United CTO David Dickey on AI adoption and strategies for brokers and agents

HousingWire

In this insightful interview, David Dickey, CTO of United Real Estate, discusses the transformative role of AI in real estate. Topics include AI's practical applications for agents, misconceptions about AI, ROI measurement, data privacy, and strategies for adoption.   Here’s a glimpse of what you’ll learn:   AI's impact on real estate agents' productivity   Myths and misconceptions about AI in real estate   Measuring ROI of AI tools in real estate   Data privacy and security in AI applications   Strategies for AI adoption among brokers and agents   Related to this episode:  David Dickey's LinkedIn  https://www.linkedin.com/in/david-dickey-7a67338/  United Real Estate  https://www.unitedrealestate.com/   The RealTrending podcast features conversations with the brightest minds in real estate. Every Monday, brokerage leaders, top agents, team leaders, and industry experts join us to share their secrets to success, trends, and the lessons they’ve learned. Hosted by Tracey Velt and produced by the HousingWire Content Studio.

Transcript

00:08You think you understand AI? You don't. Today's guest, David Dickey, he's the CTO of United

00:16Real Estate, goes into all things AI, especially building a brokerage platform. They just built

00:23Bullseye AI for their agents. We talk about myths. We talk about just unexpected costs when building

00:30this platform and strategy. So enjoy the podcast. David, thanks for joining the Real Trending Podcast

00:37to talk all things tech and AI. Thanks for having me. Appreciate it, Tracy.

00:43Yeah. So United Real Estate has had a lot going on. They recently released Bullseye AI,

00:49and you helped build that for real estate agents. So what problem were you trying to solve

00:56for agents that traditional CRMs and tech stacks weren't addressing?

01:01Well, our tech stack is our own proprietary tech stack. So we own all of basically the four pillars

01:11of our data. So we have all our transaction data. We have all our listing data. We have all our

01:17agent

01:17data. We have all our contacts in there. And so we're in a unique position because as opposed to,

01:24say, a point solution, like one specific piece of technology, we're able to apply AI across all of

01:34that data. And so really what we were trying to solve for is making our product a lot easier to

01:40use.

01:41Make it so that an agent could wake up and ask for a specific training, ask our bullseye assistant for

01:48it, and it would deliver that to them. Make CRM a lot easier. You can do things like just say,

01:54hey, I met with Tracy this morning for coffee, and I want to add her as a contact and put

01:59her on our

02:00drip marketing. And oh, by the way, remind me to call Tracy back next week. And it'll do all those

02:06things for you. So there's some really some great things that we're doing with Bullseye now from an

02:13AI standpoint. There's a couple different directions we're going. But overall, the thing we're trying to

02:20do is just make it easier to use and really engage the agents so they pick it up and use

02:25it and have

02:26better adoption with the platform and just make it easier to find things as well.

02:31Yeah. I think a lot of agents get confused like, oh my gosh, they've got this new AI product.

02:37How am I going to figure it out when it's really built into the system so it's more intuitive?

02:43Yeah. Yeah. It's like ChatGPT, really. And everyone's using ChatGPT. So the way ARDS works,

02:50it's very similar. You just talk to it. It prompts you and you tell it what you want to do

02:56and it'll do

02:57it for you. So I think if people are using forms of ChatGPT, then they're going to find this to

03:04be

03:04extremely easy. Yeah, absolutely. So there's a ton of hype around AI and there's a lot of

03:13misconceptions or use cases that really are just the tip of the iceberg, I guess. So what do you

03:20think brokers are getting wrong about AI in real estate today? Well, I mean, I think it's still

03:27early, right? It's an early technology. Well, although heuristics and AI have been around

03:35forever. It's just now being really applied in the space. And I think what people are getting wrong

03:42about it is, you know, that it's going to solve, you know, necessarily everything. And I don't think

03:48it's at that point. I think there's some things that are overhyped. And so there's a lot of experts

03:56out there now. You know, it kind of reminds me, I'm going to date myself here, but I was around

04:00actually when the internet came out, you know, and so it really reminds me of that time period.

