're engaging around asking questions or being able to, I think maintenance is a good exa— I think maintenance is a really interesting example. If you think about when a resident is reporting an issue, if they're in the tenant portal reporting an issue and they're able to say, hey, I have a leak in my kitchen, okay, then the AI can sort of prompt the resident for some more information. So because it's a leak, is it the faucet? Is it a pipe under the sink? Like, where is it? And so they can, like answer those questions. And so when that maintenance request gets sent to the property manager, they already have some of the triage questions already, you know, defined versus, you know, just an open text field where they can describe something. And if you're like me, I'm not very good at, you know, doing home projects. My description, if somebody were asking me, it'd be like, There's water on the kitchen floor. That would be my, like, description.

That's the extent of my knowledge.

A bunch of other residents also probably report things that way. So think about, like, how much time that can save the property management team, the maintenance team, because they already know, hey, the problem is under the sink. It's the, it's the pipe. And so I'm going to send the right guy out there who can get under there and resolve that, right? Versus, like, showing up and and hoping that they gave you the right information. Or also, I think from an urgency, you know, being able to help identify if something's really urgent, you know, I think that's where that kind of technology can help.

I love the idea of AI being used behind the scenes in like a silent way to be helpful in solving customer problems and not like the conversational AI is You know, it's big media. There's great PR there. It's very exciting. It's this whole new interface. But, you know, what I'm hearing you articulate and what I also think is true is some of the best use cases for AI are quiet. They're just silently doing their thing, helping solve a customer problem, helping a property manager get something done, and not being, not being flashy or necessarily even exciting, but just small quality of life improvements that help streamline workflows or automate communication or different things like that?

Yeah, you know, that is so one of our founder, one of the ways he described AI when, when I was talking to him about this, this, this role before I joined, he said AI is like raisin bread. It's traditional software development with raisins sprinkled in and AI is the raisins. Right. And so it's like AI isn't everything. It's just this— it's the sort of supporter, right? So, so I thought it was a great analogy and I think about it often. You know, it's the, the sprinkling of raisins throughout the product. It's kind of a— if Appolia property manager were a loaf of raisin bread, we want AI throughout.

Yep. Okay, great. I love that. So I put out the call on Twitter a few days ago for questions. And so I picked a select few here that I thought were really good questions to maybe prompt some good conversation here. So my friend Wolfgang Kroski, he was curious about AI, and we kind of started going down this road a minute ago, but AI for data analytics. So property managers have tons of data within essentially the database that is the property management software about units, tenants, work orders, vendors, clients, bank, you know, bank transactions. So, you know, I think there's probably insights buried in that data that could be surfaced with intelligent data analytics or AI, whatever you want to call that. Is that something that AppFolio is working on? Is there anything there that you find interesting?

Yeah, so I think some of the things that we're thinking about, like in the maintenance space, we do have a maintenance dashboard, I think in the Plus product, where we are bubbling up some insights. I think, again, when you have a large amount of data, that is a great application for AI. I think we'll see more of that in the future. But I don't, I'm not sure if we have anything like that that I can share right now.

Okay, no problem. All right. Um, so we've got some funny names here coming up from people on Twitter because some people don't use their real name. So next one is the Grumpy Landlord. Uh, I, I empathize with that. Um, he, uh, one of the questions he said is, is what's on the AI roadmap? And, you know, maybe you can share some of this, maybe you can't, but, um, Anything that you can talk about coming down the pike, I'm sure would be exciting for folks to hear about.

Yeah, I think what I can share is, you know, we are working on this conversational interface. I think the other things that you'll see is we'll continue to see developments in our existing products, so they'll continue to enhance those. Smart bill entry is gonna continue to get smarter. I think over the last 6 months, it's able to recognize more and more fields that I think then, I think it initially when we first released it, it could do 4 fields. I think it's up to like 9 fields now that it can auto recognize.

So this is like vendor, date, amount due, invoice number. Okay. Like, yeah.

