Bản chép lời đầy đủ của 037: Will AI Replace Hospitality Teams? | Stone Fish, Vancauwenberghe, Ullrich, Rubin

human guest messaging. 00:05:13.759 --> 00:05:20.079 Um I think very much so people are seeing that AI can do a much, much better job and actually improve reviews. 00:05:20.319 --> 00:05:28.720 Um and the AI never sleeps, never has an off day, um, and never forgets what the guest messages were several conversations ago. 00:05:29.600 --> 00:05:30.560 Thank you, Jay. 00:05:30.720 --> 00:05:31.680 Another great takeaway. 00:05:32.000 --> 00:05:34.000 Arlo, why do you go next? 00:05:34.639 --> 00:05:35.839 Hey, my name is Arlo. 00:05:35.920 --> 00:05:38.160 I'm one of the co-founders of Bestie AI. 00:05:38.560 --> 00:05:47.600 Um, I think one of the things that people get wrong is they think that the guest communication is just kind of like siloed into AI guest communication. 00:05:47.920 --> 00:05:56.000 Um so many of the guest issues actually come down to an issue with the cleaning team, an issue with the operations team, lock issues. 00:05:56.319 --> 00:06:02.240 So we actually have this all-in-one platform where we consolidate your software spend into one place. 00:06:02.560 --> 00:06:18.639 So we have a lock system, uh cleaning an office portal, phone system, uh, and and uh AI workflow system that just completely we aim to just completely automate every aspect of the operations of uh managing a short-term rental property. 00:06:20.079 --> 00:06:27.759 So just quick question, Arlo, without asking what do the operators get wrong, where can they get better? 00:06:30.240 --> 00:06:42.240 Um yeah, I mean they they're thinking that uh guest communication is kind of siloed as like one issue, um, where it's actually connected into like the rest of your business. 00:06:42.639 --> 00:06:57.600 And so needing to coordinate people have like all these different systems, they'll have like their locks, you know, lock management system, they'll have a cleaning and ops portal, they'll have a guest communication system, they'll be unified in box like all over all over the place, and people lock that all in one place. 00:06:58.639 --> 00:07:00.079 Yeah, awesome. 00:07:00.319 --> 00:07:00.959 Thank you. 00:07:01.040 --> 00:07:03.279 Um thank you very much, Arlo. 00:07:03.439 --> 00:07:06.879 Now, last but not least, um Anthony. 00:07:07.519 --> 00:07:09.040 Hi, uh Simon, hi everyone. 00:07:09.199 --> 00:07:10.399 Um so I'm Anthony. 00:07:10.480 --> 00:07:23.360 I'm uh the co-founder and uh CEO of Guestway, a little bit atypical, uh, but I'm also uh the founder of Stacy, which is the property management company in Belgium where we run almost a thousand units uh uh at the moment. 00:07:23.519 --> 00:07:26.959 So I have uh uh experience in a little bit of both. 00:07:27.120 --> 00:07:42.959 Um Guestway is uh a platform that um I would say is the cherry on top of the PMS, where we handle things like gas communication, but also uh operations, the guest journey itself, uh IoT, so smart log, thermostats, uh, etc. 00:07:43.199 --> 00:07:47.600 And we try to consolidate all

of that under under one umbrella, uh I would say. 00:07:47.839 --> 00:07:57.680 Um so what uh do operators uh get wrong um with regards to AI, I would say they often chase the wrong metrics. 00:07:57.759 --> 00:08:11.360 Uh they uh look often at the percentages, the percentage of messages that are effectively answered by AI, but I in my experience that's not where uh the time gets wasted by the virtual agents or by the operators. 00:08:11.600 --> 00:08:22.240 It should be the percentage of issues the uh the operator or the AI actually resolves, and in my experience, that's where most of the times get lost, where most of the time uh gets lost. 00:08:22.399 --> 00:08:29.279 Uh so that's something that uh that in our company that we pay a lot of attention to and and and that should be improved further on. 00:08:29.600 --> 00:08:30.800 Fantastic, Anthony. 00:08:30.879 --> 00:08:34.799 And you could have not given me a better segue to start opening the conversation. 00:08:35.039 --> 00:08:41.600 What are do my other panelists think about that, where time is wasted and and where things can be improved? 00:08:41.679 --> 00:08:45.440 I mean, that's such an incredible point that Anthony's raising. 00:08:45.519 --> 00:08:47.519 Uh, who wants to jump in here? 00:08:48.000 --> 00:08:54.639 I mean, I think the one thing I want to like piggyback on is people like to talk about like automation rate. 00:08:54.879 --> 00:09:00.879 But automation rate, if the customer satisfaction is bad, is is useless. 00:09:01.039 --> 00:09:09.440 Like all of us here could go vibe code an AI agent that responds to 100% of the messages and in not much time. 00:09:09.519 --> 00:09:19.200 Um, but knowing when it needs to speak up, escalate to you, and um actually, you know, really resolve what the customer is asking is the key. 00:09:19.360 --> 00:09:29.200 So I think like the biggest thing to look at is like is as your automation rate is going up, is your review scores and and conversation sentiment going in line with it and not tanking? 00:09:29.360 --> 00:09:36.480 Because if it's tanking, you've deployed the AI wrongly, and um overall it's probably gonna have like a net negative effect on your business. 00:09:37.120 --> 00:09:40.320 Would the other is agree, disagree? 00:09:41.200 --> 00:09:43.919 Yeah, I would I would agree entirely. 