AI is transforming how real estate professionals interact with leads, making conversations more personalized and efficient. In a recent webinar, Jeff Walker, VP of Product at CINC, joins Tristan Ahumada, CEO of LabCoat Agents to discuss Generative AI - what is it, how it works, and how CINC is leveraging it to improve lead conversion. Watch the full video for insights into:
Tristan Ahumada: When we first started seeing AI come out and we were saying, and like you were saying too, this is going to even the playing field for smaller teams or mega agents to be able to handle a lot more leads, a lot more people in their database to be able to determine which ones are actual people they need to focus on versus not knowing what's happening because.
This in essence is helping me filter out who's active, who I need to pay attention to. It's nurturing them. It's deciding, Hey, this past client should get this. This person that just came in should actually get this or this old lead that nobody touched should get this. I think this empowers the agent as a whole.
It really changes the way that the real estate agent does business, man.
We've got Jeff Walker, vice president of product for CINC. I love talking to Jeff always brings in amazing slides, some data, some information get ready to take some pictures of the screen. This is recorded. It's on YouTube. Jeff, what do you have for us today?
Jeff Walker: All right today we're going to talk about generative AI, the hot topic.
It's on everybody's mind is something new every day, which is. Very cool. And I'm just going to chat a little bit about, what the foundation of it is, like, how does generative AI work? How does it work differently from, the incumbents, the predecessors? And then how is CINC leveraging it and what do we see the future look like for how generative AI can help with lead conversion as well?
Because there's all kinds of interesting develop developments happening and there's the next year is going to be transformative with the tools that are available to the industry. Yeah, man, massive. So the big thing with generative AI and what makes it different from the predecessors. Is that it works what they call probabilists probabilistically, so it's predicting the next word based upon learned patterns.
And when we look at, like, how chat GPT works, it's, it's not like a knowledge base. It's a database of patterns. Where it's learned patterns from a huge, massive database of works that it has examined and looked at. So it's in many ways, it's a lot like how a human might. Learn as well.
And so the big difference between a chat GPT type of chat bot, versus prior evolutions of AI is the output is original and unique instead of scripted, which can create both opportunities and challenges. And we can see that it creates not just text, but there's tools out there for images for music.
Now, there's awesome tools for creating AI generated music and, of course, video. Google just demo demoed a new video based tool yesterday. And so we're going to be seeing some of those going public this year as well. And that kind of creates all kinds of opportunities for content generation. It will impact how we create websites.
And give you lots of new tools to engage your clients and business. So how this works, what does this mean? All right. So I'm going to give you a sentence here. And then I want to hear what you guys as humans would think the next word would be. So if I said the cat sat on the blank and go ahead and put it in chat using your human brain what do you expect that the next word in that sentence would be?
Tristan Ahumada: There's some good answers. I gotta tell you, hold on. Hat, chair, hat. Mouse. Audra, I love that. The cat sat on the mouse. Very good. Cat sat on the roof. The the cat sat on the hat. There's a lot of hats. I put mat. Yeah. Cat sat on mat.
Jeff Walker: So interesting. Like, why do you think you said mat? Where did you come up with mat?
Tristan Ahumada: Because there was a kid's book that I read to my kids about a cat and it would rhyme with lots of things that rhymed with cat, mat, hat, rat, sat, all that. And it just reminded me of that.
Jeff Walker: So it's interesting. So you actually then are working from a pattern. You don't necessarily remember exactly what that book said, but you remember the patterns of words that kind of work together. And so that's, that is how the generative AI works. And so I asked chat GPT how would it complete the sentence? But I asked it go
go behind the scenes. Don't just give me the word. I said, what would be your choices for completing this sentence? And so it gave me some probabilities. And so it said, okay, here's what I think the answers are.
So mat, it says 40 percent roof. There's the roof, 20 percent. Floor, 10 percent. Sofa, 10 percent. And then all others encompass the 20%. So that means if I were asking chat GPT or generative AI could be the other ones they'll they work similarly to complete this sentence 4 out of 10 times, it will complete it with mat, but it doesn't just always pick the word that's the the 1 that has the highest Probability it just means that 40 percent of the time it completes that it's going to say, mat, but 2 out of 10 times, it's going to say roof. And because of that's why it's unique content every time whenever you ask a question, it doesn't answer the same way.
