Heads In Beds Show - AI Update: Q1 of 2026: OpenClaw, "Prompt Tracking" Useless?, Current AI Tools We're Using

Episode Date: February 11, 2026

In this episode Conrad and Paul give their regular update on AI: what tools they're using, what's changing, and how you can use it to grow and scale your vacation rental business. They talk ClawdBot (aka OpenClaw), traffic from LLMs going down month over month, prompt tracking value (it's useless?!), and what tools they're using MORE and LESS of... Enjoy!⭐️ Links & Show NotesPaul Manzey Conrad O'ConnellOpenClawNEW Research: AIs are highly inconsistent when recommending brands or products; marketers should take care when tracking AI visibilityChatGPT vs Google Traffic DataConrad's Book: Mastering Vacation Rental MarketingConrad's Course: Mastering Vacation Rental Marketing 101🔗 Connect With BuildUp BookingsWebsiteBook A Call With Us🚀 About BuildUp BookingsBuildUp Bookings is a team of creative, problem solvers made to drive you more traffic, direct bookings and results for your accommodations brand. Reach out to us for help on search, social and email marketing for your vacation rental brand.

Welcome to the Heads of Med Show presented by Buildup Bookings. We teach you how to get more vacational properties, earn more revenue per property, master marketing, and increase your occupancy. Take your vacation rental marketing game to the next level by listening in. I'm your co-host Conrad. And I'm your co-host, Paul. Paul, good afternoon. What's going on?

What's happening? Well, you know, we've been talking about a little bit. We're recording this on Trade Deadline Day. Your team has made a trade. There's a lot of chatter about my team potentially making a real. really big splash. But we'll see how the day kind of plays out here.

Like a deer size? I've heard deer sighting. I mean, it's going to work a lot. I don't know if I want to give up what it's going to take to get a deer sighting group. But it would be intriguing at least that. The one thing that Timberwolves, the new Timberville's leadership has not been afraid to do is splash around at this time of year.

So it's refreshing as a Minnesota sports fan for anybody who is from those listens or just commiserates with me. through the podcast over the last couple years. It's not fun. So when we at least give it an effort and a try, hey, I'm on board. But how are you doing, sir? Pretty good. So you mentioned that we're recording this specifically Wednesday, February 4th.

By the time it comes out, you'll know the winner of the Super Bowl. At this point, we do not know the winner of the Super Bowl. So it'll be interesting to see. Very interesting. So always sport stuff we got going on. Always things that we got our eye on. And of course, we are a pro-patriots podcast when. I would say, as all of you know, Conrad is clearly cheering for St.

Arnold and the Seattle Seahawks. Yeah, yeah. No, I suppose not. But yeah, we are. We will be rooting. I made my prediction. I think the Patriots are going to surprise some people and at the very least they're going to cover. So I think that's, uh, you put, you put five bucks on.

We'll say, yeah, we'll see. It, uh, it feels possible at the same token. You know, I think the smart money would say, you know, you go ahead and you bet the Seahawks if you're just betting on the winner. but yeah, that's what the whole points are for if you want to get into that world. Totally different world. Totally different world. So it's funny, we hit record here, Paul,

and as soon as I was about to record a moment ago, we had a power glitch. I was thinking, like, what are some things that AI becomes useless when? It's like, oh, when your power cuts out, the AI is kind of useless at that point, right? Because it's a browser call, you know, to an external service or whatever the case,

maybe API call. But anyways, hopefully the power will hold up now. It was weird. We didn't lose power when there was five inches of snow on the ground over the weekend, which is very atypical for us for those that don't know. I live in South Carolina. I was going to say, not Paul.

This is counterhead talking. Yeah, I got five. And then I got power cut now when like the snow's all melted and gone. I don't know. Maybe the snow melting caused the power outage because it like hit something in some transformer blue. Who knows? But anyways, today's kind of a fun one.

Some of conversational, a bit of, I think we're just going to do these like every six months. We talked about this. Yeah. We're just going to do one of these every six months. And it's kind of like the AI updates. This is our AI update. Obviously, you know, February, but, you know, the very beginning here of 2026.

And let's do an, another one kind of midsummer, maybe even we do one more at the end of the year. I don't know. We'll have some fun with it. We'll do another one in a little bit. It's been a little while since we've done it. And I've got a lot of AI things to say, you know, including some new stuff that's just come out, including some stuff that I'm maybe changing my mind on a little bit. You know, what's your current LLM stack, stuff like that. I want to talk about this prompt tracking article that I put in the notes as well from the SparkToro research from Green Fishcan and the guys over at Gumshue. That's pretty interesting. Just philosophical questions about AI who owns it. And then just to check in on,

again, Google's dying. Remember, that's been the narrative now for about a year. and gone and all these people are using AI search. We'll get an update on that because it's valuable to get some more context there. But I guess, you know, level set with me now. You know, so you just wrapped up January. What was your January like AI workflow look like for your day-to-day responsibilities and tasks? What are you using AI for? Why do we even need this stuff?

Why are we talking about it? What I'm using it for primarily is getting additional insights on content, data, assets. So looking through emails and doing some things like that. for me it's it's organization because that's that's that's honestly being my setup that's that's where I need the most assistance is just making sure I've got eyes on everything because you know I don't want to be leaving anybody in the dark there so I think that is where I don't think I've gone crazy with it and you know I'm not I'm not using I wasn't using Claude greatly and still I'm not using it very much um but there are I think

my, and it continues to be my limitation. It's not my hesitation, it's my limitation is, for whatever reason, my brain can't think of cool ways to use AI. Like, I'm sure if I, when I'm presented with like options and demos and I try to go through things and try these trials and do things like that, but when it actually comes to brass hacks of what am I going to do, like doing a project, doing something using the systems and tools, I haven't been able to find a whole lot of stuff that sticks. So it is. I mean, my tech stack is filled with dozens of 14-day trials or lifetime deals and things

like that, but I'm testing. But realistically, mostly what I do is I use Gemini, I use chat GPT. I use, and chat GPT is very limited there. I use Gemini a lot. I think just the multifaceted nature of it. The fact that they're trying to go multi-modal is not exactly what I'm looking for, but they like to tell me that. That's what they're going for. So that's fun. But I do. I mean, as someone who has been Google, Google, Google kind of the whole time, I still find it to be the most resourceful. I still find it to fit my needs. And I think that's the thing. I'm trying to find areas where I can utilize it more because I think it can have a positive impact and what I'm doing, I just really struggle to put those pieces together of tool, task,

tool, task, and kind of combining those all. But I know you do, you've been using things to create, to organize, you know, where have you found it most effective? Or have you found, have you found it coming up short in any areas of, you know, where you're wanting it to deliver there? Yeah, I mean, I think so. I mean, there's so much green grass in front of us. I think we have to keep working on and improving with some of these tools and how they fit into our workflow. So I guess if I just go front to back of like, I'm looking at my browser tabs here and like what I spend time in. I made the switch a while ago.