Interesting. Yeah. And so that will continue to evolve. You know, we're working on a product called Rent Insights, which is, you know, looking at a number of pieces of data around both comps and amenities and helping you understand how your pricing or your, you know, how you think about pricing those units would change based on like moving some, some of those criteria around. So that's actually a— and it helps you with understanding your vacancies. Uh, so that's definitely using like the more of that data analysis.

I love that. I think there's big opportunities here. I think what the big property managers, or maybe it's the hotel operators, what they have this great euphemism for basically— I think they call it revenue management or something. It's basically like, what's the most you can rent this for? It's a really nice way of saying that. What's the most you can charge for this and still get it leased? They call it revenue management. Sounds very— we're trying to raise your rent as much as we can is really what it is. And that's something that's really, I think, a pain point within single-family rentals and small multifamily is pricing the unit. Now sometimes it's easy. You're in the middle of a big built-to-rent development and there's 100, you know, all over the place that are all priced right around the same amount, and you can optimize your vacancy rate pretty carefully, almost like a big multifamily property. But if you've just got a Class B, 1980-built single-family home in Houston, a suburb of Houston, where do you set the rent there? And yeah, you can go and do a bunch of work, and if you're experienced, you may have some ideas about where that should be. But the ability to have an AI help there I think would be really great and I think is a great use case for that.

Yeah, I think almost on top of that, or, or tangential to that, is, you know, what if you wanted to make property upgrades, um, and you're trying to decide like, hey, if I made this upgrade, how would it affect, you know, what I—

my—

the market rate I'd be able to charge for this, this this property, that might help you in making some of those decisions about those sort of capital expenses you might want to make in order to drive up your, your, your revenue there. Or, you know, does it make sense to replace an appliance or to repair it? Like being able to look at some of that data and, and have it say, hey, it's probably better just to repair it than to replace it. With where you're going to be able to charge the, you know, increase the rent or, you know, the other thing is, is like, does it make sense to renew a lease at a certain rate? You know, or, you know, do you want to consider like if you had, if that resident left and you had that unit sitting vacant for, you know, 6 months, you'd still be fine being able to understand what those inputs are that drive that, those, your occupancy rate or your vacancy rate.

Please send help in this area. We need help so badly as an industry. Everything you just described, and I think that latter may fall under the revenue management piece, like what should you send for a lease renewal amount? What's the impact of a vacancy? How long does it likely take to re-rent? What's the price you could get on a re-rent versus renewal? And you can even get really fancy with like different pricing for 1-year, 2-year, 3-year leases or weird random month lengths. Sometimes you see, if you've ever been on your local city's subreddit, you'll sometimes see people complaining about rent increases. Like, "My landlord sent me this weird rent increase. It's like $995 for a 14-month lease and $1,295 for a 19-month lease." And I'm always like, "Yep, there's some software back behind that trying to optimize that revenue." And we just don't have any of that, to my knowledge, in our, in our single-family, you know, small multifamily world. And I think we're leaving a lot of our, of our clients' money on the table, meaning property owners, by not having any sophistication in this area. Not because we're stupid, but just because we don't have the tools that these big multifamily guys have.

Yeah, I think this is where, you know, AI could definitely help.

Make it accessible to more, to more folks.

Yeah.

And everything you've been talking about, I think just is going to help elevate the conversations that we can have with our clients. Getting back to like the recommend recommendations on appliances, whether you should repair or replace, or even like, should we upgrade this bathroom? What would be the impact on rent if we did that? Right. This is how we can go to our property owners and position ourselves as experts and as partners rather than just a bunch of people who collect the rent and call me when something breaks at my house. That's a big deal, I think, for our industry. We got another interesting question from Twitter with a silly name. Dumb Real Estate Guy says, Do they plan to acquire any solutions as they grow this platform?

I think that's the question. I actually hear that question internally as well. You know, I think if the right opportunity arose with the right product or solution, we would consider it. But, you know, you know, further, nothing, nothing imminent that I could, could share.

Yeah. AppFolio is not afraid to make acquisitions. I think You all— AppFolio has made a number of acquisitions in our industry in the past. That was kind of before I started really closely following AppFolio, but—

I think the other thing that— I think another place where we'll leverage other vendors out there is through our Stack Marketplace. I think there is an opportunity for us to think about how do we expose these technologies, you know, across our partner ecosystem. I think that there could be some opportunity there, so we could accelerate some of this development through our partners.