00:09:44.159 --> 00:09:50.480 Um it's it's all about the the reviews and the quality of the the guest experience uh for us. 00:09:50.639 --> 00:10:09.039 And um I think there's kind of like um an unidentified uh ROI when it comes to the quality of the experience that you're providing, just because we have better reviews on Airbnb, on Virbo, on Booking.com, on all these OTAs, you're able to actually charge more. 00:10:09.200 --> 00:10:14.720 You're getting better uh higher scores, higher uh search engine optimization. 00:10:14.960 --> 00:10:19.679 You're you're showing up higher in the results on Airbnb because your response time's faster. 00:10:19.759 --> 00:10:21.039 So all these things factor in. 00:10:21.120 --> 00:10:23.360 You get the badges, you know, you're a top host. 00:10:23.519 --> 00:10:34.960 So there's all these different things that factor in where you're able to now actually charge more, uh higher uh ADR, and then you're actually able to produce more income as a property manager. 00:10:35.120 --> 00:10:36.799 So I think all of that is extremely important. 00:10:36.960 --> 00:10:44.080 If you're just focusing on like, all right, let's just automate the communication, doesn't matter as long as the answer gets out, like you're missing a whole massive piece of it. 00:10:44.159 --> 00:10:45.919 It's hospitality, you know. 00:10:46.879 --> 00:10:56.480 Yeah, one thing to also add is I think, yeah, I think we're gonna see AI going from something that's purely reactive to something that's more proactive. 00:10:56.799 --> 00:11:06.399 So for example, having AI that right now, um, you know, if there's a guest issue, you know, the AI can kind of handle it based on what the guest said. 00:11:06.720 --> 00:11:17.600 But in the future, the AI is gonna notice when things could potentially go wrong with guests and intervene and say, hey, I noticed, you know, there have been a couple of issues during your stay. 00:11:17.919 --> 00:11:29.919 We wanted to let you check out late for free, or we want we sent this package to you, or we gave you a call, um, or things like that where it's intervening proactively. 00:11:30.320 --> 00:11:33.360 I think we're gonna see that um in the future. 00:11:34.639 --> 00:11:42.320 That's a great um segue, uh Arlo, to talk about the real ROI, or is it just a feature, right? 00:11:42.399 --> 00:11:44.720 So let's ground this in money. 00:11:45.360 --> 00:11:50.799 An operator, so let's take an example, an operator with like 200 units is drowning in messages. 00:11:51.039 --> 00:12:01.600 They install one of your tools six months later, what has actually changed on their PL and let's say on their headcount. 00:12:02.000 --> 00:12:10.240 So you know, can you are you able to give me a real number, not the marketing one, but like what really happens? 00:12:11.600 --> 00:12:19.840 And we can just take very pragmatic examples within the business of adoption in terms of your technologies that you're putting out there. 00:12:20.240 --> 00:12:25.600 Um I I can speak from from oh go ahead. 00:12:25.679 --> 00:12:26.000 Yeah. 00:12:26.240 --> 00:12:28.320 I can I can speak from my experience, Simon. 00:12:28.399 --> 00:12:35.840 If uh if I would run a company with 200 units, I would say in headcount uh there would probably not be much difference. 00:12:36.000 --> 00:12:48.720 Uh since we would still need operators to handle the uh the issues, um underlying cancel uh, for example, handling uh real issues, uh refunds, things I will never give to uh to an AI. 00:12:48.799 --> 00:12:55.759 Uh coordination of the the uh the layer between the operational teams and and

the back office uh itself. 00:12:56.000 --> 00:13:06.399 Um if we grow further, then the difference will be substantial, will be real, uh, because the time lost in answering basic uh questions uh will be reduced. 00:13:06.480 --> 00:13:11.279 Uh as that 200 is for me kind of the like the uh the base. 00:13:11.600 --> 00:13:21.919 And once you start going further than that, uh go up to 500 or 1000, you will need a less a lot less people to uh to deal with the standard stuff, answering basic questions, all of these things. 00:13:22.000 --> 00:13:24.320 Uh so that would be my opinion for that. 00:13:24.480 --> 00:13:29.679 The on the other hand, I I would say the biggest difference would maybe be um uh AI upselling. 00:13:29.759 --> 00:13:32.399 That's somewhere uh something that I strongly believe in. 00:13:32.639 --> 00:13:39.200 Identify opportunities and uh where AI can uh basically spot those and uh ask for an early check-in. 00:13:39.360 --> 00:13:46.480 Maybe you want uh uh a daily cleaning, things like that, um, or direct bookings uh after their stay. 00:13:46.559 --> 00:13:48.480 Uh I I strongly believe in that. 00:13:48.799 --> 00:13:52.000 So, Jay, that's an interesting point that you're mentioning, Anthony. 00:13:52.080 --> 00:13:57.360 I and I thank you for your for your honesty and transparency here, because I would tend to agree. 00:13:57.440 --> 00:14:15.759 But one thing that Jay has alluded to, which I think is is incredible, is is other things that are material that do not that maybe affect the top line more than your bottom line in terms of or in terms of your operating costs, such as reviews and and what actually happens and moves. 00:14:16.000 --> 00:14:19.440 So, Jay, you want to take this and and go a little bit further? 00:14:19.679 --> 00:14:25.440 What actually happens uh within the 200 unit uh level of businesses? 00:14:25.759 --> 00:14:37.759 Yeah, I mean I have a couple things, and to answer your question, I guess more directly, I think there's a point um where I think all of our platforms, all of the uh individuals here, we all have upsells uh of some form. 00:14:37.919 --> 00:14:44.799 I think upsells are great and they all have their place and we can be proactive about them and and all that, um, and they can absolutely increase your revenue. 00:14:44.960 --> 00:14:58.159 But I think there becomes a time where if you're nickel and diming your guests constantly back and forth saying, you know, yeah, we have to charge you to use this specific feature that they thought was included, it kind of starts to impact the overall experience. 00:14:58.240 --> 00:15:05.759 So you do have to be a little bit careful when it when you are a host, um trying to figure out what gets upselled, um, what gets upsold rather. 00:15:05.919 --> 00:15:25.840 Um, but I do think that uh, you know, the the comparison I would I would give is like for I would call it like the spirit airlines of uh property managers, where like they now have to charge