I don't know if we even talked about this in a previous one to superhuman. So superhuman is basically a layer on top of Gmail. So Gmail is like my typical email. So you email me, Conrad at Buildup Buckings.com. Technically it's just Gmail, like the workspace product that we have. But within that, I did layer on superhuman a little while ago. And I found that to be really solid. It just would like, it sounds simple, but just it writes a draft for you.

and then you can edit it basically is what it does. So like you open up your email and there's a draft sitting there already that you can then, based on the context of the conversation and then you can then modify it from there. And there's a bunch of other features like reminders and, you know, come back later type stuff so you can like send it and kind of get rid of it, archive it,

keep the inbox a little bit cleaner. I'm losing my email battle right now, but that's not a software function. That is just the nature of being busy and, you know, getting other things done. By the way, I have this philosophy.

I tell this to my team and hopefully I'm not outing myself too bad for a listener. But I have three things in a day. I have meetings. I have to-dos like projects and basis. camp and then I have email and I can pick two of those things that I can work on to any level of effectiveness. So if I have a bunch of meetings, the email might be getting ignored. If I've not a lot of meetings, email is probably in decent shape. But then it's like, hey, me with anybody. So, you know, it's just funny to think about those three things. It's like, and I'm like thinking

like, oh, this will get better as we get more structured to built out. And I'm like, no, it's kind of been this way for like five years now. Like, it's just I get a little backlog, I work through it and then I'm on the next backlog or I'm on the next thing I got to work on and then go from there. So do I think AI is helping me? Absolutely. Is it, I think if I think about my AI use cases now is just where we're sitting right now in the beginning of 2026. It feels like a lot of software that I was into back in like college and back when I started my career, which is like these little hacks, these little optimization things that I use it for is kind of how I would feel. So for example, I've been a text expander user since like 2010, maybe 2011, which is you type a snippet and then you, it expands a snippet into a longer bit of text.

So for example, like when I do comma, that's my delimiter, and then the word kickoff that writes a kickoff email for a client when I sign them. So is that some amazing, life-saving thing that's going to save me hours and hours of time? No, you could just put that in Apple notes or you can put that in, you know, base camp or is what the PM system we use. You could just put it in notepad, open it, copy and paste them, put it in there. But it's a 20 second, 30 second, one minute, two minute, three-minute, save type task that I find stack a dozen or two dozen those things on your day. And you get so much more done than other people. Like, that's kind of been my experience in my career for the most part, is that, and I feel like AI is just accelerating that mindset for me.

Anyways, to go back to your original question software. So superhuman is like what I use for email now. But if you like snap me fingers and said, oh, you have to go back to using plain Gmail, I'd be fine. Like it would take me a little bit of a learning curve to go back to it because now I'm kind of used to the system I'm using today. But I like the fact that I have AI auto drafts in my voice too. They're not awkwardly written. Like I find some of these auto Gemini written emails are just like a little bit awkward. The chat GPT copy has gotten so bad lately.

Like they keep tweaking the model and I could literally just look at a tweet now or LinkedIn post and be like, oh, that's chat GPT. Like it has such a flavor to it that it's so obvious to me that people use chat TBT to write and I'm not saying don't do that but I'm just saying like be aware that like it's very obvious what you're doing. So if you want to make your copy a little bit more unique, probably don't just copy and paste straight out of chat chabit, probably not a good idea. We'll get to this later. But quick answer is I'm still, I feel what's the word polyamorous between chat TBT and I'm

still dating both of them, which I feel like I can do. But like gun to my head right now. What's the one for it pinned in my browser tab right now? it's Gemini. So like, but I find myself opening chat chabit and still using it. But Gemini has kind of been the one that I've been leaning more into over the last like, I would say two months or so. Ever since that three pro came out, I don't know when that was. I don't know if we can look that put in the show notes, but three pro when that came out, I kind of started testing it. It was like, I think I like this better. So that's for sure one thing. And then we have to do

an honorable mention for the new Claudebot, aka OpenClaw system. Yeah, I don't, I don't, I don't know if I have enough to share here because it's so new. You know, again, I'm recording this beginning of February. It's, I've kind of had it now, I think, for two weeks. But the idea of for those that haven't used this system. So open claw formerly known as mold bot, formerly known as Claudebot was like how it got popular it initially is essentially allows you to have a system control your computer on your behalf.

So like you could tell the system to go in there and like go to Google and research 10 different property management companies, go to their landing pages, tell me all their phone numbers. And like it'll do that for you. You can and then there's a slew of APIs and you know, agent style capabilities that you can do with it. So you can basically ask a. to accomplish something and it will try to actually accomplish it for you. Unlike most of these

interfaces that we're talking about, like chat, GPT, Gemini, etc. They're pretty limited to like whatever your pay whatever context, whatever content you're giving it. And then sure, it has some level of like internet search capability. Open claw or or cloudbot is so much more powerful in advance than that because it literally controls your computer, which yes, is a huge security risk and risk and you should be aware of those things. But anyway, is more to more to kind of say there. But what I'm using that for right now as far as practically is it's doing kind of, I think it seem like you touch on this as well, the organization elements. So like, for example, I, in the, there's, there's basically jobs that open clock and run on like a daily basis or

or weekly basis or whatever. But I want to sound simple where it's like, go in every day at 7 p.m. and tell me all the meetings that I have the next day and then pull the last 30 days of emails with those particular people and give me a three minute or a three bullet point, excuse me, summary of what I talked about most recently with those people. That sounds simple, but like just going in, pulling those three things, pulling them into that, um, that little, um, like summary kind of executive summary style message I get every day at 7 p.m. It allows me to flag myself and be like, do I have a call tomorrow at 9 that I'm not prepared for? Because yes, that does happen. I'm like, I said, you're in prison and it doesn't happen.