Yeah, I'd love to hear a little more about that. We haven't talked about the Stack Marketplace, but that's, that's been a really big deal for AppFolio and for our industry. Um, yeah, are you, are you envisioning like some sort of a, um, Realm API where stack partners could query, or, you know, we're getting a little above maybe my pay grade, but yeah.

Yeah, so I mean, we definitely have talked about like, what could that future look like? Is there a way for, you know, some data sharing to enable, you know, powering these tools? I don't know that we've come up with exactly how we would do that yet, or what the constraint might need to be, but definitely something that when we've been brainstorming and talking around a table, it's definitely come up.

Yeah, love it. Mithril Property Management had a great question. Mobile app voice commands possible? And he gives an example. He had a question mark and he gave an example. Send a violation notice to ABC Street and issue the fine on their ledger. Sounds like maybe he's asking with the AppFolio app, you know, will this conversational AI be available there?

Yeah, so at NAA, I demoed the conversational UI from my phone. Oh, cool. And I didn't use a keyboard. I talked. I spoke to it the whole time just using the microphone on my phone. So yes, I am a huge— like, myself as a user of AI technology, I hate typing on my mobile phone. I was a BlackBerry user for years, and I did not want to give up my keyboard. And so to this day, I love my iPhone, don't get me wrong. And I would never want a keyboard again on my phone. But I found that I've become a really slow typer on that little keyboard on the iPhone.

And I like— I feel like that keyboard's gotten worse. It's gotten smaller. I feel like it used to be better.

And I hit the wrong letters all the time. And so I pretty much always use voice dictation. I send all my text messages through voice dictation. I'm that annoying person in the grocery store talking into my phone. Yep. That's awesome. So yes, that, that will be a, I think that's the benefit of, of having, you know, a mobile device is that you have those, you know, that tool. And so we would definitely want to piggyback on that with anything we're building.

So was this like at the booth at NAA? You were on your phone because there was, was there a big presentation on stage or was this just kind of at the booth?

It was at the booth. We were demo— we had demo stations where we were able to demo it on iPads. I had my iPhone and then also workstations. And we had some things that we were showcasing, but people were also just coming up and typing in questions. And I think my favorite thing that happened at the event was we had I don't know if he was a customer or prospect. I think he was a prospect. I don't think he was an AppFolio user. He was there and he was spending a lot of time talking with the team. And then maybe 2 hours later, he came back with a bunch of his colleagues and he was the one demoing to them. Oh, that's what you want to see. Which is great because it means that it's— it made sense to him. It was easy to use and he could start thinking about problems that this technology could help him with. And so that was really— it was really exciting to see as a technologist who's building something, you want to see people get value out of it.

So that was a highlight for me. Yeah. Any other interesting feedback from customers who were testing this out or just learning about it that you could share?

Yeah, I think the thing that I heard was something you mentioned right in the beginning was this could really accelerate new hires getting using the product and this like not requiring, you know, a lot of training. That was definitely things that resonated with people. They kind of, they recognized that pretty quickly. You know, I think there are certain things that don't make sense to use this technology for. And I think that's also really insightful. You know, I'd have people who would say, well, it takes me like 2 minutes to like do this task, so I don't think I would use this for that. And that's, that's also really good feedback, you know, because maybe we don't need to spend time perfecting that particular workflow or case, you know.

This is reminding me a little bit of like in the finance world, there's kind of this thing where if you're, if you're really good at Excel, you don't use the mouse. You like do everything from the keyboard. Right. And like people really pride themselves on like, I never touch the keyboard when I'm using Excel. I'm not one of those people, by the way, I'm using the mouse all the time. And to even further go backward in time, you know, I grew up with DOS, right? And you're typing, you know, commands like to see the files in the directory, like dir and cd space and cd backslash to like navigate through folders and directories. Um, and there was times when I first started using Windows that I was like, this was way faster in DOS. Like I knew the command, I know how to like pull up and edit a text file really quick. And now I'm like clicking around and this feels really slow and clunky. And in some weird way, a conversational AI is like almost getting back to that command prompt where if you know the system and you know the bare minimum that you can say to get it to do what you want. I think there's no question it's going to be faster than clicking around and waiting for pages to load. And like, like you said, you're pulling up some report and like the, the numbers there, it's in column GG2 or something, and you're like scrolling sideways. And so I could, I could see this being beneficial not only to new users but potentially also power users.