a lot less for the flight, but once you book it, they charge you for the bag, they charge you for the seat, they charge you for your boarding group, all these little things, right? 00:15:26.000 --> 00:15:29.279 So um they're not actually pulling in more revenue total. 00:15:29.440 --> 00:15:34.639 Maybe they're getting a lot more upsell revenue, but the the quality is is lower of the guest experience. 00:15:34.799 --> 00:15:37.360 They're not able to actually charge as much. 00:15:37.519 --> 00:15:48.399 Um, I would agree with what Anthony said about, you know, it does depend, I would say, on your for a 200 unit um uh property management company, I think it depends on how you have your team set up. 00:15:48.559 --> 00:15:56.240 So like I I do I've seen uh many property managers who have virtual assistants who are dedicated specifically to guest communication. 00:15:56.480 --> 00:16:12.879 In that case, that's like a very clear ROI where it's it's a direct swap of a virtual assistant for an AI because all if all they're doing is answering messages all day and maybe they're escalating to other people on their team, that's something that the AI can already do. 00:16:12.960 --> 00:16:16.720 Um, so when we project and we estimate, you know, how much can you really save? 00:16:16.799 --> 00:16:20.320 It's it's usually over 92% on guest support costs. 00:16:20.480 --> 00:16:29.919 And some of these 200 unit companies can be spending over$30,000 uh US on virtual assistants just doing guest communication uh per month. 00:16:30.080 --> 00:16:35.440 So that's a crazy amount of savings if you consider over 9% uh savings there. 00:16:35.840 --> 00:16:38.399 So, Arlo, you want to take that ROI question? 00:16:38.559 --> 00:16:42.080 I think that's a lot of our conversation in the industry goes around that as well. 00:16:42.240 --> 00:16:51.360 Creating efficiency, we're facing margin compression, we need to automate things, and we'll get to the human aspect in a minute uh in terms of automation ceiling and everything else. 00:16:51.519 --> 00:16:59.759 But Arlo, where where is your head at when we're having this ROI conversation and adopting some of the tools that you provide as well? 00:17:01.679 --> 00:17:10.960 Yeah, so where we're actually trying to take it is replacing all of your guest communications staff and all of your operations staff with AI. 00:17:11.119 --> 00:17:12.240 That's where we think it's going. 00:17:12.319 --> 00:17:13.839 That's where we're trying to take it. 00:17:14.000 --> 00:17:32.000 Um, and we just launched a new initiative to do this, actually, where uh people are replacing, you know, not so instead of the instead of you the old method was you use a software tool, the software tool escalates to a member of your staff, your staff handles it. 00:17:32.319 --> 00:17:43.200 The new system is the member of your staff isn't there anymore, and the AI does everything, and you're you're only filling it in when it doesn't know what to do,

uh, because it's information that it lacks. 00:17:43.519 --> 00:17:55.599 So what we're uh working on now is the ROI is gonna be get rid of all of your software, um, your unified inbox, your cleaning and ops, your locks, your uh you know, anything like that. 00:17:55.839 --> 00:18:01.680 Um, and also get rid of your operation staff and your and your uh and your guest communication staff. 00:18:01.920 --> 00:18:12.799 And the only people who are working in your operation are the kind of uh uh the the the managers and the humans who actually need to do things physically. 00:18:13.039 --> 00:18:16.400 That's where we think uh things are going, and that's what we're working on now. 00:18:17.359 --> 00:18:17.759 Wow. 00:18:17.920 --> 00:18:31.119 Okay, so that's a pretty big statement, and uh not um missing an opportunity uh to put a pitch out there, um, but that's massive and and definitely come back to that uh call. 00:18:31.200 --> 00:18:47.680 I mean, I'd love to hear your immediate response to what Arlo just mentioned because I think it goes uh quite in a different direction than than what Anthony has raised originally, and we'll talk about you know the balance and automation and the automation ceiling in a minute. 00:18:48.000 --> 00:18:48.400 Mm-hmm. 00:18:48.960 --> 00:19:02.240 Yeah, uh on what what Arlo's talking about, like I think that's you know, very in line with where like a lot of software is going just generally, is like with what the the more access you give the agent, the deeper it can go in the workflow. 00:19:02.400 --> 00:19:08.720 So, you know, it's not just saying, yes, you can have a late checkout, it's actually like checking your PMS. 00:19:08.880 --> 00:19:09.839 Can we accommodate this? 00:19:10.000 --> 00:19:17.920 Checking with the cleaner, telling the cleaner, hey, you're coming later, sending the guest the Stripe link, making sure it gets paid, updating the status on the reservation. 