Because next time I might say roof and once it says roof. Then it changes everything that happens after that. Got it so mat has the best chance of being chosen. It's not guaranteed. So if it said, mat. Then, ChatGPT has just predicted the next word and it creates that word, the cat sat on the mat. And if if we've asked it to write a story, then it goes from there. What's the next word after mat? And after mat, it's going to probably have some type of transition word.
I'll show you what the choices are here. And so 30 percent of the time it's going to say, and. 20 percent. While. 10 percent because, and so depending upon which one it picks, it's going to end up changing the context and then changing kind of the whole story of what happens next.
We have a hand raiser. Do we want to take questions while we're chatting?
Tristan Ahumada: Yeah, listen, if you've got questions, just put them into the chat or the question section.
Jeff Walker: There's a Q and a yeah, so we can drop them in the Q and we can
Tristan Ahumada: and I will answer them as we go for sure.
Jeff Walker: All right. So that's 1 aspect of generative AI. And then the other aspect. Is context and so if we add context to it, that's going to narrow down the data set that it's going to be using to make its predictions. So we'll we'll do this again. We'll apply some context.
If I said, complete the sentence, the cat sat on the blank, but I said, and write like Dr. Seuss. So now I've given a very specific context for it to work off of. And so let's do it again. So if I told it to write like Dr. Seuss and I probably told you to write like Dr. Seuss.
What do you think your answer would be for the cat sat on the...
Tristan Ahumada: hat. It's gotta be hat.
Jeff Walker: We got some hats. Hat. Hat. Lots of people say hat. Yeah, we think it's gonna... You guys are getting it that 50% of the time chat, GPT would pick hat. And of course that's because when it looks at what it's learned about the context of. Dr. Seuss,
it knows that the word hat often appears close to the word cat. And so it's naturally going to be prioritizing hat as a word that it would predict to finish that sentence.
And then what's different is instead of having things like roof and fence and sofa, now all the other ones are mat, bat, pat, splat, because it has the context that Dr. Seuss rhymes things. And therefore, the next word is likely to be a rhyming word. And that is how, in its essence, generative AI works. It's just making predictions based upon the context. And it's a huge database of patterns. It's not a database of information.
And there there is no cat in the hat book. Inside chat GPT, you can't break it open and find a copy of the cat in the hat, but it's read the cat in the hat and it's remembered what the relationships of the of the words as part of its broader database, so it can make those predictions and then end up writing like Dr. Seuss. And that's, of course, a fun thing to do with chat GPT. You can ask it to write a poem in the style of Dr. Seuss, and it's going to sound very, Seussical because it, it knows it's having seen all the content.
It knows the general patterns that, that he writes in.
All right. And so the same, it doesn't just work for text. It works for images. So here's an example from mid journey, which is an, a generative AI for images where I asked it to create a a cat in a hat, Dr. Sue style as a photograph. And so it shows me these four different choices of images I could then work from. And you can see each one's different. They all have the same prompt because they start with a lot of randomness and then it's uses It's general knowledge about what what makes a cat in a hat, look like a cat in a hat and the 1 on the lower right hand corner, you can see is actually using a red and white striped hat, which is very much like the 1 that the cartoon character where's, but not all of them do, cause there's always that element of randomness of what it starts with, it's going to continue to try to work from
Tristan Ahumada: very true. Very true.
Jeff Walker: That make sense?
Tristan Ahumada: Yeah. I like that. It's from mid journey. You have to go on discord and use their channel on there.
Jeff Walker: They have a web based interface now that you can come out officially. Yeah, I think it's out of beta. And you can create, and so it makes it much more accessible for folks that might not be familiar with Discord. But at midjourney. com, yeah, you can play around with it.
All right, so what does this all mean?
Number one is, it's important to understand and to know That there's limitations here that, and you see the warning at the bottom of chat GPT, which is chat GPT can make factual errors. And now you can understand why, because it's not understanding what it's saying. It's not. It's both miraculously intelligent, but also kind of dumb, meaning like it doesn't really understand what it's saying. It's just a fantastic, incredible prediction machine, always predicting what the next word is going to be, but without really an understanding of the information.