04:07It's interesting because when the internet first came out, and again, it was, you had these really

04:14kludgy websites and people were like, oh, there's no way, you know, people will buy things on the

04:21internet. And there's no way Amazon will take off. And I remember all those things. And there were a lot

04:27of people that were so-called, you know, experts. All of a sudden, they were, you know, internet experts.

04:33And I think there's a lot of that right now. And you just have to kind of be wary of

04:37what's real

04:37and what's not real. And so, you know, what we're trying to do for our agents is just really deliver

04:46concrete, real world solutions that they can touch, feel and understand. And obviously, it's going to

04:53evolve and it's going to grow. And this is the first iteration of what we have with with Bullseye AI.

04:59But we've got some really, really exciting things that we're going to be rolling out.

05:03Yeah, well, I'll age myself as well. I remember having an AOL account. And at my first job,

05:10we had, it was supposed to be email, but it was more like a message board system

05:15that we had until they eventually got email. So, yeah.

05:21Right. You know, that's what you see with evolving technologies is they, you know, they change and

05:27they grow. And, you know, there's, it's super exciting. It's a great, you know, it's a great

05:34time to be, it reminds me of that. I was, you know, in my 20s when it came out and

05:41I was super excited

05:42about what, you know, that was going to do. And I feel the same way now. You know, I'm really

05:47excited

05:47about it. But I think at the same time, you got to kind of temper the results a little bit

05:53with,

05:54with reality, my opinion. Okay. And that kind of plays into my next question, which is really

06:02how or where or how is AI actually delivering measurable ROI for agents and brokers right now?

06:09Or maybe we just talk about agents first and then brokers.

06:12Yeah. With agents, I think the main thing that we're seeing is around content creation,

06:17you know, whether it's, whether it's, you know, obviously the most basic one is,

06:23is listing descriptions. But you've got social media posts. I think there's some great

06:28technologies around that image creation. I think there's some really cool technologies out there

06:37and use cases and applications for AI are really around content creation. And that's a lot of

06:45different forms. That's marketing, social media, you know, core listings, images. So there's some

06:52great things there. You know, we're applying it to lead nurturing. So we've actually created an agent

07:04that is we call Sophie, it works your sphere of influence. And I think that's a really exciting

07:10area. And that's where you can get some immediate ROI. So it's if you have, as an agent, let's say

07:17you've got 50 leads, contacts that you haven't talked to in the last, you know, six months, you can

07:24hand them over to Sophie, Sophie will work them for you. And then Sophie will bring them back to you

07:29when someone says, Hey, yeah, I'm ready to talk now. And so I think there's some great ROI around,

07:35you know, content creation, and around lead, you know, maturation and working leads. So

07:44and this is a question I've had is, what's the difference between AI and like predictive analysis?

07:53Like, isn't AI actually using the same type of, you know, information?

08:01Yeah, yeah, you know, it's, it's interesting, we've, we've gone to all of our brokers and said,

08:06Hey, what are some things that you would like us use cases to build agents, because we're,

08:11we're building a suite of agents, that that the agents, I'm sorry, that the brokers can access,

08:17and they'll come back with these use cases. And much to your point, Tracy, they'll, they'll say,

08:21Oh, well, I want to be able to have, I want to know when a listing is going to go

08:25on sale,

08:26I want it to be predictive in nature. And, in a lot of cases, they'll call that AI, but it's

08:31not

08:31really AI, you know, the best, the best way I think of it, that's just software, you know, that's

08:38just that. And so they've kind of come back, and they came back, and we, you know, we, we got

08:44a

08:44whole litany of ideas from our brokers. Let's call it 50. And I'd say half of them were just

08:52really more software, as opposed to, you know, AI, AI is something that learns, it changes, it grows.