Yeah, well, and I think the interesting thing about this technology is that it understands natural language. So back in the DOS days, you had to know the command, right?

Yeah.

The backside of the horse or the front side, right? And so with this technology, you can just talk to it like you would talk to anybody. And I think that's really where this is exciting and unique because you could be a power user or you could be a novice user and you can get the same results. And I think that is— it sort of democratizes the technology a little bit.

To all the user types.

Super.

Let's see. So what's got you excited about the future here? I know we've covered a lot of different topics here, and I love the concept of sort of rebranding all the existing AI products under the realm so that sort of like people can have a mental model of everything that's included there. I know obviously you're excited about the conversational AI piece. Anything else that, that you want to talk about regarding, you know, your work at AppFolio in general or AI or any, anything we've discussed?

So I think we're just scratching the surface with what this technology can do. You know, one of the things I think we talked about with conversational AI is, you know, being able to discover, being able to execute, you know, tasks in, in the application. But I think the future is actually going to be even more powerful and being able to allow the system to do work for you without you even needing to ask it to do it. I think that's where the future is. And I think you described this really well in the very, very beginning. There are just some things that you have to do all the time that wouldn't it be just great if the computer just did it for you? I think that that is possible. There's obviously work to do to get there, but if you can think about, there are going to be certain activities that the property manager should never have to do. The application should be smart enough. It should learn the things that you do all the time, these sort of routine behaviors, and it can say, hey, I noticed that the first of the month you run these 10 reports. Do you want me to do that for you and then summarize the insights and send you a note? Um, like, that's where I think the future is, is going, um, or, or could be going if, if we, if we want to harness that.

Yeah, or I'm picturing, like, um, for a new client, like, you bring on a new property owner, um, I feel like there's a lot there where the AI could be like, hey, looks like it's been 4 weeks Since you onboarded this owner, the property's still not leased. Just wanted to bring that to your attention. Do you want me to email the owner? Do you want me to review the listing with you?

Something like that. Yeah. Think about when you first log into the application and you're on your dashboard. What if that dashboard actually becomes something that's really powerful, which is like, hey, these are the 3 things you need to pay attention to. I bubble— I, I'm bubbling up this stuff, you know, to help you be more effective. And these are the things that you need to, you know, either have, um, you need to take action on, or like, hey, did you know, uh, here's a valuable insight. That is where I see like, like, how do we make that dashboard like just super powerful for, for a user? Or maybe that's the only place they ever have to, to go.

Um, yeah. So much of property management is just completing small tasks. It's not a project type of business. It's a task-based business. And I often think about how if everything's going well or if everything's happening correctly, there should be nothing for anyone to do because just all the rent gets paid, all the owner disbursements go out, all the leases get renewed, and everyone's just sitting around. And that's actually proper. Um, if you like, in contrast to like a steel factory where like if you went out on the floor and there was no steel being produced, you'd be like, this is horrible. This is, I'm going to go bankrupt. This is an emergency. But if I go out onto my floor and look at my team and they're doing nothing, I'm excited. I'm like, this is great. They must've gotten caught up on everything and all the leases are renewed and all the rents collected and all the owners are happy. And that's ideal. I would love if we could reach that at like 3 PM every day. And send everyone home early. I think the challenge as a property management company owner is so much of the tasks are not visible or surfaced in an intelligent way. There's just so much contextual knowledge or just intuition that folks have to rely on. Like, oh, I emailed the owner, he never got back to me. Let me dig that up. Or man, I feel like this Property's been on the market for so long, we need to do something. And so you're constantly second-guessing yourself like, yeah, I better go log into this software and check on this, or I need to go over here and talk to this person to figure out what's going on with that work order. And I feel like the more software can surface these things and yes, automate them, but just letting us know as property managers what's going on and what we need to be thinking about is going to be a really big deal. That's the direction I see our industry going. It's such a reactive type of business. And so anything that's going to help us not drop balls is going to be really, really big.