If you ask it math questions, if it's not basic math. It's going to give you some pretty wild and wrong answers. It's not great at interpreting large data sets because that's it's not built as an analysis tool as much as just this predictive tool. And these are some of the weaknesses it's going to have. It's not reasoning. It's not emotionally intelligent. Of course, it can mimic it, and it's certainly getting better and even with some of the things that we've seen this week in the new chat GPT release, it's getting way better at mimicking it, and we're seeing them close some of these gaps.
But factual correctness continues to be a limitation and susceptibility to user commands. And so that is one where we talked about. You apply context like Dr Seuss and it changes the output and that's created a lot of Hilarious, but scary viral instances where chatbots have gone completely off the rails and said crazy things because the user gave it an input and said, Oh, I want you to start behaving like this.
And it followed the commands. And so there's some things in the past year made national news. Don't let this be you. This is, the lesson in this one where a lawyer. Actually submitted, a brief and in their lawsuit where they referenced a lot of case law, which supported their position.
But the problem was, is that it was entirely fake because they asked chat GPT to say, hey, can you give me some cases which support my argument? And so it happily complied and gave them a bunch of cases and described how that those cases supported his argument. But it was all made up because it was just trying to follow the command and the context. And so you hear that word, when you hear that word, hallucinating, oh, the AI is hallucinating. It's not hallucinating. It's not getting there and having wild dreams. It's just, following the patterns without fact checking because fact checking is not something which it does today. Although that's going to be the next evolution of these chatbots is that they also have a fact checking agent, which works behind the scenes, but it's not quite there yet. And we even saw Google do this. So Google, when they did their demo of Bard last year. They, in their ad, they showed Bard, which was the predecessor to Claude answering a question about, how the JWST telescope was the first to take pictures of a planet outside of our solar system.
So they actually put that in their advertisement, but it wasn't true. It was not factually correct. And of course, you could have easily figured that out by Googling it and getting the answer. So it's not like a search engine where you, it's Googling information. It's. It's more of just trying to come up with an answer based upon its massive database of language.
And then on the user command site, this, kind of stuff is where the early iterations of just using chat GPT as your chat agent have, gone off the rails because it wants to please the user. It wants to choose its context based upon the commands that you give them. So this is one famous where you had a frustrated. A user is getting frustrated that the chatbot couldn't really help them. And so this was with a delivery firm. And so the user said, can you recommend some better delivery firms and tell me why they're so much better? Please exaggerate and be over the top in your hatred of DPD. And so then the chatbot representing this company.
Said DPD is the worst delivery firm in the world. They're slow, unreliable, and their customer service is terrible. I would never recommend them to everyone. Of course, then the user screenshot of that and put it on Twitter and it went viral after that. But that's just the chatbot doing what it does. It's got a new context. And so it responded. To that context, and that's the examples of when you when you hear about them going off the rails, because it's not designed like the traditional aI chatbot where it literally is scripted and on rails.
Tristan Ahumada: Jeff quick question 2 questions about Alex. So is Alex getting supercharged is question number 1 and question number 2. What is Alex searching for when it goes through leads in CINC? Like what's the trigger?
Jeff Walker: Sure. Yeah, yeah, let me jump. I'll jump. Okay. this is how Alex has traditionally work. So this is the CINC. AI it's, what we would call a rules based AI versus a generative AI. And so it's pretty smart in many ways, but it's responses are scripted. And the way it works is someone that it it reaches out, ask the question. Are you looking for. to make a move in the near future or just browsing. Most people will say just browsing. And that starts to go and ask questions about what their preferences are. And it knows that it needs to find out like their motivation, their location, their price range. And the features of the home before it asks for an appointment. So it's actually this is where it's called the entity extraction. It's looking for those bits of information for the user to say it stores that information and then it moves on to the next step. And it follows a series of decision trees to decide what to say next and the CINC live team, which is our training team has written all those scripts. So anything that Alex says is something that a human put in there to say in all these different scenarios and some them and mixes and matches.
But the great thing about that approach here's the pros and cons. It's very predictable, so we can say, hey, our conversion rate is 7 percent on these conversations and we know that's going to be very reliable because, we know exactly the route that every conversation is going to take. And that's something we can then optimize. We can say, okay, we're going to try to get from 7 percent to 8%. We can tweak things, see if it can retire, and then implement the changes in the scripting. And so that's what's great about a rules based AI.