09:00So for example, Sophie, when Sophie is talking to, you know, prospective buyer, and they say,

09:09Hey, I'm, I'm interested, not in Dallas anymore, but I'm interested in Austin. Sophie's smart enough

09:15to say, Okay, that's great. What, what, what part of Austin are you looking to live in? You know,

09:21you can't write software to do that. It's, it's, it's learning, and it's growing. And every time it

09:27interacts, it's getting smarter. So then the next time it may say, Are you looking to live on Lake

09:32Austin or Lake Travis, you know, and then, and so that's, to me, you've really got to, you've really

09:38got to look at the use case, and what someone wants to build, and understand, is this something

09:43that, you know, the AI agent can sort of have this continuous righteous loop of learning and getting

09:49smarter and smarter? Okay, thank you for explaining that. Because I have, I've never asked that question,

09:54but I've always wondered, like, the difference. I thought, well, they seem very similar. So, so yeah,

10:01definitely. And there's different types of not to be nerdy, but there's, you know, there's supervised

10:05learning, and then there's unsupervised learning, right, where you can actually take an agent,

10:09and you can supervise the learning. So for example, you can say, I'm the only person that's going to

10:14teach it. So when we turned Sophie loose, we turned it loose with our, with our internal sales agents

10:21who do this, you know, constantly, right? And so they were, they were teaching it. They're like,

10:26no, no, don't say it that way. Say it this way, or move it, move Tracy to close, you know,

10:33now go ahead.

10:34She's already said she's going to meet with you tomorrow at 12. So stop, stop talking to her.

10:39You know, she's, she's, you've got the appointment. And so that's, that's supervised learning. And then

10:46they, you can actually turn these agents on to do unsupervised learning where they're, where

10:51they're learning on their own. So I don't know if that, that helps, but that's, that's kind of how

10:56we distinguish what is an AI agent versus just what's software. No, that makes sense. And just to

11:02clarify for the audience, a lot of times when you're talking about aging, you're talking about

11:06an AI agent and not a real estate agent. Yes. Sorry. Yeah. I do that. I run into that all

11:10the

11:10time. I need to say artificial intelligence agent versus a real estate agent. No, you're right. You're

11:16right. Yeah. I think most people will, will get that by what you said after it, but just to clarify.

11:23Yeah. Thank you. Yeah. So how should brokers be thinking about their tech stack, um, build by

11:29partner, especially as it relates to AI? I know you built, you know, United built, um, not everyone

11:36can do that. There's a lot of smaller brokers out there. Um, so what should they be thinking

11:41about? Yeah, no. And we do buy sometimes, you know, if, if there's a commodity technology,

11:47you know, we're, we're going to buy it. We're not going to build it. You know, I think of it

11:50like, um, and I would, I would say, you know, brokers should think of it like, think of it

11:56like an F1 car. You know, we want to own the engine, you know, we want to own that, that

12:01core component because we want to be able to modify it. We want to be able to get feedback

12:06from our brokers and take it in different directions. And, um, we want to move rapidly to make

12:12changes and we want control, you know, and I think, I think it really comes down to data.

12:16So as a broker, I would just make sure I am in control of my data, like my contact data,

12:23my, my transaction data, my listing data. So we think of it that way. And I would encourage,

12:29um, brokers to think of it that way. Um, you know, the tires on an F1 car, that's a commodity,

12:35just, you know, whatever, you know, we, we, and I, and I'm are the best example for us.

12:42We didn't rebuild an LMS system. We use an LMS system that we've partnered with. It's a fixed

12:49cost. So, you know, we've got 25,000 agents and that's the other thing I would, I would warn,

12:54you know, brokers out there, you, you, you got to watch out for all these, um, variable costs

13:00where, you know, they'll put minimums in place and then they'll charge you X amount and they'll

13:05say, Oh, well, you're going to get a hundred percent adoption. So I'm going to charge you

13:08for 500 agents. And then that doesn't happen. And I, I've seen that over and over again.

13:13And I've been on the other side of this, you know, um, Dan, our CEO and I were from the

13:18tech

13:18space. And so we, we would get involved in all those conversations from both sides. So we

13:24try to do things to where, you know, when we're buying technology, it's, it's got a fixed

13:30to cost so that we're not caught as we grow into a situation where we get our margins compressed.