00:19:18.079 --> 00:19:23.279 Like AI can say, yes, you can have a late checkout, but you know, that's not that's not really where the work is. 00:19:23.359 --> 00:19:25.599 It's it's the six steps behind that message. 00:19:25.839 --> 00:19:35.839 And then coming back to the ROI point, um, undeniably, like the the most opportunity in AI, um especially at the enterprise, is for like a headcount reduction. 00:19:36.079 --> 00:19:50.000 And you know, it's it's great when you can you know reduce your your human spend$20,000 a month, let's say, but the real value comes to the value just added to the business, where you know you didn't just save$22,000 a month. 00:19:50.319 --> 00:20:01.440 That over the year at the you know six, eight X multiple your business is going to trade at could be a million millions of dollars you just added to the value of your company. 00:20:01.599 --> 00:20:10.720 So if anyone is has like MA top of mind, for sure uh be getting your your AI strategy in place before you you're bringing the business to market. 00:20:10.880 --> 00:20:17.279 So you just have the the most, the leanest financials possible and you can get the the most value for your business. 00:20:18.079 --> 00:20:21.039 Yeah, I mean, I would have a lot more questions in relation to that. 00:20:21.200 --> 00:20:23.839 I mean, one is why does not everybody adopt that? 00:20:23.920 --> 00:20:27.920 I mean, the your value proposition is just uh too good to be true. 00:20:28.000 --> 00:20:31.200 But let's move forward about the automation ceiling. 00:20:31.279 --> 00:20:50.480 So, what percentage of guest messages should be fully automated, or any process within the basis should be fully automated where uh whereas the line you should never cross in terms of the human capital within the hospitality industry, which is equally important while we're running these businesses. 00:20:50.640 --> 00:20:55.680 So we were like you in your mind, we can take just automate everything that is happening. 00:20:55.759 --> 00:20:59.839 On the other side, let's not forget we're still in the hospitality industry. 00:20:59.920 --> 00:21:02.640 Uh and Anthony, I want to throw that to you. 00:21:02.880 --> 00:21:15.519 Um, also coming from the operating side, and where is your mind in balancing um that uh adoption in terms of AI functionalities and where's the human capital and where's the ceiling in automation? 00:21:15.680 --> 00:21:17.359 Where is the ceiling in automation? 00:21:17.920 --> 00:21:23.519 Um I really think this question depends on the the size of the company that you are. 00:21:23.599 --> 00:21:34.000 Uh if you are a very large uh company, already have a big brand uh behind it, you probably already have people uh dealing with uh certain stuff uh like this. 00:21:34.319 --> 00:21:39.519 Um and uh the damage to the brand will be substantial if if things go wrong. 00:21:39.759 --> 00:22:03.759 Um so um if if you are an operator uh like us, have a thousand units, then uh I wouldn't go too far or too aggressive on the autopilot mode or uh however we we all like to call it, uh the basically uh the lower the confidence uh you accept uh the The faster it will answer, if if I can say it like that. 00:22:04.880 --> 00:22:27.599 So I would say from what we've experienced with with our clients and from smaller property managers, they um tend to um set the autopilot much more aggressively, uh where they um where they basically uh say, oh, you can respond to uh issues, you can uh agree to cancellations, all of these things. 00:22:27.920 --> 00:22:31.359 Personally, for me uh as an operator, I would never do this. 00:22:31.440 --> 00:22:41.440 I uh would only let the AI answer on factual things, uh knowledge questions, where the answer is uh almost 99% uh uh sure. 00:22:41.680 --> 00:22:51.920 Um but uh I I I guess it it really uh depends on how you feel in the market, if you are a hotel, if you are a pure short-term rental player, the size that you are, uh etc. 00:22:52.160 --> 00:22:52.880 So yeah. 00:22:53.119

--> 00:22:56.319 So how do you find out the answer is 99% true? 00:22:56.400 --> 00:23:02.160 You get me going here, and and who wants to take it, Anthony or anybody else, because I could not agree with you more. 00:23:02.480 --> 00:23:07.119 Yeah, so uh um I can only speak from uh from my experience, of course. 00:23:07.200 --> 00:23:17.119 Uh how would we know this uh is if the answer is specifically um mentioned in uh the knowledge base, in the guidebooks, in the safe replies. 00:23:17.279 --> 00:23:30.880 Uh what for what we would never accept as a high confidence answer is something that was mentioned by uh an AI previously or by an operator in the in a past conversation, this would uh gain a much lower confidence score uh for us. 00:23:30.960 --> 00:23:34.640 So when the machine gets it wrong, guys, who wants to take that next? 00:23:34.799 --> 00:23:37.279 I have a very recent example. 00:23:37.440 --> 00:23:49.039 I've had fantastic friends from Australia coming over, I put an itinerary together on an LM uh to do visits and travel in Ducerne on the lake and everything else, gave me recommendations of a restaurant. 00:23:49.200 --> 00:23:52.000 This restaurant was closed for two years since two years, right? 00:23:52.960 --> 00:23:55.359 How do we go and deal with this? 00:23:56.400 --> 00:23:59.039 Cole, you want to go next, or who wants to take this? 00:23:59.359 --> 00:24:00.799 Yeah, yeah, I could take this next. 00:24:00.880 --> 00:24:06.640 So the way Conrad set up, uh it's a little more black and white in terms of what the AI can respond. 00:24:06.799 --> 00:24:10.880 We're only responding if we are 100% confident in the answer. 