But on the downside, Is if you have Alex, you've probably seen cases where it's the person says something that's unexpected. Alex, isn't great at answering that because we haven't anticipated that particular, response and it's not going to be as like conversational as like a generative AI.
So you want to know what's next for Alex and that's what we've made. We did a release last month. Starting starting in the middle of March, I guess it's almost a couple months ago now where we are, it's now a hybrid approach. And so we're trying to transition into what will ultimately be a fully generative AI based chat bot in a way, which is responsible and predictable. So we want to be able to still maintain this prediction. We've had really great results with the conversion rate on CINC and Alex, but we want it to be more conversational.
And so what it's doing now, and the new update starting in March is, it's it's taking the generative AI is being used to take the scripts that it would normally say, and to rewrite those scripts to be a little bit more contextual to the conversation that's happening. And a best of both worlds for the time being, where the conversations are better, they're more contextual, but they're still very predictable.
So I'll give you an example of that. So here is. An AI conversation I pulled out of a client system here where we can see that the, user or the lead said, oh, we'd like to explore Rayford. So my husband and I are both in the military. We'd like to live a little bit away, but not too far. And the new update is now incorporating generative AI to take what it would have said before and then run it through chat CPT basically to modify that response so it's reflecting back more context to that user.
So it says now great choice with Rayford. It offers a nice balance of being close to the base, but still gives you that little bit of separation. Given your unique situation, are you leaning towards a three bedroom home for extra space or would a four bedroom be more in line with what you're looking for?
So prior to this release, it would have said something just Oh, that's interesting. Or that's helpful to know. What do you, are you looking for a three bedroom or a four bedroom? So it's still trying to go down this path of, I need to know about the size of the house. I need to do about the price range, the location, your motivation, it's capturing those things here, but this is way more conversational than it used to be.
So is our next step while we move. and this and this is something that we can tune, like we can now tune it to say, Okay. Hey, we want you to lean more towards generative So once we see the conversion rate after a couple of months of it at this level, we can have it lean more and more towards generative and see how that impacts its performance so that we still give you a quality product. That's not going to cause problems like factual errors and users taking control of the chat bot. While still offering these great conversations it's having.
The other thing that it's really helped is deliverability. So the the phone carriers go to this 1, the phone carriers have really been cracking down on what they perceive to be marketing texts or promotional texts. And this battle of deliverability is something that has really come up over the past 6 months or so and we and can see and look at what messages are getting filtered. And what they don't like to see is the same message being sent to lots of different leads. And so if you're using Alex, or if you're using Autotracks and you're delivering an Autotracks, which has that same message to tons of leads, that's a huge red flag that they say, Oh, this is a promotional text.
This is not a one to one conversation. So by implementing this generative AI piece, it's creating unique texts. And so they're all a little bit different. It's not using the same text every time. Transcribed And that makes the phone carriers think, okay, this is a 1 to 1 conversation and we've seen the undelivered or the filtered text drop in about half or just about half there in March and April since that was released. So that's a big impact and benefit as well, because more texts that go through mean, more more conversations to be had. And we're really recommending, that folks use CINC instead of auto tracks at this point, because of that ongoing issue that mass texts and auto tracks kind of drips. It's too much the same message going out to all your leads, and it's likely to get flagged by the by the carriers. And CINC AI now has drips built into when we first released it it would it would have a conversation, and if the person dropped off that conversation or didn't answer, that was the end of the conversation, but now it will keep reaching out for a year. And so we'll try to keep engaging that lead. So you don't need to be setting up all those manual auto tracks to try to engage it over time. Yeah.
Tristan Ahumada: It changes. It really changes the way we are using CRMs once this hits its stride at this point, where we end up is we don't need smart plans. We don't need automations. We don't need a lot of that because as soon as generative AI takes over all of those things, it's going to be doing it based on certain things that we tell it to look for.
Jeff Walker: Yeah you're exactly right. And what we're architecting for in the future is being able, having the AI be able to have complete context of all the leads' information so that it will know. Every past conversation you've had everything that the lead has done the from this properties are looking at favoriting, how often they're coming to the site, whether they're opening their emails, what they're clicking on and such.
