Why Great AI Tools Get Rejected by Good Teams

Shane Harris, Director of Harris & Wood, gave his team powerful AI tools.

The reaction was immediate.

“They don’t work.”
 “They’re not accurate.”
 “This isn’t how we do things.”

From the outside, it looked like a familiar pattern.

Good team.
Good tools.
Poor result.

So the assumption is obvious: Something must be wrong with the AI.

But when Shane looked closer, that wasn’t what he found.

In this episode, we unpack what actually happened inside the business:

What we believed would happen when AI was introduced
What actually happened when the team started using it
What broke in the process
What Shane changed
And why that shift led to over £50,000 of new business in just over 30 days

The tools didn’t change.

The team didn’t change.

The way the business worked around the tools did.

This is the part most agencies are missing.

AI doesn’t fail quietly.

It exposes how your business currently operates.

And when teams expect certainty from systems designed to operate on probability, rejection is almost guaranteed.

If your team is resisting AI…
If the tools feel inconsistent…
If adoption feels harder than it should…

This episode explains why that’s happening—and what it really means.

Transcript

SPEAKER_01 0:02

It's the AI operating system diaries. I'm not going to just download the digital AI impact on business. For the past few years, we've been building AI systems first agency by transforming our own company into an AI first business. This podcast documents what actually happened during that transition. The resistance, the mistakes, and the breakthroughs that change how businesses work. Okay. I wanted to do a podcast today with an actual agent who is experiencing some of the stuff that I've been talking about with artificial intelligence, both in terms of productivity, but also in terms of the challenges that you face with the team and internally with your own sort of thoughts on it. So this is the first time I've had a guest on the AI operating system podcast. It's Shane Harris from an estate agency called Harrison Wood, very successful estate agents in the Essex area. Shane, welcome to the show. Thanks for having me. Yeah. So I want to just jump straight into some of the sort of thoughts and questions because a lot of the people who listen to this show will know you. You're in a WhatsApp group with all of our clients, and we started sort of people started picking your brains on some stuff that you were saying the other day. And I said I'd do this podcast because I thought it might help people. So just a quick uh recap. Can you just remind uh the listeners, or if they haven't, if they're not in that group, like what it was that you were talking about the other day?

SPEAKER_00 1:39

Yeah, so we had a uh, in fact, you and I did had a bit of a back and forth on predict, which is uh AI scouts in the life cycle system and how effective they were or weren't really. I was being told they were more the latter than the former. Uh, and obviously you and I sat down, we spoke about it, and both really couldn't get to the viewpoint of like, well, why aren't they working? Like, what is it specifically that's not working? So a bit of back and forth, and uh figured out that it was more the user case. But you know, what I put in the chat the other day was about how we had found almost £20,000 worth of listings in a day from just using the system in a more effective way. And that was one office and one use case. And I could have put in a few more in the same chat about other offices. Uh, new starts. I think I spoke to you personally, Mark, on new start offices that have had they've got very little uh foothold in that marketplace that they're in because we've operated in it gently, but not really enough to generate any real brand awareness, and it's still working in new locations, which for me was the biggest eye-opener. Um, and so yeah, I just wanted to share it in that group chat.