13:37So, um, yeah, it's, it's, it's control. It's owning your data where, where you want to, where

13:44you want to have, you know, that, that kind of build. And I understand a lot of people can't

13:48necessarily build. We've got 30 engineers, you know, and we're actually using AI to actually

13:53build a lot. That's where we're getting a huge advantage out of AI. But anyway, um, so yeah,

13:59that's kind of how we think of buy versus build.

14:02Yeah. I love that analogy. It makes it very clear. Um, for sure. So one concern, uh, concern

14:08we hear a lot is that AI can replace the agent's role, the real estate agent's role, I should

14:13say. So where do you see AI augmenting agents versus displacing them?

14:19Oh, well, I definitely don't think AI is going to, uh, replace agents. That's, it's just, you

14:25know, I'm, I'm a big believer in, um, you know, the last mile concept. Uh, I don't know

14:30if you know, uh, Mike DelPrede is, he's, he's got a lot of work on that. And, uh, you know,

14:36and I, I'm, I'm a big believer in that, that you've, you've got to have the agent, um, uh,

14:43shepherding a transaction and it's so emotional, you know, it's someone's biggest, uh, decision

14:49they'll make in their life financially. So of course they're going to need a person. I just

14:54think AI is going to make them a lot more productive. It's going to make them a lot more,

14:58um, uh, their, their, their job. I think they're going to actually be able to actually, um, use

15:03technology a lot more now, right? They're going to act, they're going to be able to interact with

15:08it in much more meaningful ways. You know, just the example I gave earlier about talking to

15:14bullseye and saying, Hey, add my contact and set them up as a seller and put them on drip

15:18marketing. That's huge. Um, and it really helps, you know, the agent, uh, uh, to, to be more

15:25productive, uh, to do more things, to engage with technology more. And, and so I, I don't

15:32see it disintermediating the agent though at all. Yeah. Yeah. In fact, I know a recent story

15:36where a seller claimed they did the entire transaction through a AI. Um, but the buyer's agent disputed

15:45that and actually said they left money on the table because of it. So yeah, I'm sure,

15:50you know, there's a concept you've probably heard of it. Um, human in the loop, uh, around

15:56AI where you've got to have a person watching AI cause we've already seen, you know, what

16:03AI can do. Yeah, definitely. Yeah. Um, so what do you think are the biggest adoption barriers

16:09you're seeing among agents and how do you overcome that? Uh, you know, I think, I think the biggest

16:17barriers to adoption are around, um, just being kind of afraid of AI, you know, or, or being

16:24afraid of the technology. Uh, you know, there's a lot of different emotional reactions. You can

16:30have the new things you can, you can cower, you can sort of say, Oh, it's not real. Um, or

16:37you

16:37can embrace it, you know? And so our, our top agents are, um, you know, embracing it and

16:43they're curious about it and they're learning. And, you know, again, you just kind of have

16:47to have the right, uh, I think the right disposition towards it and, and, you know, and sort of embracing

16:54and, and, uh, liking what's, what's to come and being curious, you know, more than anything.

17:00Yeah. I think being curious is, is true. I mean, I even see that at housing. Why are the people

17:05who, you know, really dug into our, we have a, um, proprietary system, um, and the people

17:14who dug in and the people who are like, I don't, I don't know how to use that. I'm not

17:17going to try it. I'm going to screw it up. Um, and so it's, it is interesting, you know,

17:23um, but definitely being curious helps a lot. And, and for your, how do you, um, encourage

17:32adoption at United? Um, you know, are there specific campaigns or, um, training? I assume

17:40there's training, but how do you encourage it? Definitely. So I think, you know, we, when

17:45we rolled out, uh, Bullseye AI to our brokers, uh, and this is pretty recently, we got some

17:54really good feedback and I, it's going to lead into what you're asking. The biggest thing,

17:58and it was an aha moment for, for me is, um, having local content, you know, again, we're,

18:07we're, our home office is, um, you know, we've got brokers all over, you know, Nashville,

18:14Atlanta. And so having that local content in Bullseye and having local, even things like

18:21events and announcements, um, but having the, the, the, the policies and procedures and what's