00:24:11.119 --> 00:24:22.000 If we're not 100% confident because we're missing context, we need a human in the loop, or it's simply you've said we don't want AI to handle this, it's gonna come back as a to-do in your inbox. 00:24:22.240 --> 00:24:27.200 So the concept of reviewing what the AI got wrong is is hopefully not a thing. 00:24:27.279 --> 00:24:32.960 It's just reviewing what the AI could not handle, and you're teaching it to get that automation rate up. 00:24:33.119 --> 00:24:54.880 And um I I'm not so aligned with Anthony's point of view with the kind of reluctance to roll out this level of automation because at the enterprise we're we're seeing these customers reach 80% with higher sat guest satisfaction scores than they came to us with, with faster reples, with better rankings on the OTAs. 00:24:55.119 --> 00:25:02.559 So I think the the time has passed where people need to hold back on on what they give the the AI control over. 00:25:02.960 --> 00:25:10.559 And um the the biggest bottleneck for automation rate is is really just like access to the other tools in your stack that it needs. 00:25:10.720 --> 00:25:12.400 Like sending the message. 00:25:12.640 --> 00:25:18.400 Yeah, um, if if it can't go check the the PMS to see when that room block is, like it can't automate that. 00:25:18.559 -->

00:25:22.480 Um and that's just not like things you're but what what about when the source? 00:25:22.880 --> 00:25:28.240 Sorry to interrupt you there, Cole, but what about if the source is like for example, the internet, right? 00:25:28.400 --> 00:25:46.240 So it like how can you make sure that all the information that you provide in the guest comms um circle within the short-term rental industry is what the is is you can a hundred percent be sure that any information you provide is relevant information to the guests. 00:25:46.960 --> 00:25:51.680 So, like the information that it goes and finds online, how can we confirm that it's correct? 00:25:52.079 --> 00:25:52.640 Basically. 00:25:53.440 --> 00:25:59.599 Um I mean to play devil's advocate with that, I don't know how a human would confirm it's correct either. 00:26:02.799 --> 00:26:03.119 Okay. 00:26:04.559 --> 00:26:05.680 Our law. 00:26:06.559 --> 00:26:11.920 Yeah, I I agree with the we're actually trying to go um to answer your original question. 00:26:12.240 --> 00:26:16.240 We're trying to go completely automated, 100% automation. 00:26:16.559 --> 00:26:19.440 Uh AI is getting, you know, 1410. 00:26:19.599 --> 00:26:22.799 GPT 4 got a 1410 on the SAT, which is like a while ago. 00:26:22.880 --> 00:26:27.680 I don't know what the latest models would get, but would definitely be smarter than most of us. 00:26:27.920 --> 00:26:32.160 Um it's passing the LSAT as an MCAT, it's passing all these standardized tests. 00:26:32.319 --> 00:26:42.480 So there's no reason why you can't teach AI to run your business better than you can if you give it access to all the information, if you give it access to all your tools. 00:26:42.720 --> 00:26:46.640 Uh, the AI should be making a lot of these decisions. 00:26:47.039 --> 00:27:13.599 Uh so we're actually we're we're working on just having the AI ultimately uh not only work according to the uh workflows and formulas that it's given, but what it's gonna do in the future is it's gonna learn actually what are the best policies that I should be following based on the things I've seen from similar guests, based on similar situations, similar scenarios. 00:27:13.839 --> 00:27:26.880 It's gonna be coming up with its own policies for what to do, and it's gonna actually, there's gonna be times in the future where a human would have done something differently than the AI, but the AI is actually gonna be right. 00:27:29.119 --> 00:27:33.359 Okay, so let's put the truth to the to the play here. 00:27:33.599 --> 00:27:39.039 What is the worst answer your your technology has ever provided to a guest? 00:27:39.279 --> 00:27:40.319 Let's be honest here. 00:27:40.559 --> 00:27:41.920 Who wants to go first? 00:27:43.359 --> 00:27:45.039 I have a story I can share. 00:27:45.279 --> 00:27:47.519 Um Go Jay. 00:27:48.880 --> 00:28:03.920 I think um, I mean, I think we're all kind of talking about it a little uh maybe directly but indirectly in some circumstances where it it really does all um the the way that the AI responds, it relies on the information that you give it. 00:28:04.079 --> 00:28:13.039 And if that's it responding to 98% of the guest messages that come through because you provided it the knowledge to do so, that's great. 00:28:13.279 --> 00:28:23.359 Um and to kind of like go back on the last question with with host buddy, it's very much more of the the polarized like whether it's right or it's if it's right, it's giving the answer. 00:28:23.519 --> 00:28:31.599 If it's wrong, you get a notification, it goes into your action items dashboard, you get to figure out why don't we have this information, how do we add it, suggestions for how to do so. 00:28:31.759 --> 00:28:34.240 So I think we're all pretty much aligned on all of that. 00:28:34.400 --> 00:28:36.480 Um AI shouldn't hallucinate. 00:28:36.559 --> 00:28:41.759 Like a good AI is an AI guest messaging system, it's not going to just make things up and send it to a guest. 00:28:41.920 --> 00:28:47.839 Um, you do give the guest or the host the ability to customize the AI. 00:28:47.920 --> 00:28:50.559 Um, and I'm sure a lot of our lot of our platforms do this. 00:28:50.720 --> 00:28:53.279 What we do is a specific custom tone section. 00:28:53.359 --> 00:28:57.839 So you can make it embody a certain type of person, a character from a movie. 00:28:58.000 --> 00:29:06.799 Um, we have themed Airbnbs that some of our hosts run that they want it to sound like um like a character from Disney, for example. 