Yeah, I, I like to think of it as you're very interrupt-driven. So, you know, you'll be doing something and someone walks in the office, or the phone rings, or you get a maintenance call. You know, all those things interrupt you from, you know, whatever task at hand is, and I don't know if you're like me, if I get distracted, it takes me a really long time to get back to the thing I was working on. Um, and, or I forget, and then like I get home and I'm like, oh man, I was working on that thing.

I never finished it. Okay. I always remember these things like in the shower or, or at the dinner table with my family.

Right.

And I'm like, I'm sorry, I just have to run and like jot this down real quick.

Like, I don't want to forget. Yeah, yeah. Wait, I, I voice dictate into my phone just to set a reminder the next morning. So when I get in back to my desk, I remember to do it. But it is, um, you know, I think if you didn't have to worry about those things, and when you did get interrupted, it's fine because it's probably something that's more important anyways, right?

Yeah. Last question I think I have for, for you today is around Um, like the conversational AI, is that going to be English language only, or is there going to be support for different languages there?

Great question. Um, you— it was very— it was something we were playing with, with it, you know, at NAA, and we're like, well, I wonder if we ask it to write the response, uh, to write the bulk message we're going to send out to like about the elevator, let's say. What if we asked it to write it in Spanish? Will it do it?

And it did.

And so that's the power of tapping into a large language model is that oftentimes, there's multiple language support. And so we're just getting the benefits of that. So I think there will be some opportunities to have some multiple language support. I think the challenge there is we got to make sure the quality of that translation is really good. But I see that English and Spanish is like, pretty much a no-brainer.

We want to be able to support that. Yeah. So something interesting about your answer that I wasn't expecting is that you didn't know what would happen. And I think that probably highlights something wild about large language models and AI is that you can't always predict exactly what's going to happen. So you're the head of AI and you didn't know what would happen if you asked in Spanish or if you asked it to produce a response in Spanish. That is fascinating to me, and I think that that probably shows the power of like, there's things that it's going to be able to help with that we haven't thought of yet.

Yeah, you know, I think this technology is moving so fast, you know, in hindsight, of course it's going to be able to respond in Spanish, right? Um, but you know, I think when you think about traditional software, you— we would have had to tell it explicitly, hey, make sure that we can translate this into Spanish.

Oh yeah, and it would— you would have had all this massive localization effort with all these professional professional translators and/or a big API lookup thing, or it would've been this massive, it would've had to have been part of the spec for the product, like on day one. Very interesting. And I, I'm also curious about the ability to, um, converse with the system in Spanish or in another language specifically because a lot of property managers are using global talent and yes, they speak English generally, but for them to be able to engage with the conversational AI in their native language might be a big help.

Yeah, yeah. I mean, there's no reason it shouldn't be able to do that because it does understand, you know, what the user is trying to ask it or get pretty close.

Yeah, I think that could be a really big deal and help, again, help global talent be more comfortable, come up to speed more quickly, be productive more quickly.

Really interesting. Yeah, that's the question I had for you is, you know, when you think about You know, you say global talent. I know a lot of companies now, you know, work with virtual assistants, or maybe they've outsourced some of their data entry to the Philippines. I know that's a pretty common locale. Where do you see more of like this sort of outsourcing headed? Or do you think that people will be willing to move outsource to an AI versus to, you know, an offshore team?

Well, like, are you asking, do I think there's a potential for AI to replace some of the global talent that folks are currently—

like, I think I'd be curious to know if there are certain things that they would want to replace versus things that they want to keep, you know, a human, the human touch, you know, there.