And I can be like, oh, yeah, I do have that call tomorrow. What things would I need to maybe prep either in the evening or in the morning of to make sure that I'm hopefully in decent shape for my responsibility for that particular recliner project? So hopefully I'm not giving too much secret sauce away about how the sausage is made. But sometimes, sometimes, you know, I will admit my college habits sometimes, you know, bite me hard. Or it's just like, oh, I need an update on that. Let me go talk to the team member on my team that is. working on that project. So when I get on the call, I'm informed about where that website is at or where we're at on the ads or whatever the case may be. So that's like one specific example.

I could do a dozen more, but that's kind of the AI stuff that I'm playing with right now and kind of where I see this smart agent assistant direction going in is kind of a system having access to your accounts, your authentication, your systems that you use to manage your day, manage your work, etc. And then what's unique, I think, about OpenCla or CloudBoc compared to all these other AI tools that we've mentioned so far is an element of productivity is that it's going out. You can tell it Every day, it's 7 p.m. do this for me. Or if you see a problem in this message, look at it every hour, and then flag it

for me if it's a problem, that sort of thing. So that's what's unique about that system, is how the, and it's very simple to control. You're just typing up what you want it to do, hitting enter, and then it's doing this sort of agentic things behind the scenes. I'm very bullish on that. But yeah, that's kind of where I'm out right now.

Hopefully that all makes sense. Ultimately, there's a lot of power there. Again, because it is, we're still in that system area, time of, or piecing together solutions. There's not like one. Even, I mean, even the chat GPT initial offering was little GPs and a big GPD. And then the big chapities.

We're all tying all of these AI elements in together. So I've talked to, you know, quite a few people about what they use. And I don't think I've ever run into two back-to-back conversations or any conversations that everybody's using the same stack or everybody's using the same tools. Everybody's using the same systems. And that is one thing that I think. think is maybe more unique to the AILM models, tools, whatever we've got here is that there is so much, there's so many new technology.

There's so many new people who are trying to uncover like that all in one solution. And I do, I think it's allowed for ingenuity past just a big three, a big five, a big 10. And you do have like, I think Cloud's a great example of that. That is where it started just as this over here, the CloudBot. And now you've got an open source system that can help you run business, life, whatever you want, and out of your own machine there. That's the fascinating thing I see on Twitter, by the way, is people are using it for a personal use cases. I don't find myself that, I don't know, I don't feel like that's my constraint in life.

Maybe I'm just oversharing, again, personal information here. But I don't feel like people have talked about you can actually create a chat with your agent, like your OpenClaw, claw bot agent. and you can send it things and then ask it to do stuff on your behalf, again, once it has the right access. I think people keep skipping over that on Twitter, by the way. They're like, oh, do this and it'll control everything. And I'm like, you realize this is like two days of set up to give it access to everything. And you should be.

Is that way? Yeah, I don't know if you run into issues there. But anyways, some of these people are the use cases they have for it. Like I saw an example this morning yesterday on Twitter where someone took a picture. They're like, I'm out of this. I'm shampoo. Go buy this on Amazon for me and like, you know, get it. Like, I'm out of this.

Go reorder for me. And like it went to, I guess it did it. Right. like the the claw bot agent or open claw agent went to Amazon. It already been authenticated. Previously, re-orders shampoo. It's going to show up the next day. And I'm like, okay, that's kind of cool, I guess.

I could see a future for that maybe. But I'm like, I don't know. I don't feel like me ordering shampoo is like my constraint life. Or I don't think that's what AI is really going to help me make a better life. But again, people have different use cases for this stuff. So I think that's why we're learning. This is maybe the downside to AI and the technology is that we're like creating use cases for AI instead of solving the use cases that we currently

like for the problems that we currently have and that's great. I'm not saying that you couldn't do that same thing probably take a picture within the Amazon app. Get it. Yeah, like how much time would it take me to open Amazon search for the product? You know what I mean? Like, I don't know. Like I'm not saying that it wouldn't save you time. I'm just saying

like is that 17 seconds valuable? Whereas like me going and pulling, let's say at five meetings in a day to go back to my project briefing tasks that I have. I have five meetings. So I have to go pull five emails, go back 30 days, look at those threads, read them all, come back to it. Like, that's like 20, 30 minutes easily, if not longer, depending on that.

And then honestly, my email and then I, oh, something else alerts me. Like, I know, like, I'll get ADHD. You know, and I'll get like on the other arm. So that's part of it too, right? Agent doing stuff for me and then preventing me from going in my email inbox and finding some problem I need to then go to go to it instantly. So like, to me, that's like, wow, the value I see and that is so high. The value in me taking a picture and reordering shampoo, I'm a little bit less like

bullish on that as like the use case wine. But some people are super excited about it. I think that's part of it, too. Any new technology, people get excited by it. oh, this is so cool. You know, what are all the ways that can use that? I think that's part of this one. Early adopters, you've got to love them.

Yeah, yeah. Let's move into this whole conversation of why am I not showing up for L.M. prompts or why am I, why am I not showing up an AI? Or why, I mean, have we seen much movement? And I think this is, we can jump into the Spark Toro article as well, because I think Grand has a great job of talking about this. But, for sure. I mean, let's do a little recap here.

how are you seeing LLM AI traffic move on your end? You know, where is it moving well? Are you seeing big gains? Are you seeing small gains? What does it look like right now? Well, okay, so there's two different layers to that. And I don't know which one you want to start with. Do you want to start on the traffic side or you want to start on like the Sparktor article?

Let's go traffic side and then we'll go into the Sparktoro side. I'm not seeing much meaningfully, meaningful increases in AI traffic, at least not significantly more or a massive increase from what I saw. let's say towards the tail end of last year. Now, again, a lot of reasons for that. But I'm going to actually open this live. I'm going to open up the biggest account that I've access to, analytics-wise, in terms of volume of visitors. I'm not saying it's the biggest customer we have,

but the volume of visitors that I have. And I'm going to pull some of this stuff live because it's been a minute, you know, we did that episode at some point last year. And the data was super clear where it's like, you're going to get a fraction of referrals from these systems compared to Google. And that hasn't really changed much. So now this is not me saying, I want to be clear. It's not me saying people are not using these AI search models,

even to search for vacation rents. I'm not saying that at all. I'm saying that these platforms do not refer much traffic to most brand websites. There's some exceptions, but that's kind of what I'm seeing right now.