00:29:07.039 --> 00:29:13.599 So I have one story where we had um a host who was managing near Disney World. 00:29:13.759 --> 00:29:19.599 They had the AI that they wanted to sound like Mickey Mouse, and so that's what they put in the custom tone section. 00:29:19.680 --> 00:29:24.000 So it's a little bit lighthearted, you know, it doesn't really take a lot of concerns very seriously. 00:29:24.160 --> 00:29:33.839 Um, this host had the air conditioning at the property went out, and the guest was asking about it, you know, pretty upset about it, pretty late at night. 00:29:33.920 --> 00:29:49.440 And the AI is responding as Mickey Mouse, like, you know, um laughing about the situation until it realized, you know, that they mentioned, oh, we have a baby, like we really need to get this figured out, like we can't sleep, and it kind of cut the crap and uh and switched gears. 00:29:49.599 --> 00:29:55.200 So it all of that to say, like it really does depend on how you inform the AI to follow instructions. 00:29:55.359 --> 00:30:00.319 But I think it's just so cool the the type of experience that we're able to provide uh guests nowadays. 00:30:00.400 --> 00:30:06.480 Like, how cool is that to provide a family with kids, the experience of being hosted by Mickey Mouse, like all these different things. 00:30:06.640 --> 00:30:08.720 So um I think that's where it comes from. 00:30:08.960 --> 00:30:15.599 But uh I wouldn't say that there's any particular

situation where I would say, like, oh, the AI told the guests to go to the wrong address. 00:30:15.759 --> 00:30:18.079 Like that just shouldn't ever happen, right? 00:30:19.359 --> 00:30:20.000 All right. 00:30:20.240 --> 00:30:33.680 Yeah, I mean, I heard um, I heard of stories where you know there was like love affairs happening and love talk with agents uh during conversations with people that had to stop that at the same time. 00:30:33.839 --> 00:30:36.640 Uh interesting, but there is a lot of stories out there. 00:30:37.039 --> 00:30:39.680 Let's let's switch up a gear here. 00:30:39.839 --> 00:30:46.400 Um, I want to talk about why do standalone tools survive going forward? 00:30:46.559 --> 00:30:54.480 And my argument would be uh in today's world, every PMS host of a guest D, a lot of them is building AI messaging natively. 00:30:54.559 --> 00:30:56.480 We've just seen a big announcement of Guest D. 00:30:56.720 --> 00:31:00.160 Every guest has Chat GBT claude in their pocket. 00:31:00.240 --> 00:31:06.079 So why does a standalone AI guest communication company exist in three years? 00:31:06.319 --> 00:31:12.960 And what's the moat that it just, you know, we did it first, follow-ups. 00:31:13.119 --> 00:31:19.119 If a major PMS ships this is a free feature next quarter, what happens to you? 00:31:19.200 --> 00:31:22.400 And what does that mean for where you guys came from? 00:31:22.480 --> 00:31:33.759 And I mean, obviously, I met all of you uh when you uh started the business coming into the industry at at a big bang, but but now the bigger question is how is that going to evolve? 00:31:33.920 --> 00:31:42.240 And I think uh, you know, we heard it already from Arlo going forward to say, hey, we're we're now expanding our rally proposition. 00:31:42.400 --> 00:31:43.359 Who wants to take that? 00:31:43.440 --> 00:31:45.039 It's a it's a big one. 00:31:45.920 --> 00:31:47.680 I I could take that one first. 00:31:47.839 --> 00:31:54.640 Um so like how does how does a third party like stay defensible? 00:31:54.799 --> 00:32:01.759 Um so we're focused on you know the guest communication today um and all the other communication in the business. 00:32:01.839 --> 00:32:08.799 So like we're not just trying to automate business to guest, we want to automate business to vendor, business to cleaner, business to homeowner. 00:32:09.119 --> 00:32:18.400 So now that we have like all the communications brought into one platform, the rest of the work that happens in the business has to come up in a conversation. 00:32:18.559 --> 00:32:19.359 It's triggered. 00:32:19.519 --> 00:32:31.359 So if all the conversations are happening on our system, we can be the source of truth for all the the work that happens after the message, and we can effectively be orchestrating the whole tech stack. 00:32:31.599 --> 00:32:45.759 Um, the PMS, the pricing system, the task manager, like a lot of these best in class point solutions will serve a purpose, and these AI layers

will be sitting at the top, orchestrating all these tools below, uh, which is what we aim to do. 00:32:47.200 --> 00:32:49.039 Who wants to go next? 00:32:50.079 --> 00:32:55.279 I can add, um, I mean, I would say I think the gap can only get wider. 00:32:55.440 --> 00:33:08.240 I mean, I I know the PMSs have resources as well, but I think what we're starting to see is I know, for example, a lot of the the major PMSs in the in the US, um, they're starting to release AI guest communication as a feature. 00:33:08.319 --> 00:33:14.400 A lot of times it's just that draft mode where it's suggested, you get to edit it, you move it forward. 00:33:14.640 --> 00:33:21.440 Um with my platform, uh host buddy specifically, there's basically no option besides to go on autopilot. 00:33:21.519 --> 00:33:27.519 Like you can go in, we have a we have a unified inbox where everything brings together, you can see the guest sentiment, add notes, all that stuff. 00:33:27.680 --> 00:33:33.519 But it's um, you know, the the quality of the guest responses is always going to come first because it's the hospitality industry. 00:33:33.599 --> 00:33:39.440 But I do think the gap can only continue to widen just because we're only focusing on the guest communication. 