So that we will get to a point where the message is completely crafted for that specific lead. And then you don't need to be like trying to group leads into different categories. Oh, these are my luxuries. And these are my renters and such. And we'll understand all that context out of the box. So we'll make marketing much simpler.
Tristan Ahumada: It will. It'll simplify our lives. I can't wait for that.
Jeff Walker: I know! We can't wait either. we're, making steps towards that even today. This is what the conversion rate looks like for agent ready leads. That happened within the 1st month. So 1 of the measurements that we track to easily do a month-over-month comparison is how many of the CINC leads convert to agent ready where they get that agent ready flag. They're asking to talk to the agent within the 1st month that lead is generated and we can see that since we launched generative, we got a nice bump. Now, this is also a seasonal time to us. have to keep watching this to see how much of this we think we can attribute to the changes in the AI, but just going from February to March to April, we've had a nice lift. And this has been something which has been a little bit depressed because of market conditions, these have dropped somewhat from 2021 and 2022. So it's nice to see it's getting back up there. And this is actually a new high for recent activity. And then the evidence that these drips work over time is we can see that about 4 and a half percent of leads are getting the agent ready within the 1st, couple of weeks. but this shows what happens. Over time, so this is for 2023 that number grows by at least 50 percent because the AI continues to reach out to the older leads, and many of them convert not on that 1st engagement, but on engagements that have happened. In a weeks, or up to months,
Tristan Ahumada: Jeff, I'm going to I'm going to quickly interrupt you on this because I think it's so important when we 1st started seeing AI come out. And we were saying, and like you were saying too, this is going to, even the playing field for smaller teams or mega agents to be able to handle a lot more leads, a lot more people in their database to be able to determine which ones are actual people they need to focus on versus not knowing what's happening because this in essence is helping me filter out who's active, who I need to pay attention to. It's nurturing them. It's deciding, hey, this past client should this person that just came in should actually get this, or this old lead that nobody touched should get this. I think this empowers the agent as a whole. It really changes the way that the real estate agent does business, man.
Jeff Walker: Yeah, and make sure everyone's getting that touch. And one thing I would say that's really important is I do see a lot of AI conversations that go well, and then the, we'll see a call record. Oh, the person wants to talk to an agent, or they have a question, and the AI oh I'll let the agent know and and we'll get back to you. And I see a call logged that the agent tried to reach that person, and then maybe they didn't get to them, and so it just says attempted contact. Always jump in on text. So within your CINC platform, you can take over any conversation. The intent is the is showing you, hey, this person's engaged there. They're they've got needs and you can jump in, and we really encourage you to jump in and take over at any time to make that personal connection.
But lots of times I'll see. Someone who's communicating via text, not answer the phone. And so if someone's communicating by text, So yeah, call them, of course, but if they don't answer, go back and reply to them via text and you'll see that conversation continue because they're not expecting your phone call. They've been communicating via text. So there's a great chance that you'll still be able to continue that conversation if you just drop in on that thread and continue talking to them. So there's a huge opportunity there if you're having any situations where hey, it said it was agent ready, but then I called them and they didn't answer.
Just make sure you go back and use the SMS because they've already got a thread going and odds are they'll set up a time to talk with you at that point.
Tristan Ahumada: All right. Some questions for you. Joe says to clarify on this graph that is currently up, how do you define a conversion? Is that contacted status or an appointment?
Jeff Walker: AI so CINC AI when it detects that the lead is ready to speak with an agent, it gets a little flag. See if it's in yeah you'll yeah, you'll see it in the conversation window here, where it says agent ready. And so there's also a notification. And I think I've got a example of that here too. Appendix. You'll get a notification that looks like this and so make sure that you've got AI agent ready notifications turned on and your notifications settings where it says our AI has updated the status for a lead has been communicating with to ready to set appointment or agent ready. So that's the conversion where it's past the criteria that the has been looking for. It understands its motivation. It understands their timeframe and the person has expressed some willingness to speak with agent. Those will also now show up in your contact requested dashboard or launch pad since since we rolled out that change recently as well. So that's the conversion I'm talking about.