When I say not much, I mean less than 1% in almost every single account that I've access to. So that's what I'm seeing. I'm not seeing this massive increase or flow additional value of traffic coming in

from these platforms. What are you seeing from your side? You also have other other data as well. It is. I will say I have seen growth. I have seen increases. But these increases, again, are accounting for even in the best case scenarios. I think the one we talked about before, I think, is 2%.

But even when the winner is 2% as opposed to 1%, yeah, you're doubling it. But we're still talking about a very minuscule portion of the traffic that people are getting. I mean, I made the comparison to it's like the amount of traffic you get from tablet. Are you going to optimize your website for tablet traffic? Probably not. Now, again, this is the different context, different medium, all that stuff. And yes, people are using LLMs to find businesses to do research, to do all these things. But I do think that ultimately the websites that I've seen improve, see that steady improvement on the LLM side, whether that is in Gemini, whether it's chaty VT, some clod, some perplexity. We've seen it across the board. We're doing it. SEO best practices. We're improving the technical SEO or making sure schema looks good. We are writing blog posts that, yes, I will say probably are a little more in tune at answering questions. I think that's something that it's difficult to, you know, and we'll talk about it on the prompt side of things, the prompt tracking and whether or not that's valuable. But I do think

understanding, which doesn't change at all from what SEO is, is understanding those pain points of your homeowners of your travelers and being able to create references, guides, helpful information that demonstrates your authority, expertise, trustworthiness, say, we're just talking about these SEO principles. Those are the ones. So, like, I would say, yes, we are seeing some increases, but not as a result of having tried to do it. It's because we're still continuing to follow the content strategy, the SEO strategy, and it's paying off in dividends. not only on the organic search side, but on the AI and LLM side as well. So that's kind of what I'm seeing.

But I think that's got to be kind of universal. If you're doing well organically, you're probably going to be improving on the, on the LLM side of things as well. Yeah, it's a bit of a guessing, guess and go kind of thing. Guess go, get more data, come back and then see what's happening. So I did pull that data, by the way. So this is a large, they're in the vacational space and the affiliate space. So this is their 90 day, we're 90 day traffic.

So reasonable sample size, down 15% from chat GPT. So 203,000 visitors in the last 90 days. The 90 days prior, 242,000 visitors. So if they get a lot of traffic from chat chit, let's compare it to Google, shall we, loading up right here, 10.7 million in the first 90 days and then 11.3 million in the 90 days prior. So Google traffic up 5%. Chat Chit BT traffic down 20% will be the headline.

Hold on. Let me give you an exact sense of sessions. This is session data. 47 million sessions over the last 90 days. So pretty good sample size, I would say. I would argue in the vacation yeah in the vacational space of traffic and yeah they are seeing a decrease in chat dbt referrals and an increase in Google referrals but again people keep telling us Google's debt so that's the piece that I don't understand to be fair they are running paid ads I want to mention

that so that's all Google referrals including paid so you know that's a whole other layer or nuance to that but I digress so yeah I mean I think again to this idea of like I've said this before I'll say it again people are not replacing in the way that some people would claim their usage of an AI search platform tool with with Google. Like that is clear. They are supplementing. They are augmenting. They are using both platforms or tools depending on the job to be done, which we can have a whole

section about that in a second when we talk about the Sparkora article. Should you be being aware of how the stuff works? Should you be aware of your visibility in these platforms? I think so. Absolutely. I think you should be aware of it. You should talk about it. You should, if with your marketing agency, you can kind of talk about a little bit.

But I think you shouldn't expect this very clean, neat, you know, answer of, oh, we just do these two things and we're good to go. We're perfectly optimized because. again, we keep referencing this article. We'll come to it in a second. Some of the largest brands on literally one of the large companies in world, Apple doesn't even consistently show up in chat TBT. if you ask it, what are the best headphones?

So, and that was the example from the article that we can get into. So, yeah, I mean, I think my current thinking, this is not done. So I might eventually put up like a page in the website about this to this effect. But here's kind of where I'm at right now, maybe a little bit. Authority or kind of like trustworthiness is one angle to it. You have to explicitly show these platforms why you're trustworthy, why you're authoritative, with proof. I've hosted X number of guests.

I have X number of properties. I have X years of experience. I have X, Y, Z reward. Makes me actually think back to your inventory days, Paul, when you were doing those homeowner landing pages, that's social proof for kind of proof concepts. So that in many forms or different formats, etc. That makes a lot of sense to me. Story, what makes, what's your story?

Why should I pick your, you know, particular company out of the many, many out there? So Heather Bears beat this drum for months and months out. probably years at this point, this idea of your aboutus page is really important. Sheo talked about it long before AI showed up about people. People would go to your bodice page and look at it. Very true. And they're still doing that. But also now the bots are going to go to your about us page.

What's the story that you're telling you, your about us page? So like there's authority, their story. And then I'm kind of trying to figure out what that third one is. I feel like it's questions. That's kind of my current thinking. The last thing is, ever on the website, I'm going to have more opportunities for people to get their questions answered. So it's a Q&A block or an FAQ block or something like that. And it's on this property, I have little.

collapsible sections or accordions that I'm going to play with. Or at the bottom of a blog post, I have, you know, that type of thing. So, again, bots can crawl that, find that content and potentially show it in the LLMs. Because people are often going to LMs to ask questions, right? They're not putting a keyword in an LLM, like, you know, chat, GPT, like, you know, score of the Patriots game on Sunday when they beat the Seahawks and win the Super Bowl, right? They're not doing that.

They're doing stuff that's more, like they're asking long form, mostly asking questions, right? That's the type of thing that would return a web response is more of a question, who was the best property manager in ABC location and why, why should I choose them, that sort of thing. So that's kind of where I see it as like questions, authority. And then, you know, I think it's like this, it comes down to these other, you know, pillars around that. So I think we've made almost a full circle in coming from Ask Jeeves. Yeah. Which, I mean, it was. If those were questions and now you're going to search engines, which were long tail, ultimately the same thing.

I mean, we're just not putting a question mark at the end. We're sorting it up and making it easy. and now we're going back to we've made the full circle, full loop and now we are asking for full recommendations and asking for sources and for this, all this stuff. I think it's just, I mean, it's that human nature of

how we, everything comes in trends, everything comes in cycles, and we've cycled back through on the intent-based searches on how we are trying to find the most information. That's the one thing. So that is one thing that As we were and search console, that was something that when Google was trying to push, you know, pushing out, hey, this is where you're going to start to see all these longer tail keywords and all these more AI type of prompts.