00:33:39.599 --> 00:33:48.960 We have so much more data right now to use on not only what like what Cole mentioned, the PMS, but also you're coming from text messages, from emails, from all those other sources. 00:33:49.200 --> 00:34:01.039 Um, and I think like it it's to make a comparison, I would say like you wouldn't go to like your your general care practitioner for a heart surgery or something. 00:34:01.119 --> 00:34:09.519 You know, like you're gonna go to a heart surgeon to get to get that work done because it's a a specialized service that this is the only thing that we focus on. 00:34:09.599 --> 00:34:10.880 It's what we live and breathe. 00:34:11.039 --> 00:34:14.880 Um so I would say because of that, we're gonna stay 10 steps ahead. 00:34:15.760 --> 00:34:17.920 That's a big statement, Jay, and I love it. 00:34:18.000 --> 00:34:25.519 And it gives me a good segue to go a little bit deeper and and picking up on what um Cole said in this terms of social truth. 00:34:25.679 --> 00:34:27.599 And I want to throw that at you, Arlo. 00:34:28.000 --> 00:34:34.800 I mean, we had we started to have a big conversation about the source of truth, and PMSs would claim they are the source of truth. 00:34:35.199 --> 00:34:53.679 But you obviously, as in like building what you're building right now and going a lot broader, who ultimately within the STR space owns the sort of truth today, and who is going to own the source of truth in the future in your U point when it comes to a property management uh company? 00:34:54.800 --> 00:34:58.239 Yeah, I don't think the PMS um is going anywhere. 00:34:58.400 --> 00:35:00.639 I think they're gonna be the source of truth. 00:35:01.360 --> 00:35:04.000 Um there's a lot of other data as well. 00:35:04.159 --> 00:35:11.519 Uh, like a lot of data right now is in people's cleaning and operations platforms and their SMS and their email. 00:35:12.239 --> 00:35:20.480 Um so we do, you know, we do consolidate that, for example, and I'm sure a lot of the other uh people here consolidate it as well, the other vendors. 00:35:20.800 --> 00:35:27.440 Um so it sort of lives in both places, but the PMS at the end of the day is the kind of underlying pipes. 00:35:27.760 --> 00:35:34.000 Uh so I do think a lot of that uh data is Okay, so then what does that mean for your differentiation? 00:35:34.159 --> 00:35:50.320 If we if we look at the uh at the announcement of guests yesterday with their agentic agents and then the platform that they've announced, what does that mean exactly in the in the in the context of the source of truth and the way you can differentiate your services from Yeah? 00:35:50.400 --> 00:35:57.440 I mean the PMS is a source of truth, but that doesn't necessarily mean that they're the intelligence layer or the decision-making engine. 00:35:57.760 --> 00:36:08.400 Um what what happens right now is you've got the kind of so you've got the software that's just these pipes, and then you've got people who use the software making decisions. 00:36:08.719 --> 00:36:12.559 In the future, the decision making is gonna be done by the AI. 00:36:13.039 --> 00:36:19.840 Um and so the software is still gonna be the pipes, but then the AI is gonna be making the decisions and actually not in business. 00:36:21.199 --> 00:36:22.159 Anthony. 00:36:22.880 --> 00:36:24.400 Yeah, I I agree with Arlo. 00:36:24.639 --> 00:36:31.360 I also do think the the PMS is uh is a source of truth, but it it doesn't mean anything uh uh to me. 00:36:31.440 --> 00:36:39.760 We we are still uh or all of us I think uh have a platform that interacts uh with the PMS, pushes data, gets data, can update it. 00:36:39.840 --> 00:36:49.840 Um and we can be the layer, the intelligence layer on top that uh orchestrates things uh and arranges things for for uh the inside the BMS or on top of it. 00:36:50.800 --> 00:37:00.239 So let's move on to that because I think that's a great uh segue there, Anthony, to talk about AI as a revenue and not just as a cost as well. 00:37:00.400 --> 00:37:11.679 And and how can we, and I think that's where a lot of opportunity lies, especially with you guys' platforms like everybody sells AI comms as a cost saver in terms of operation. 00:37:11.840 --> 00:37:15.360 We talked about that, automation, VAs, et cetera, et cetera. 00:37:15.440 --> 00:37:18.960 But the bigger price is the revenue, the upsells. 00:37:19.199 --> 00:37:25.679 Some of you have alluded to that already, early check-ins, paid late checkouts, repeat direct bookings. 00:37:26.079 --> 00:37:33.039 So can anybody of you show me the proof and where is your head at because of plugging into that? 00:37:33.119 --> 00:37:42.320 Because at the end of the day, we're

constantly driving additional tech costs, we have margin compression within within our industry, through distribution, through OTAs. 00:37:42.480 --> 00:37:44.400 It's becoming more and more expensive. 00:37:44.559 --> 00:38:00.320 But I think I think we need to have a mind shift, what AI actually can do for us going forward, because you know, I think Anthony brought an interesting point to say, hey, and I like that in in all total honesty, yet let's just not here to automate everything. 00:38:00.400 --> 00:38:06.400 Um, but we can also improve things with the revenue, the reviews, and everything else that Jay alluded to. 00:38:06.559 --> 00:38:18.719 So I think the up cell and and the other opportunities that this industry has, which is actually after 20 years I've been in the industry, never been well capitalizing that is probably even the bigger opportunity. 