18:28going on, the broker support and having that AI work at a local level really, really is helping

18:36with our adoption. Um, it's, it's bringing in the agents because it's not that they're just,

18:44there's a lot of great things. We have, you know, hundreds of different programs, referral networks,

18:49um, uh, financial, uh, you know, planning programs, and we have all these things nationally,

18:56but the good old fashioned, Hey, there's a pizza party next week, and we're going to go over

19:01training. And by the way, you can talk to Bullseye assistant and it'll tell you,

19:06you know, about compliance and all those issues. So really getting that local content

19:12into our platform, which it supports that. And then having, um, Bullseye assistant work with that

19:20local content is really driving a lot of adoption and making it easier too. We talked about that.

19:25You've got to make it easy, you know, having it so that it's, um, so that you can interact with

19:31it

19:31and tell it to do things that before you used to have to type in and, you know, log in

19:38and type in

19:38and go here and do this. And, you know, the traditional training, ah, you don't have to do

19:43that anymore. You can just tell it. So, yeah, yeah, that's amazing. And I think now things are

19:50changing. You know, States are making laws about biorepresentation agreements. They're making laws

19:55about private listings. So every state is doing things differently. They, they already did, but like

20:00it's constantly changing. So having that local knowledge, um, obviously helps the agents for

20:08sure. Yeah, it's pretty, it's pretty cool. We we've got, um, we've got a concept that we call

20:13bullseye drive. It's like a Google drive and the local offices can load up their content, you know,

20:19um, into that. And then the bullseye assistant, the AI can actually, um, go to that first and,

20:26and answer questions, uh, at a local level. So that's, that's been a, that's been a great

20:32feedback point, um, you know, from our brokers, which is another point of having your own tech

20:38stack. You can listen, you know, we listen, our only customer are our, our brokers and our

20:43agents. And so we can control that roadmap, which is pretty cool.

20:48Yeah. And so you have 24,000 agents. What do you, what, um, what do you see top producers

20:55kind of patterns that you see top producers, um, doing when it comes to AI and embracing

21:01when it comes to AI?

21:03Yeah. And again, the top producers, we kind of hit on it earlier, but they're, they're the

21:07ones that are embracing it. You know, they're the ones that are curious about it. Um, it's

21:12okay to go out and, and, and mess with it and make mistakes and, and try it, you know,

21:18so that's what we're seeing our top producers do. And, um, you know, it's, it's, it's those

21:24folks. And again, it goes back to the content creation where they're, they're taking advantage

21:29of it. And there's a lot of, of, of great products out there other than, you know, bullseye,

21:34um, AI that, that allow them to take advantage of that. And so, again, I think it's the top

21:39producers, I think the ones that go out and, and really are curious and embrace it are,

21:45are the ones that are doing the best.

21:46Yeah. Um, so there's a growing conversation around data ownership, privacy compliance with

21:52AI. Um, what should brokers and agents be, agents be paying attention to right now?

21:58Yeah, definitely. Um, boy, this is a big one too, but you know, kind of a couple bullets

22:04on it. Um, first off, be careful about using like free chat GPT, putting your customers in

22:10there or whatever, because you're basically putting it in the public domain. Um, you know,

22:16so, and I've got all kinds of stories about that, but, but, uh, you know, it's, there's,

22:21there's some crazy mistakes you can make about, um, putting sensitive data, you know, into the

22:27public domain. So, but what we're doing with it is, um, you know, our bullseye assistant is actually

22:36working off of large language models like chat GPT, um, and Claude and various ones. Um, but then

22:44what happens is it actually creates what we call, um, a narrow language model, an NLM. Um, and, and what

22:52that is, is it's, it's, it's basically using that technology of those large language models, but

22:59the guardrails are within the bullseye ecosystem. And so, you know, we have, we're protecting that

23:08data inside of there. We've got, um, the privacy protections in there and then we have the guard

23:13rails set up, set up that, that keep it, keep that data inside our four walls. Um, and you definitely

23:21have to, to watch that. Um, so it's, it's a, it's a big thing. And then, you know, there's the

23:27whole

23:27token conversation around, you know, cost of AI, um, that, that we're, you know, we're learning a lot

23:35about right now. Um, so, but anyway, uh, the guardrails are the main thing and, and just make sure

23:41you're using, um, you know, like a pro version of, of, uh, chat GPT or whatever, and then you should

23:48be

23:48in good shape. And then the token, if I don't know a lot about it, is it something like if

23:53you,

23:54certain search words or something cost a certain number, like a certain token, which is money?