I have seen on some of the websites. And it's funny because they're not necessarily the websites that are talking about the prompt tracking moving all around. And all those rankings changing and serving up different answers for every. I have seen as you do some of those REJX formulas within search console, some of those longer tail, clearly more prompt-based searches happening. Now, the one thing, and in AI mode, it kind of builds the prompt for you a little bit. So this is the one area where I'm not sure, and this is where I think prompt tracking, in addition to what Rand is saying, is just, is so kind of convoluted is that because AI

mode kind of helps you create that prompt in Google, and then you start to get some of these longer prompts that are being the output in search console, are these really the searches that are happening, or is that almost like a people also ask question like Google has, or in the searches

and stuff like that, those structured snippets? Are we getting again, is the model feeding the model to a certain extent here and we're not actually getting that true human intent of what's coming through. That is the one, like, that's something that I have. I've seen, especially in, up in the Pacific Northwest, there's, you know, there's definitely a lot of those, I don't know if it's

just the audience up there, but I see a lot more people who do those longer, plan me a trip or give me all the possible recommendations of what I can do in this area. So I don't know if you've seen that on the search console side, like outside of what we're doing everywhere else. Well, it's a good question. I don't know the answer right now with the data that I have in front of me as far as how people's behavior is changing. So let's go to this article, though, because maybe some of the insights from this article, I think are pretty interesting and they bring up, again, this idea of like where people are

at. So we'll put a link in the show notes, but it's at Spark Toro. It's probably going to be the most recent blog post by the time you're listening to this. If not, the title of it is new research. AIs are highly inconsistent when recommending brands or products. Marketers should take care when tracking AI visibility. So that's the title of the article.

I can post it on January 27th, 2026. So I can find that later if you need to. But yeah, the, so, okay, a lot of things to say about it. I think you bring up a good point. People don't use quote-unquote search in AI anywhere near like they use search in Google. So Google, if I told you, hey, Paul, you're coming down this year for my birthday for the summer. You're going to come spend some time during the summer for me. Man, I would love a pair of new headphones my birthday.

That's like the example I think that's given in the article to research there. And you go, yeah, I should buy Connors to Matt phones. He's a nice guy. I'll buy his headphones before I get there for his birthday. So you go on Google and you go, Obrad needs noise-canceling headphones. He's more of an Apple user. He really wants noise-canceling because he works from home

and he has children that are very loud and, you know, during his calls and he wants noise-canceling. And then, who knows what the AI would come back with? Probably AirPods Macs, which funny enough I already have. But then do that same prompt again or give that person a new prompt and say, oh, no, I need to buy some free ROM and she's actually going to split him 50-50 with you, but she's on a budget. Like that's kind of their parameters.

So, okay, I've got to find something nice for my mom. mom. She flies somewhat often to see her grandchildren. So she wants decent always canceling. She's on a budget. So we want to spend more 200. Like you'd, you would massively change the prompt. And that's what users do when they go and use these platforms like OpenAI or Chitbt, Gemini, etc. So that's, that makes it really problematic to say, I'm going to do prompt tracking because what prompt you tracking to your point? Well, like, how are people even putting that together? So, so what the, what the article kind of concludes to get to the kind of end state here is that you basically need to build a persona, like you need to understand who is searching your products

and create some actual realistic logical use cases for how someone would look for it, come up with probably a few variants of that, like, you know, add in some additional specific terms or questions that people would ask into that prompt. And then you basically have to run it like 100 times. And then what you're tracking is not, do I show up one or two? Because during that data set, you will show up one or two a few times. The question is, how often am I mentioned when I do this 100 times? And that's what the product Gumshoe does, which is actually an AI prompt tracking tool

that we have been testing with a few clients. And I like their logic. I think their logic is relatively sound with how they're trying to get this data. So if that's the case, if you have to go into a prompt tracking system, come up with a few different ideas of who's searching, then generate a half dozen prompt kind of ideas,

then run them 100 times to actually get this data out. For 1% of your traffic, may I come back to? Like just for a little of your traffic that's actually coming in, you know, one question is this something that, again, should be taking a lot of your time, effort, energy, and focus. I'm having our time justify this for most of our clients because I just think it's a little bit more of a distraction than anything else.

Is it something we can do one time and then make some changes and come back six months later and do again? I think so. Is it something where you need to run like prompt tracking, 100 prompt tracking requests every single day or every single week and then look at how things are changing?

I suspect that's not the case for 99.9% of the listeners on the other side. I mean, as far as prom tracking goes, I think what you should be understanding is what prompts your competitors show up for and figure out, the content that you need to write to answer whatever prompts those. And I think that that's what it is. I mean, ultimately, you want to know what people are looking for in your area, but that's creating those personas more than anything else, I think. So that does get back to nuts and bolts

marketing 101 type of stuff as knowing your target persona audience, knowing their pain points, knowing their psychographic tendencies and all that stuff, and then building from there. Because what is, when I look at like a seminar, rush or an ATA traps and even the reporting that they're giving, it's nice. I like seeing it. I mean, I have, I kind of take it with a grain of salt because I don't know. They're not, they're not also consistently running thousands and thousands of queries. So they know, they have any better idea.

And honestly, I think Rand's probably got the best tool for actually testing and proving some of the stuff out and understanding, what is that percentage? What are the standard deviations of, not only the percentages, but we're talking about 15 to 20 percent standard deviations between some of these most mentioned brands. I mean, this is not, this is science at a bare minimum of a science definition here. So this is like the data is, and I think that's where even when we were kind of planning this out last week, planning this episode out last week, while a lot of my question still comes down to on AI,

traffic. Like, there's so much unknown out there. It's not even a matter of why this site does versus this site. It's, it does. It's more of a why do people use these AI tools. What answers are they actually trying to find? What answers are they trying to find? I still don't think there is a consistent theme trend across travelers per se. I don't think there's a consistent. theme, like, regionally. There's nothing that we can still quite point to. And maybe we never will be. I mean, maybe that's the, like, search, we can make these assessments.