SPEAKER_01 2:53

So, like let's just try and figure it out then because it's still a little bit baffling, really, isn't it? Like, um, and we're still in the midst of trying to figure out because the next AI thing will come along and we'll have the same problem, right? So um on the face of it, we've got this uh system in place whereby obviously, you know, maybe someone's listening to this who isn't even an estate agent. So, you know, from an estate agent's point of view, they're trying to get uh as many houses on the market as they can. So we've got this system in place whereby it's looking across all the properties that are being advertised on the internet, bringing in the full addresses of those properties, matching up with any contact details that you might have uh gathered over the years that match against that address so that we know who to contact. It's looking at how many buyers you've got that match against that property, how much how many properties you've sold nearby, creating the pitch uh via for letter, email, WhatsApp to send to the person, or just for the person to read to get an idea as to what to say to them. So on the face of it, it's kind of like a no-brainer, right? Even when I first announced it at the events and stuff, you could see people are like, oh my god, this is gonna be like utopia. And you give it to people and uh they go, that just don't work. That's what people would say. Uh and so you when we spoke about it, you you're not obviously on the tools every day, but your team had said to you, like, it just doesn't work. And we then, as you said, like we sort of sat there scratching our heads going, I don't get it. I mean, I I feel like if we just sat down now and did this for the next month, that was our only job, we're gonna come away with loads of listings. And I had the same conversation with another agent. Actually, I've had it with quite a few. One of them um has got quite a lot of branches. Um and we went to talk to them about it, and they were like, Yeah, it's just the whole time that we've been using it, which is probably now well it's gotta be like six months, maybe five months, the whole time that we've been using it, they were like, you know, we've only we've only got two listings out of it, so we're not really sure it works. Now bear in mind they're covering a big area, and and I was like, that's really interesting. And then I was trying to get to the bottom of it when I was talking to them, and yeah, we were they were saying, Oh, a few things didn't match up, and this was not right, and that was not right. It was like, yeah, I kind of see that, but it's not a big deal. The things they were saying to me were not a big deal. For instance, like maybe maybe it matched uh the address with an old tenant, and they were like, you know, obviously, like that's not the owner because they're just a tenant. I was like, I know, but you don't really need me to tell you that, do you? Like, you know, it's pretty obvious. Um, and they would show me things like, oh, look at this one, for instance, this one says the property is available, but when I look on right movie, it's SSTC. I'd be like, Yeah, I know, but so what? Um, and then eventually we sort of got to the stage where I was like, hold on, let's just have a look at this properly, right? How many leads have you guys worked on? How many have you closed over the last five months? And we looked in this at their system then and they'd they'd closed over a thousand, it was about 1100 and something. And I was like, okay, and how many you got open at the moment? It was like 250. And how many have you not assigned to anyone? It was about another 250. I was like, so let's just think about this logically. That's about 1,500 uh people that have been on the market for more than eight weeks was the criteria. Uh it matched the content details, uh, and it's giving you the reasons as to why they should switch. And you're saying, yeah, but some of them weren't right. Let's even say like 50% of them weren't right, which would be outrageous. Like it was probably more like 10%. It's like 750. And you're telling me you got two listings out of it, and they were like, Yeah, I was like, I I don't know how else to say this other than just that you're shit. I mean, I don't know what else to say. I'm not trying to be nasty. Yeah, and I had another agent that did a similar thing. They sent me this spreadsheet of 43 listings they'd been working on over the time they'd had uh the scouts, and they'd they'd filled it out these listings with like, you know, what they'd done, left message, sent letter, done this, like, you know, one action for each one of these 43 items and said, like, it doesn't work, got didn't get any listings. Like, what are you not getting about this? Like, the job isn't isn't leaving messages, the job is getting listings, right? All we're doing is supplying you with a bit of easier information, but your your job is to get listings. Coming to me and saying, Oh, you gave me 43 and I got none of them, so that's your fault, Mark. It's like so so how did that translate in real life? Because you know, we both felt like that when we looked at the system, but I didn't get to go back and then you know implement it. Like, what happened when you went back and spoke to the team and had that conversation? How did it evolve into something that's now working?