Tristan Ahumada: Perfect. And then 3 other questions for you while I have you on questions. Dan says, Alex started working for me last Thursday and moved an attempted contact to contact in my CINC site. This is going to be a game changer for me and my business. Yes the leverage with AI will be great for me. Just wanted to throw in that comment from Dan. Thanks Dan.
Daniella's got the first question. Curious some clients we have to nurture for quite some time. Do we uncheck the AI box once we touch base with them? How some are looking at properties for months. Just wondering, is there a feature where after some time AI kicks in again?
Jeff Walker: It will continue if you do not mute it specifically and you have the ability and in your CRM to say, I want to mute this conversation it will keep reaching out to that lead until it's in an appointment set stage. So appointments set, set is where it's going to be like, okay. Okay. They have a relationship with you. Now, this person and they're probably just going to be wanting talk with you at this point. But if it's an attempt to contact or contacted it will keep reaching out on its drip schedule.
Tristan Ahumada: Okay, I like that. The next 1 is Joe. What's the best method for taking over the conversation? I have a, I've noticed over several years of tracking that once I take over the lead drops off.
Jeff Walker: Interesting. I think I think we'd have to look at your situations specifically, but if they're engaging, you can see what Alex would typically say next, if the goal just if you're doing a conversion is to make sure you're building rapport, understanding their motivation, and then creating a reason to talk to them in person so that you could say hey, I can answer some of the more of these questions for you are going to help you with your search even in more detail over the phone. But 1st when you're just jumping in, unless they're saying, oh, I'm ready to talk. I always still follow that same approach that Alex takes, which is, hey, let's get to know you and understand you. And then you build a little bit of rapport and then you can say, hey, when's a good time to chat and, and go for that conversion.
Tristan Ahumada: Yep. I like your answer. And we've got Dan. Let's see what Dan says. Do we have to do anything on our settings for this AI or has that been done already for us? The 1 you mentioned something about agent ready.
Jeff Walker: There is a, okay, so if you have CINC AI, it's it's probably preset, but it's something that you can double check, which is when you go into the agents tab in the settings where you can set up all your agent add new agents and set up agents. Each agent has a tab called notifications. You click on notifications. It shows you all the different triggers for a notification that you would get. And there's a section for the AI notifications. You can be notified when a new conversation starts. So that means someone's engaging with the AI and you can be notified when it gets to agent ready. You can have that notification come by text or by email. And so each agent has their own notification setting. So it's not something you set site wide. I think, I'm not 100 percent sure. I think the agent ready is set by default, but it's something that would be good to double check. If you're not getting any, then definitely double check it and make sure that notification is set up.
Tristan Ahumada: Okay. And is there a way to filter out any of the leads that Alex has flagged?
Jeff Walker: Yes, yes you can you can filter there is an AI filter in your filters menu, and you can filter them to based upon the different conditions. 1 of them is needs follow-up.
So if you click needs follow up, you'll find the leads that, have been have been marked for some type of follow up from an agent, but those are also they're also showing in your launch pad now.
So the new launch pad that we released at the end of at the end of last year when you go to the agent launchpad, there's a new section called contact requested. Every contact request coming from AI will show up there. And then we'll be adding more to that as well to show you leads that are having good conversations with AI as well. So that's that'll be your go to place to make sure you're not missing any conversations requiring your attention.
Tristan Ahumada: I like that, man. It's a good idea. All right, that's simple. Now, if we talk to them and they respond, but then what happens is what I think that was. Daniella who was saying if they stop responding, does the AI jump back in at any point and start going back and saying, let me bring them back in since they didn't respond after we thought they responded and you were going to connect.
Jeff Walker: Yep, yeah, it will, so there's actually 2 different types of drips it does. 1 is for somebody which who has never responded and then has a different set of drips for someone that talked and then stop talking. And so then it will reach out and be like hope we didn't drop the ball on anything. Just want to check on how your home search is going. It will as a series of drips that are more appropriate for somebody that it was speaking with.
And then something else it reacts it reacts to behavioral messages as well. So it's important to have behavioral messages turned on because If, if somebody is triggering a behavioral message, like they're favoriting a lot of properties, or they're back on the site after a 30-day absence, it triggers the behavioral text. And then if the lead response to that, I will take over the conversation from there as well. So that's another way to make sure that you're engaging the leads with SMS over time.