We can hypothesize. We can do all these things. But AI is that black box, maybe a little more or less, where we don't have the same ability to do that. But, I mean, I'm not sure. That's the wax poetic moment here. But, like, where you kind of sitting on that spectrum there. Yeah, well, I mean, you bring up good points again, which is like, okay,

Knowing that there's some element of randomness to how these things even work in the first place, there's an element of, you know, I would say unpredictability. It's not that it's not predictable, but an element of I don't know exactly what's going to happen. Again, so the example in the article that I referenced a few minutes ago is in some of these headphone prompts, Apple had like 70% coverage or something. So Apple, the multi-trillion dollar, you know, whatever consumer electronics company that's probably sold more headphones than presumably any company in the history of the 6 billion, you know, years we've been on this earth.

Granted headphones have been around for, you know, fraction of that. I'm being now just exaggerating for the take of the point I'm making here. They are not mentioned 100% of time, nor is any brand, by the way. I think the highest was Bose or Senheiser, which was like 90-something percent or whatever the case may be. So, yeah, your best case scenario, if you are, you know, your job at Apple is, hey, make sure we're mentioned LMs by, you know, when people are shopping for AirPods or headphones or whatever the case may be, is. Yeah, we're showing up 70% in time, boss. Like, imagine you go to the CMO of Apple and you say that. I don't even know who it is.

But I just say that. They'd be like, okay, that seems like we could improve that a little bit. So a lot to dig into, you know, with respect to that, you know, how those systems are working. And it's just so early. I mean, we're still, you know, we're still just kind of in the first inning of this stuff. You know, I think the whole thing, by the way, kind of ignores the idea of how much personalization goes on. So we're talking signed out prompts. That's I think what Gumshu does, too, by the way, which is signed out prompts, you just run into a free version of, you know, chat, TBT, which, by the way, as soon as they announced as soon as they announced, but we need to mention that.

chat dbd is announcing that they're bringing as to the to the platform you'll be able to pay them roughly $60 CPMs very pricey by the way to get your brand mentioned when people are searching again how would they know what they're searching right you're like they have to be scraping those queries and then serving them back to you in some ways you know if you're actually getting an influence right amongst these people i think so yeah they have to right like they have to store every single one of them and then tell you here's the search term or here's the prompt term that triggered your ad so i know if i'm even showing through the right stuff so that'll be fascinating to see if they end up doing that to actually see the prompts of people

and how they're looking for these types of things. But it'll also assume that you're getting some level of traffic or brand awareness from that. Great. I'm not opposed to it. But again, is that what's moving the needle? Is that how people are actually booking travel? See, to me, Chattibouti and Gemini, even stuff like that, Gemini probably does a little bit better job of it.

We talked about this previously. They're not really good travel research interfaces anyways. I think if there is a use case for Chatsybtee in the travel realm that we're in, it's maybe in finding a destination. So if he's had something to the effect of like, I go into chat chitp t and I say I live in a Myrtle beach. My airport code is MIR. Show me all the direct flights that I can take in the month of February to get out of the cold. We've got a lot of cold here. I want to get some more sunny.

I know there's a few like Fort Lauderdale, you know, Orlando, something like that. But like, show me all them. You know, and then, you know, make a list of them and get them back to me. That feels like a chat GPT use case I could get behind, right? Instead of be going into Google flights, clicking in my, NYU airport and then clicking out all the different options. Like that's a good use case. But again, you're not booking. That's not intercepting the booking of like, oh, yeah, go ahead.

book this flight. Like obviously you're going to do so much research outside of that, which is why I still think you have to have the website. And may I add, last thing that I'll say in that. I'm sorry. People like looking at vacation rentals. When they're booking a vacation, they enjoy the shopping process. They like looking around an Airbnb. They like looking around on your direct booking website. These things are legitimately enjoyable. In the same way that my wife going to like T.J. Max is legitimately enjoyable to her. Like she enjoys looking through the clothing and finding stuff. People like the act of booking a vacation rental. It is not. I just need a place close to the airport for, you know, a one-night stay because I'm staying over here at a

layover. That's a different type of travel entirely. That is not what you and I focus on. Maybe that could be more agentic, but I just don't think people are going to book a vacation rental. Just like we were told a while ago, oh, you'll book it, you'll book stuff based on voice. I don't think you're going to do that for vacation. Because you want to look at it. You want to look around. It's fun. Like people enjoy it. So it's, I mean, I think clarity, somebody who has clarity implemented on their website, clarity tells you that story. Like, I can't tell you how many sessions I see that are 10 plus minutes with 100 plus clicks. What are people doing?

Photo galleries, baby. All of your photo galleries and looking at every single photo. Like, that's it. It's fun. It's enjoyable to actually watch people navigate through. And like that little moment where you see someone like circling like a really cool picture and this is what happened. Like this is a real life case of what happens during travel planning. And we've said many times over the last few years that this is the one area that Google really hasn't tried to improve because they have so much ads been coming back in.

And to compensate for that, there's really no way that they can give the end traveler a significantly better experience without dramatically decreasing the amount of ad revenue that they're pulling in through the travel brand. until that weird break-even point, like meta-search is fine. Like, if you do a search for hotels right now on Google, it's a reasonably good experience. If you do a vacation rental search and start going into the map like a meta-search would be, it's clunky. It doesn't feel like you want it to feel. And I don't know. I mean, this is outside the AI side of things.

But ultimately, I think it does, I mean, it does speak to my. more to the travel trip planning experience and why we're not AI proof. We're not LLM proof. But the way people use, unless people are saying they have very specific specifications and how many people go in with, I have this many guests, I need this, this, this, this. Sometimes, sometimes not. But that's what it's going to, like, you're going to need to put in the specificity of what you're looking for. This is where if anybody at OpenAI, chat GPT is listening,

I would recommend actually having proactive sessions where, hey, could you use chat GPT better? Here's 10 ways to do that. Like put pop-ups. Like make people utilize this product more because I guarantee if you gave people good things that they could do with it, that they didn't know that they could do with it, that they didn't stumble on on TikTok or they didn't stumble on on some type of social media,

that's the key. I really do think that you could, I mean, people would get more out of it. And I think you could get people, hey, this is how you plan the perfect trip. Put in these things, your potential destinations, what you're looking for, what, you know, how you travel, how you do these things. Absolutely. You could get a good travel itinerary and some good recommendations. Now, I think ultimately you are. You're still going to use that search engine.