SPEAKER_00 8:08

Well, I think I think one of the biggest things that we realized, and I think what you just said there is the expectation of what the system should do. And I think there was a really big misconception around well, this thing will get contracts signed. Like in the end, I think people were under the impression that these scouts are so good that all we need to do is press go, and then it will come back and say, Mark Burgess is ready to put pen to paper, just send in the contract. Yeah, and the expectations were wildly uh over optimistic because uh we didn't set clear expectations from the start of probably what it should do. So I think once we realigned the fact of what you just said there is like, look, it isn't there to do all of the work for you, it's there to just point you in the right direction of here are the people you should be doing the work for. Because I said to the guys, like, if it did all of that stuff, you'd probably be redundant then, wouldn't you? Because we don't need you if all that's gonna happen is we go, yeah, send contracts out. We'll just pay someone 70 grand a year to press that button. Unbelievable. We could just probably automate that being sent, and then we don't need anybody. Like if it's as easy as that to win listings, it's crazy. So it was just a bit of a mindset shift from the guys as to what the expectation was from it, but it boiled down to, I think I spoke about it in the chat the other day, it was really like I wasn't particularly clear with the team on what they should be doing and what they should expect the outcome to be. And when we defined those parameters and that boundary, which was similar to what you just saying there, our marketplace really is six to ten week contracts, and if we're farming stuff, because there's different elements to predict, which I'll let you talk about how that works because you'll know it a lot more intimately than me. But if you're farming things and you're fishing in a pond where there are not that many fish, well then of course don't expect to catch anything. But if we're going six to ten weeks, we have the contact details for these people. So when you've got the contact details for these people, for anybody who's not on lifecycle, it's people who have actively engaged with you for one reason or another. So they want to work with you. So we already know who they are, whether it's a viewing, whether they've had evaluation from us in the past, whether they're a land or whatever the scenario is, they have given us their information. We now know that they're on the market to sell, and we have their contact information. And so six to ten weeks is the optimum time frame in our market. They're probably going to get pissed off around four to six, be really annoyed at 10, and then anything outside of that is just like, well, is it still on the market? Has the agent just left it on for their own market share, et cetera, et cetera? So let's just go in that window, six to ten weeks, because we know our marketplace moves in that time frame, and speak to all of the people we have contact details for. And you're right, if you can't convert those people into listings, then we just have to hang our gloves up now because we're not good enough. But that is the job. The job is sales. And so if we can't convert those people who will be frustrated at that point, then there isn't really much further we should go because it's it's really is just food on the plate. Yeah.

SPEAKER_01 11:35

And then and then and the team agreed, and what just if that was it, they just then suddenly started banging in listings.

SPEAKER_00 11:41

It was literally as simple as that. Yeah, I said to the guys, like, imagine, because we use this AI word all of the time, yeah. So just imagine that this was a member of staff. Everybody's recruited one new member of staff in their office, and then you go to that member of staff and say, right, can you win me some instructions today? And they come back with nothing, we wouldn't go, oh, you're sacked. See you later. Like, we just wouldn't do that because we'd go, oh, well, obviously they haven't won us any instructions because we probably need to train them. And if we trained them and showed them what to do and how to do it and where to find instructions, we'd have a much better chance of winning instructions. And so it was just a mindset shift on well, what are we doing? How are we doing it? And think about it as a this isn't a magic tool of like it's just gonna win you business and send contracts out because we don't need anybody for that, but it's enough that says here are unbelievably qualified leads with information that we wouldn't be able to just be aware of all of this stuff at all times, and it makes it a lot simpler for you, but you still need to do your job, which is the sales part. That's what you were here to do in the first place. So those two things just made the world of difference. And I wish there was a more like a like a bigger aha moment for people, but it's really was as simple as that.

SPEAKER_01 12:59

No, I think it is a big aha moment, and I think you continually have it. Like today, I was looking at um the contact intelligence in life cycle. So, again, for anyone who doesn't not familiar with that, um the AI looks across everything that the contacts done in terms of their web visits, their interactions, the notes, the emails they've looked at, or anything that they've done with the company, it summarizes it uh into an overview so that when you land on that person, you know where you are with them, and it summarizes like the next best actions to take with them. And I was looking at the contact intelligence today, um, and I was getting a bit annoyed with it because you know, it was saying certain things, and I was thinking, if I dig into this really deeply, that's not really true. Like, you know, it was saying, like, we use lifecycle, it was saying this person has been highly engaged with all your stuff, like they've done this, they've done this, they've done that. And I looked and like, yeah, it kind of depends on what you call highly engaged. You know, they've been sent a lot of emails, they've opened a couple of them, like, not sure that's highly engaged. And I sort of started having a go at my developers about it. But it's really no different to what was happening with your stuff. Like, if I was an agent saying that to back to me, I'd be going, get over it. Like, yeah, you've you've had a look on the timeline, you don't agree with it, move on. Like, what's the big deal? And so I think that's the thing. It's like we're just going to constantly keep finding this with AI because our systems are not perfect. Like uh, the fact that I don't know, uh, predict is matching up an address with somebody that no longer lives there because the humans haven't updated the address isn't Predict's fault, but we just accept it. We just go, like, yeah, because we're not robots, like we didn't update the person's address, but it's not a big deal. We phoned them, they say, I don't live there anymore. We moved on, you know. And did you update the address then? No. Well, that you probably should do, because otherwise it's gonna happen again. But it's just that that the idea of like this, I think we've all been led to kind of believe like there's this incredible AI that thinks for itself that will just figure this out. It's like, no, no, that's not really how it happens. I you it just it's just reading the information that's already there and speeding that process up for you. Like you said, like if I said to you, well, just I don't know, just sit and look across the internet at what's on the market and then try and find those people in lifecycle, see what you can find, see what buyers we've got that match, see what we've sold nearby, and see how you get on with that. You'd be like, that's just gonna take me forever. It's like, okay, well, I'll just speed that bit up for you then. But it might be a bit wrong, but it would be mostly right. And then it's like, oh god. Like if you'd actually been doing that other job, I guess you'd immediately see the uh the huge benefit of it. But when someone just brings you the finished version and goes, look at that, yeah, you know, like ChatGPT is still mind-blowingly magical, but we're kind of over it, aren't we? It's like, oh, it writes me the wrong thing sometimes, like, you know, how stupid.