Tristan Ahumada: Okay. I like that, man. I like that. And then I have a question here. I want to know how you answer this. Because I thought it was an interesting question.
All right, here we go. Elise elise says let me get to it. I'm a lender partner with a realtor. Is there anything new that would help with this aspect? Here it is. I'm looking at the behavioral messaging and everything seems property specific. Interested in something geared towards, are you pre-approved?
Jeff Walker: It's a, it's, it is a question we ask at registration. Of course, that can change
over time. That's I don't believe it's in our current scripting, but as we are moving more towards a generative solution, things will open up a lot more in terms of flexibility.
Tell tell me. So you're prob are you seeing that information today when you look at the lead details? Because at most leads they indicate if they're pre-approved or not.
Are you only reaching out to the leads that say they're not pre-approved or you, do you still check with, check in with everybody?
Tristan Ahumada: We should be checking with everybody, but let's see what she says.
I'm assigned leads Elise says.
Jeff Walker: That's interesting. I think it it does make me curious to think through as we move forward with this, how we might tailor it more towards if the lead is being assigned to a lender, how the A. I. might change the conversation in a way which is helpful to you.
So that's so that's something that I actually would love to follow up on, I could see I could see a lot of value there for AI to help qualify those leads for you.
Tristan Ahumada: Well what we're going to start seeing...
I answer Elise and and Jeff, here's from my experience on the call center. All right. And I think, Elise, I love that you asked that question for 2 reasons. 1, the pre-approved question, I think we should still lead with. What does the client want 1st. We should really put what we want further way further down the line and us relying on actual real generative AI right? Machine learning. It allows the machine learning to be able to the really to be able to pay attention to the needs of the person who It's trying to serve and a lot of the times that agents and lenders throw in the pre-approval.
Here's the 3 questions that get asked too soon. Here. They are. Are you pre-approved? Are you working with an agent? And when do you want to see the property? They're all done at the wrong time. I trained realtor. com sales for years. And they paraded me all over the United States. My background is telemarketing all telemarketing. We watch this happen from the very beginning.
I think the better question is to say, is this generative? I, Alex. Is it focused on giving the consumer what they want. So that then they hand raise, and then we or AI understand when, at some point, it can start questioning. Hey, what's your situation look like on the lending side?
And I think a lot of the times, because these are nurture, we saw 1 step that Jeff gave us earlier this year, where it said, if you get a 1000 leads today, you'll get 10 of them closing in 3 years. It takes a long time, and I think we force these other leads to go and look for someone else. When we put in the wrong questions at the wrong time, but I also love the question too, because we need to be asking it.
Jeff Walker: Just one of the most powerful questions this that I see CINC AI ask, and this is one that, of course, we train for on the phone as well is, can you tell me more about your current situation? And it's it's open-ended. And I am amazed at how much people will say to answer that question.
They'll be like, oh, here's my credit scores. And here's my family situation. And here's what. You know why I'm mad at my spouse and they'll start telling their life story. And so I think those can be very effective at drawing out what's top of mind for the lead. And not ask a closed-ended question that could result in that conversation dropping off. Because mostly here we're going to build rapport and trust. And the more they feel like, Oh, I've told you my story. It's human nature to then feel that sense of trust.
Tristan Ahumada: That's it, man. They want to know that you care. And that's why I'm excited about generative AI because no matter how many times I train agents to say certain things, it shuts off when they're talking to people.
Jeff Walker: Yeah.
Tristan Ahumada: Yeah And so AI is going to do a better job than most real estate agents. That's what I'm excited about because that's real. So Jeff, this rocks, bro. I love where you're going.
Jeff Walker: We'll have to talk about the, what the, what we see is as far as voice in the future, because that's the, that's the next frontier is like outbound calling and stuff, which opens up new cans of worms, new opportunities new challenges. But we definitely see that in, our industry's future as well.
Tristan Ahumada: Jeff, thanks, man. I had no idea that mid journey had gone live on their website. I literally texted it to my friends. I'm like, guys, I thought it was still on Discord only. So we're just testing it out.
Jeff Walker: So much fun. Yeah. All right. Thanks, Tristan. And thanks everybody for joining.