24:01Yeah. Well, well, AI is not free. Um, that's the, uh, imagine, imagine that they're going to want to

24:06get their money back. Um, you know, the, the hyperscalers out there that are doing all this,

24:11they're putting these billions and billions of dollars into AI. And, um, it's real interesting

24:18because, because what basically anytime you talk to, um, you know, chat GPT or bullseye assistant

24:25or anything that's using it, there's a cost associated with it. And it's based on, um,

24:31no, not to get all again, get too in the weeds, but it's based on basic how much you're asking

24:36it,

24:36how complicated the ask is. There's this notion of vectors and, and then there's costs for upstream

24:41and downstream. And, um, you know, we've actually built tools inside of our software that can track

24:49those costs by user. Um, because, you know, we've got to watch it because when you've got 25,000

24:55agents and if one, eight, and if an agent per agent is costing 10 bucks a month, well then

25:00it's a quarter million bucks. So, um, but so we're, we're monitoring it and we're watching it,

25:06you know, this is more at scale though, you know, but it's an interesting thing because, uh,

25:11and you'll see what they'll do. You know, I've, I've done some, I've worked with some AI platforms

25:16to do like vibe coding and whatever. And what'll happen is that as you interact with it, after about

25:2310 questions, it'll say, okay, Dave, you've, you've hit your limit for the day. Why don't you take a

25:29break or put some money in the machine, you know, and that, that's, that's kind of how they all do

25:35it now. So, yeah. Okay. I think it's a cost that most brokers don't think about. So, um, yeah,

25:43exactly. But it's there and it will be there the more and more that people use it. So, yeah. Um,

25:49so if you could offer a broker to CEO, one piece of advice about AI strategy over the next year,

25:55what would that advice be? Yeah, I would, I would, uh, just get started. Um, and there's some great,

26:02um, I use a, uh, an email list that I subscribe to. It's called the rundown AI and, um, I

26:12get it

26:13every day and it's, it's cool because, you know, it's, it's, it's got, it's more broad, right? It's

26:20more about AI in general, not necessarily about real estate AI, but it's pretty cool because

26:25it's easy to read. It's a five minute read. They have free classes you can go to and learn about,

26:30and it's just a great way to stay on top of, uh, what's happening, but you've, you've got to get

26:37in

26:37there and just kind of learn a little bit about it. You don't have to, um, have the anxiety or

26:42the

26:42FOMA, you know, about it of, Oh my gosh, I'm falling behind or, or whatever. But you know,

26:50my advice to the brokers would be just to, just to learn a little bit every day and, um, it's

26:57not

26:57going away. We're going to be dealing with this for, you know, from now, from now on. So just start

27:04learning about it and, um, you know, uh, get in there and get after it. So, all right, I've got

27:10some,

27:10uh, lightning around quick hit question for you. Um, the biggest AI myth in real estate in one

27:17sentence. Oh, the biggest myth, uh, that it solves everything. Uh, it does not solve everything and

27:25it's not going to replace everybody. Great. Um, AI replaces which task first for agents?

27:31Content creation. Again, we talked about that. So content creation, doing listing descriptions,

27:37pictures, social media posts. So in three years, AI will matter most in which lead gen conversion or

27:43client experience? I think lead gen, I think it basically takes over and can do that for you.

27:51Well, great. Well, David, thanks so much for getting on the real trending podcast today.

27:55I really appreciate all your insights. I learned a lot and I'm sure our audience will too. So thanks.

28:01Thank you, Tracy. I appreciate it.

Category

Transcript

Comments

Recommended