SPEAKER_00 16:08

Well, I think it's probably what's worthwhile remembering for a lot of people is like it's already good, but this is the worst it's ever going to be. And but if it works now, if you consider any rate of improvement, even the smallest rate of improvement, think about how much further on we'll be in 12 to 18 months' time when it's just better. At the moment, there is that whole, yeah, well, it's it's not magic enough. I mean, but it's still magic, it's just not magic enough because we're so used to it, and yeah, the odd thing comes in wrong. But like, of course, the odd thing comes in wrong. Like the person does things wrong at work, but you don't just sack everybody every time they get the odd thing wrong because it's mostly right, but it it will get better from where we are now.

SPEAKER_01 16:55

If you was I'm just trying to think how you how do you quantify it back, like since you went back and had that conversation with the team and they started using it properly. Yeah. Um have they won many listings, do you think? Yeah, loads, yeah.

SPEAKER_00 17:13

I reckon probably at least 50 grand worth.

SPEAKER_01 17:17

Wow. I mean, and that's we're talking like what is a few weeks, yeah. It's not at work. Just over a month, maybe. Yeah, probably. That's pretty scary, really, isn't it? That's the sort of scary that I was talking about when I launched the the product, wasn't it?

SPEAKER_00 17:32

And that was the sort of thing that I'm bearing by the way, like they are they are sorry to cut you off, but they are they're the wins that we're aware about. So in predict, there's the well, you get all the data analytics, right? So like here are the amount of instructions you have won, which is they're all the ones where we have controlled, but we've won more listings than that based on the Predict data, but they might just be it, might just be serendipitous that they've managed to find their way to us, which is look, we did give you all of this information weeks and weeks and weeks and weeks ago. If they've just fortunately called up and arranged a viewing or booked evaluation on their own or whatever the scenario is, well, you've won that as well, but yeah, these are the ones that we put on the radar.

SPEAKER_01 18:17

Yeah, it's it's it's it's an important point because that's the way the stats work in predict as well, isn't it? In predict, it says, I told you about this before it became an instruction. Whether you did anything about it or not is is up to you. Yeah, um, and you know, people don't like that obviously because they don't necessarily agree with it, don't want to be made to look silly, but it's just the truth. Yeah, if you did have enough manpower to have jumped on all of those people when they came in, like you know, they did end up leading to instruction, but there's probably more in there that you haven't yet spoken to, right?

SPEAKER_00 18:48

We'd be giving it a lot more credit, wouldn't we? We'd be going, Look how amazing this is because every the second it came in, I jumped on it and we were over. Like, yeah, it's there's there's too much, of course, for everybody to jump on at all times. But if you were to look at the data and go, Well, yeah, these are the results, you can't really argue with it, unless, of course, there's a really unique circumstance that I'm missing where it just really isn't working.

SPEAKER_01 19:11

But so then I guess like if there's 50 grand worth of listings in in uh just over a month or so, does it does it change thinking on internally about what we're gonna do about this? I mean, like, you must you must still now feel like, well, how much I mean this is a gold rush, right? We're the only ones fucking on this. Like, what should we do about it?