Well, I'm trying to think. So right now, property managers use global talent Generally, they start with using global talent for the most straightforward use cases, which is usually backend, either accounting or basic customer service, sometimes maintenance coordination or supporting the leasing department by processing applications and things like that. I think property managers are extremely pragmatic. I don't think they necessarily have an attachment to using global talent for this case or that case. I think they're going to use the tools that make sense for the job at the lowest cost, to be frank. And so if they're able to accomplish tasks using an AI instead of paying someone, they're definitely going to do that, whether that's a low— whether that's replacing local talent in the US or global talent. Um, one thing that popped up for me when you said that is Sometimes I hear property managers are concerned about using Global Talent due to security, right? Like, do I want these folks to have access to my client's bank accounts, numbers, or our applicants' Social Security numbers and things like that? So I could see that being something that would be really of interest is like anything, any, anything Any place that sensitive data is moving around, using an AI or a computer to do that versus global talent, I could see maybe that would be, like, interesting.

Yeah, you know, I actually, on the note of security, one of the things that the AI factory team has done is they've helped us build what we call our responsible AI framework, which is sort of a framework about how we think about developing these tools in a responsible or ethical way. And security and privacy is a key tenet in that framework, right? And that's something that I hear often, especially with these large language models. Is my data going back and training these models? And how are you protecting my data? And we've definitely thought about that. And how do we build the right safeguards and guardrails to make sure that we're not exposing any sensitive, you know, PII data or anything like that. And I do think that, you know, there are, you know, when you think about outsourcing, I think that is a good example of, hey, that is maybe an area for us as an organization to focus on when we're thinking about how do we release functionality that's going to allow people to feel confident that their data is secure, and then potentially, you know, move some of those, those, those outsource activities back into the software.

Yeah, I appreciate that you got— that Appfolio is not rushing this to market and is taking time to make the product good, put the right safeguards in place, You know, you, you've been working with AI, so you've got this framework that you mentioned around security and privacy. Like, I think that's gonna resonate with, with our industry. Like, it's the big buzzword, and yes, it would be exciting to be the first company to actually have it live, and I'm sure you're just so anxious to get it in the hands of customers, and we're anxious to try it out. But, you know, you know, I, I want to commend Appfolio on like Let's do this right. Let's, let's take our time and make it good and make it safe. You know, that, that's, that's all, all to the good, I think, for our industry. Because the last thing we would want is our clients, our clients, like property owners, to have some sort of a weird or bad experience with a property manager who was using like a shoddy AI product. And now all of a sudden, every new client's like, you're not using any AI, are you?

Like, that would be really bad. Well, it's interesting. I, I mean, I've talked to quite a few PMs who are using ChatGPT on the side as a productivity tool. Yeah, our team does some of that as well. Yeah, yeah. And it's interesting to hear the different, different use cases that, that they use it for. So I had a question for you, like, what is—

what are you using today to make your job easier or better or more efficient? So personally, I'm not using AI very much. I do pull up ChatGPT here and there and ask it random questions. I do know that my team uses ChatGPT in communications specifically to help soften language. Property managers, and my team is no exception, tend to be very, uh, driven, task-oriented, get it done right-to-the-point type of people. That's how they be successful in property management is, is getting stuff done right, getting that property leased and getting that vendor to call me back. When you're communicating with property owners and residents, there needs to be a shift in tone and an approach in your written communication. And ChatGPT has been really helpful for my team to come across the way they want to with, with our external comms.

Yeah, I think that's a great example. And I think that's an area where, you know, when we think about communications inside AppFolio, like that's a perfect example of a use case that could be really valuable. So if you're crafting a message in the application to send to a tenant, you could use a sort of, you know, communication generator. Like, here's what I want to say. And, you know, make it sound super professional and it could return something that you could then edit or polish up.

Well, this has been great, Kat. I really appreciate your time and expertise and helping me kind of go deep on AI. And it's been fun hearing about, you know, what AppFolio is up to on that front.

Great. This has been fun.

It's something I love talking about, obviously. Well, thanks again. And well, I'm sure we'll chat again. You know, maybe we can have you back, you know, once this is out and and kind of get some more secrets out of you. I wanna hear more about the AI factory and, and all that.

Thanks again, Kat.

Great.

Thank you.