You're still going to do the due diligence like we've talked about. but that's where AI and travel can have a nice marriage and they can form something, but as it is right now, that's just not the experience we're getting. That is so frustrating for me as a marketer in this space. I'm sorry. That's it. Yeah, you bring up a good point with how people are using the interfaces. Like, they're just not doing it right.

The joke I made in the chat was, yeah, bring back Clippy. I want Clippy back. That's what I'm saying. Yeah, it sounds like kind of absurd, but, sometimes you have a good idea, you know, 20 years ago and just like the technology wasn't there. And then you bring it back and it works again. I mean, think about it. Apple failed with like a PDA.

What was it, the Newton? Like Apple made the Newton. This was like in the 90s, I think. And it kind of failed. And then they're like, you know, repivit, whatever, come back. Okay, iPhone. You know, biggest success consumer technology product ever developed, right? Presumably, or one of the top few.

So it's like they took a swing at it and the tech just wasn't there. My dad had one of those, the early PDAs where you had like, I don't know if you remember these. It had a stylist, first of all. And then it had like its own alphabet. So like you couldn't just write the letter A. It was like there was like an abbreviation for it. You know, you say that allowed now and it's just absurd. You just laugh at it.

But the time that was like, wow, I could like actually type an email while I'm traveling, you know, going between instead of having to be at my computer, which is plugged in. I don't even have laptop at that time that would have a good battery long enough to last on like a train ride or a car ride or something. You know, so you think yeah, PDA makes tons of sense. Like it's a personal device assistant, you know, and then, you know, now it's like, could you not see someone carrying like a and that there's, there's AI startups that have tried this like the whole necklace thing where there's,

like you're carrying a microphone essentially with you and then it's recording everything all day and then it dumps that in an L-LM model and tells you what happened that day. Like, again, a little creepy, a little odd. I don't know if that's the solution or not in the future, but I know there's a lot of stuff that we do now that we would have assumed as absurd that we're normalized or we're used to. Like Apple making watches. Like could they not just put a microphone in the watch and then put a super battery in there? And then it's like a watch on your wrist that looks very normal.

Like I wear watches all the time. No one ever looks twice at a watch. People would look at a weird little pendant on your neck. I don't think they'd look at a watch that way. So I don't know, just to get out a little bit at this point. But yeah, it's a makes sense. Even you talking about getting your power blipping out. I saw a woman going, you know,

women going, buying her groceries at the grocery store. And she pulled out her wristwatch and put it up to the, and like. Oh, Apple Pay, like on her watch. Yeah, Apple Pay through the watch and doing everything like that. And for just a second, I thought, okay, what if there was no Wi-Fi? What if there was no internet? We have really, not overshot, but it is. Technology just runs so much of our lives that there's little,

and it's where I have the questions about AI and buying and having all these things is that what if the rug gets pulled at some point? And then we're just kind of left here, hoping, stink and pray and, uh, uh, uh, uh, uh, uh, uh, well, it's, I mean, you hear stories about this where, like, you go to a restaurant and their systems down and it's like, oh, could I just like give you money in exchange for this goods? that I'm selling and they're like, no, like our system's not anymore. Yeah, it's like, you know, whereas like I can think of like a very opposite example. I'm sure this is like you're today,

but this was my childhood where there was like a farm stand sitting in the side of a road. And it was just like, yeah, like take a pumpkin and put like three bucks in here or take like some fresh food and put a few dollars in it. Like just such a different trust-based society, you know, then sometimes we probably live in now where it's like, yeah, tap your apple pay, you know, and you're out there in the middle of the forest in Western Massachusetts where I grew up. Like, yeah, sure, we'll take apple pay why not? Sounds good. Just funny. I don't think in Sherbourne, Minnesota, the town of less than 1,000 that I grew up, I don't think Apple.

I still don't think Apple Pay is taking on it. Yeah, yeah. We'll see. Well, you'll have to go back there. I bet there's a decent chance whatever gas station. Because there would be one gas station in town, right? Maybe two. Yeah.

I mean, it's not that small to town, right? All right. There's two gas stations in town one or two. I bet one of takes Apple pay. I put one box up right. When you go back home that time. I would think so. That's probably fair.

But they're not actively. I don't think they, they're probably resisted pretty hard. Now, no, here's a whole. If we zoom out for a second and come to a close here, I know we're probably at time. And if you're listening to this rambling, then you're an awesome listener. We appreciate you.

We really do appreciate you. Yeah, I think if the person behind the counter, what would he know about AI, right? So like how deep into our society has AI gotten? And sometimes I think like a lot, but back to your earlier point, I think you made a really good point earlier, not just is someone aware of chat TBD,

now they use it once or twice. I think that number, it's almost like, tie-in referencing like how aware people are vacation rentals now than they were 10 years ago. That number is probably pretty high at this point people have gone up and used them. But are they using it well?

is a totally different conversation, right? Of like, yeah, they went in there and like asked to do one or two things. They used a free version, which is kind of not very good. Or they had one or two prompts. They ran out of credits. Like, I think I saw there was some day that was released not that long ago. Maybe we could try to dig this up. What percent of, um, users on chat, TBT are paid users?

It's like eight percent. Yeah, five percent. Less than 10. For sure, less than 10. I was also surprised when I, when I read that. That signify. And, and I know, ultimately, I do think that they did want to make. get a free resource. And they, you know, money, yes, obviously they're trying to make money,

but I don't think they initially had the intent of truly monetizing it in that manner with membership and stuff like that. But yeah, it's a low number. Yeah. So this is an article from the register.com written a little while ago, to be fair, back in October. So maybe it's the touch dated. But they said, yeah, 95% are on free accounts and 5% are on paid accounts of like the typical chat dbtee web interface that you would use to interact with it. So yeah, we're We're so early in this process. And think about this too, by the way. Like, think about the number of people that subscribe to a paid service or give you

money in general. Open AI and chat GPT has produced probably one of the most interesting and technology, you know, shifting web, whatever you want to call it, interfaces in the last, since the mobile phone maybe. So since what, 2006, 2007, like iPhone coming out with people? Other than that, like, what can you think of before that? The internet. So, like, the internet, mobile connectivity, technology, and AI models, like, those are

three massive things that have happened in our lifetime, for sure. Like, I remember a life kind of before without internet. My kids certainly won't, but barely. Like, we had to dial up when I was a kid. So think about that for a second. And yet only 5% of people are subscribed to it. So we're so early in this process, you know, of like actually paying for it. Or the free models will end up being very important. Like if you're doing prompt tracking, you better check in the free model, because that's where 95% of your users are actually coming from, most likely, right? They're doing searches in the free model as opposed to the paid model for something like a like a chat GPT. So we're so early. There's so many more things we're going to learn from this

process. I think, again, this idea of like, you know, I know exactly how the future is going to go. I just don't believe it. Because like this CloudBot thing to put a close on that idea, too. It came out of nowhere, really. Like if you go study this guy like who built it, I mean, he was building it and he released it a few months ago, I guess, maybe like an initial version of it, didn't tell anyone about it. But I think if you look on the, the, um, what's the word of looking for here, the GitHub repository for all the code. It's open sourcing. Go look at all the code and see how it works. I think it went to a million stars like faster than any other piece of software ever that has been released on GitHub.