SPEAKER_00 19:34

Yeah, so it's a it's it's one of those that we raise the point. It's I've gone for it with you personally in our own sessions, but it raises a point of like, yes, it would be great to have more listings, but can we can we deliver on the service that we need to deliver on because there's a breaking point where we get oversubscribed. So the listings if they're not selling and all that sort of stuff, right? Yes, exactly. So there is an element of like, yeah, we could really milk it, but do we need to? But there's certainly a conversation at the moment of like, well, this could be a full-time department, but there would just need to be other things in place to make sure that if we really if we improved our listing result by 15, 20%, then we'd need to improve our experience by 15, 20%. So it takes more than the than than just oh, as simple as put more house on the market and we'll be in a better place.

SPEAKER_01 20:24

Yeah. Yeah, it's probably not it's probably not as narrow as you know just thinking about predict and winning existing listings. It's just a wider, it's a wider conversation, isn't it, about like, you know, as the AI starts to be able to predict other things inside your system, you know, if it's bringing to the surface the problems in sales progression, if it's bringing to the problems like, you know, frustrating phone calls that are happening across a network of branches that can be highlighted or you know, any of that sort of stuff. Uh predict is just maybe allowing us to see what what the new world of work might be like. Yes, exactly. AR is not gonna AR is not gonna do it for you, it's not gonna do the sales progression for you, but it's going to allow you to progress a hundred sounds through to completion much faster than if you'd have just tried to stay on top of those hundred sowers yourself.

SPEAKER_00 21:18

Yeah, you're in a really unique position, I think, Mark, because obviously you're in this stuff all day, every day. And so you've got a different mindset on it than almost everybody else because you're so close to it that you understand it deeply. And I think that's where a lot of people go wrong, myself included, is the the lack of understanding of what the capability are, uh capability is and what that means for your business, because it it really can make such a huge difference, but you need to deeply understand it yourself first. And I keep putting out, you know, I don't know if you've seen them or not, but I keep putting out on LinkedIn one of my guys said to me today, Oh, you didn't put um knock with AI on. This post on your last post. And it sounds counterintuitive, but the only reason I don't use AI to write my posts is because I know I'm going to need to use AI for almost everything else in my business. And if I can't distill my own thought process, I believe the AI is only going to be as good as the thought in the first place in order to help me execute on how and what we should be doing in the company. And if I can't figure that out in my mind, it's no good me just relying on AI to figure it out for me. It's a commodity, then, and anybody can do that. I need to be able to really distill my thoughts and think, well, ah, okay, well, if it does that and we do this, I could probably get it to do that for me, which is going to help our business move forward exponentially. So you've got to spend more time in learning and understanding it instead of just. I think AI's come along and it's like, oh great. Um Laura got promoted. Write me a post about how pleased I am for Laura to be promoted. Yeah. And it's like I've gotten, I'm not saying I'm not bashing using AI. I use AI everywhere else pretty much in my life all day, every day. But for clarity of thought and distillation of an idea, I don't think it's helpful that we keep farming that off to chat.

SPEAKER_01 23:15

Yeah, no, I I completely agree. I mean, like, I don't have any problem with uh making writing posts that were written by AI. I wouldn't even try to hide it. But the posts that are written by AI, I'm happy with because I dictated to AI what it was I was trying to say. It just did the the labour of writing it for me. I didn't say uh write me a post about wishing Shane a happy birthday. I spoke to AI about like, you know, and this is Shane, this is me, this is our relationship, this is what we do, this is what I want to wish him a happy birthday about this and this and this, can you just make me a short LinkedIn post about it? That I don't have any problem with. But like you say, you've got to be able to, you've got to be able to have the thought in your own mind because otherwise it's just generic bullshit that like, you know, what's the point of this, you know?

SPEAKER_00 23:59

And you can say you use AI every day, so you can like you can open a post up in a minute and go a second, first second, you'd be like, Yep, chat GPT like that. Or it's disingenuous. There's nothing that I don't see from anywhere else, and it's hard to really start to make yourself appear any different in a marketplace where it's all such a commodity like that.

SPEAKER_01 24:22

So, all right, that's brilliant, Shane. Look, I'll probably get you back on the podcast because there'll be more to talk about as we go forward. But um, I hope that's useful to agents and I really appreciate you know you sharing that with us.

SPEAKER_00 24:33

No, thank you very much for having me, Mark.