the most popular co-chairing platform again that's ever been invented so it's like think about that for a second this one from no one having it no one using it to a million stars and then probably many many millions of users using you know cloudbot in the matter of a course of a few weeks or something like that two three weeks that time frame so think about how fast things can move now and if you ask someone a month ago oh yeah there's going to be a command line interface for these platforms in the future where you can like build agents and have it do stuff for you in your computer and people are like yeah i can see that happening maybe next year or the year after two weeks later there it is and it has million the users. So it's crazy to think about when you kind of put it into those, you know, those,

those ideas or those concepts. Any other odds-in-ends we get to pick up here? We kind of hit on the main ones. So all these AI interfaces are changing, you know, be careful, prompt tracking. We have a lot more links in the show notes, I would say this time that we normally do. So I'll try to grab a few of these and pop them in there so that we get everything dialed in. But what other odds and ends would you want to pull up there before we come to a close here? I mean, I think just looking at where does it sit in your organization? You know, who, I think that's important. No, there should be, you should have someone in charge of AI.

Whether they're doing a lot, whether they're doing a little, there should be kind of some central repository, I would say, or some central resource or person. Yeah, I think how are you using it? Are you replacing people? Are you trying to remember we're in the hospitality space? I think, you know, any way you can use AI to make hospitality experience better for homeowners and guests, I think that's important.

But I think it is. I think these checks, I think we're going to see a lot of movement and we're going to continue to see movement in search and traffic and stuff like that. We're going to continue to get questions about how do I show up more frequently in LLMs. We're going to try our best. But ultimately, I do. I think that what we're going to continue to say is when you look at the final bullet point, chat GPT versus Google, we don't need to worry about investing in chat GPT heavily yet. because it's just not a big needle mover. And that's not just, I mean, yes, that is a vacation rental space,

but it's still the greater ecosystem. People still, the use case for ChatGPT is finding a quick answer or talking to someone. It's conversational. It's not yet to the point of, you know, when we look at the informational, educational, monetary or brand-driven, that's how, like, Semrush breaks things down.

That's not really how people are using those yet. And again, I think Open AI and Chat-GPT should, more on that. Get better. Make us power users. Make us at least baseline users. And that's not it. So, you know, level yourself up. Rip a page from the Claudebot, you know, playbook that people are enjoying and proactively reach out to me, right? Chat Chachabit doesn't do anything for me. If I don't do anything, chat Chachabit does nothing. It's an app on my phone. It's a tabby browser. It's an interface sitting there. You know, and I think if there is a version of this that we can, you know,

I got a taste of it. I got a shot glass taste of it with this cloudbot stuff and kind of playing with it for a few weeks. If there's a version of it that I think is way more interesting, as far as your personal productivity as a user engaging with AI and using it to help your business or help you be more efficient or more productive or whatever, it's here's what I want to accomplish. Here's some high level state of what I'm hoping to achieve. Help me achieve it.

Ask me questions, figure out, work with me, et cetera. And Claude, this actual software platform is far better at that too, by the way. I think then Gemini and ChatDBT and stuff like that.

I think Claude is trying to be like the new version of like a Microsoft Excel where it's like this ubiquitous thing that you would use in your day-to-day workflow. you know, chat chbtee still feels like a little bit more of like a broad use consumer brand to me. But blows my mind, this is totally a sad bar. We're just riffing at this point. The number of people who use chat dbti like therapy. I knew that was the thing. I didn't realize how big of a thing that was and people get really mad and they use their chat GPT interfaces, almost like a friend or they're chatting with like a again like a therapist or something in that realm, which my sister's in that space.

I know how much training it takes to actually do that to human beings and like how skilled you should be and how empathetic you should be. It's a little bit scary. But if I think about that for a second, And it's like, that's a very different use case than I have this Excel file. Help me make this Excel file better. And I think the truth is probably there'll be a bunch of companies that spawned from this that have, you know, specific AIs, right? Like there'll be a therapy-based AI that'll actually really be trained well and be battle-tested and be able to help people.

And yeah, imagine you could like talk to your problems with an AI and you wouldn't feel judged and maybe you'd be more honest and you could work through your issues. And if it's trained in the right way, maybe it could help you better than a person almost. And it's instantly available. It never sleeps. It never turns off, et cetera, et cetera. Maybe there's something there to that. just one small idea to pull out. Or like, you know, you see the vacational industry kind of playing

with this. It seems like guesty's going in this direction. Guess he's making an AI. Their co-pilot thing I saw the other day. It was like they're trying to make an AI that works in your guesty system to help you. Yeah. Oh, this property isn't going to booking in 10 days. What if it was like flagging that for you and saying like, you should take a look at this. Like this seems odd to me, you know, but it's only sending you the ones that need a booking. It's not like, hey, this property is fine. I'm not going to send you lure on that one. There's no, you know, action you need a day as long as the property's working well. So a lot of layers to go in with that. But all right, We keep going.

We should probably put it on it. So we trim it off and don't make this two hours. I don't think people want the Joe Rogan style podcast links. Leave us a review. Go to your podcast, have the choice.

iTunes, Spotify, get the most downloads. Click five stars. Leave it there. Five stars. You don't it be fun? Paul, if they could do seven stars to represent the seven super roll rings

the Patriots now have after the BBC Oxon Sunday. That'd be kind of cool. I'm wishing. We'll see where it all nets out. That's going to look really bad if they lose, but I'll get it hoping for the best. We'll get you in the next episode.