00:38:18.960 --> 00:38:20.800 Cole, you want to go for this one. 00:38:36.159 --> 00:38:38.400 We can't we can't hear you hear you back. 00:38:40.880 --> 00:38:44.159 No, you go for it. 00:38:44.880 --> 00:38:45.599 Yeah, sure. 00:38:45.679 --> 00:38:46.480 I can step in. 00:38:46.639 --> 00:38:48.719 So I mean, I think there's a lot at play here. 00:38:48.800 --> 00:38:58.880 I think there's obviously the upsells, then there's the the experience that you're giving the guests to actually get you better reviews, get you higher ADR, um, ultimately ultimately more revenue. 00:38:59.039 --> 00:39:09.280 Um, I think what's it kind of going a little bit deeper is um kind of how the industry is is changing a little bit as well, and how we're seeing um guests are booking. 00:39:09.440 --> 00:39:18.559 So we're seeing uh obviously guests are really loving Airbnb, still using Virgo, but a lot of guests are now going to be just searching on Google to go and find their stays. 00:39:18.719 --> 00:39:27.199 And the way that the the SEO is is showing up for them, um, a lot of times now you can get your direct booking sites to show up on Google search results. 00:39:27.360 --> 00:39:29.519 People are feeling more comfortable to go and book there. 00:39:29.679 --> 00:39:36.239 If you have an AI agent sitting on your website helping with conversions, you can actually get more bookings. 00:39:36.320 --> 00:39:41.280 Um, and I think that the amount of people that are doing direct bookings is actually increasing over time here. 00:39:41.360 --> 00:39:42.800 And that's something to kind of look at. 00:39:43.039 --> 00:39:55.440 So um rather than host just hosting one group of guests and then never seeing them again, you get to create this direct experience with them and really offer them the best that you can so that they're going to be returning guests. 00:39:55.519 --> 00:40:01.760 And now your calendars are booked out nine months in advance with the highest pricing possible, so you're able to capture maximum amount of revenue. 00:40:01.840 --> 00:40:03.920 Um, there's also the upsells portion of it. 00:40:04.000 --> 00:40:10.480 And like I mentioned, like I think there's a place for it and the AI knowing the right opportunity and what to upsell and things like that. 00:40:10.559 --> 00:40:12.559 And there's a lot of opportunity for that. 00:40:12.800 --> 00:40:34.239 Um specifically, I don't have the data on the on the upsells portion, but we are doing a report uh here soon with Air DNA to show the uh the increased revenue and review scores related to uh hosts who are using our platform specifically to automate over 95% of their guest messages and how that actually impacts their business and their revenue. 00:40:34.800 --> 00:40:35.119 Okay. 00:40:35.599 --> 00:40:38.559 But I want to come back to the to the upsell opportunity. 00:40:38.639 --> 00:40:53.519 You know, in the ideal world, that's where I can see AI can play a massive role managing the calendar, seeing, you know, seeing gaps between bookings, sending automated messages to the next guest, say, hey, you can check in two days earlier. 00:40:53.679 --> 00:40:56.800 The property's available at X amount of price. 00:40:56.880 --> 00:41:04.880 Um, you want to come earlier because you see the operation has already cleaned the property, it's ready four hours or earlier than anticipated. 00:41:05.039 --> 00:41:11.840 I mean, that's money that currently is being left on the table entirely because it's just so cumbersome. 00:41:11.920 --> 00:41:13.199 It's labor intensive. 00:41:13.280 --> 00:41:16.719 You know, the cleaners have to message back the property's clean. 00:41:16.880 --> 00:41:30.079 Okay, now we can we can we can address the customer coming in and say, hey, by the way, your unit is ready five hours earlier, you can you can come in for free or you pay, or we see a cancellation post-booking. 00:41:30.159 --> 00:41:33.760 Say, hey, you can stay four days longer, the weather, the forecast is nice. 00:41:33.840 --> 00:41:35.039 Uh, you can stay longer. 00:41:35.199 --> 00:41:36.559 We automate these processes. 00:41:36.719 --> 00:41:50.079 I mean, that's where I can see AI being a no-brainer in maximizing upsells next to um um obviously automating guest comms in in in related uh operation issues. 00:41:50.480 --> 00:41:51.280 Yeah, totally. 00:41:51.440 --> 00:41:54.960 I mean, just to touch on that real quick, and I can let other people kind of speak on it as well. 00:41:55.119 --> 00:42:00.400 But um, those are the easy ones I would say is the early check-in, late checkout, gap night upsells. 00:42:00.480 --> 00:42:04.239 So you're able to increase your occupancy because now you're filling those gap nights. 00:42:04.320 --> 00:42:08.719 If you have a two-night, three-night minimum, maybe there's a few nights in between that you can't usually book. 00:42:08.880 --> 00:42:11.840 Now you're upselling that to a guest, maybe giving them a discount. 00:42:12.000 --> 00:42:15.679 But those are the ones that generate the most revenue, in my experience, from what I've seen. 00:42:15.840

--> 00:42:27.519 There's other ways to connect to different local uh providers for um experiences, guest experiences, uh, going to local restaurants and bars, doing activities in the city, things like that. 00:42:27.599 --> 00:42:28.960 And then you can get a piece of that. 00:42:29.119 --> 00:42:32.559 And the AI knowing when to upsell those things is also really important. 00:42:32.719 --> 00:42:41.280 So we have a lot of integrations directly with like private chefs and um like the host co is one where they have a ton of different uh available upsells on their platform. 00:42:41.440 --> 00:42:49.119 So then they can provide uh the AI, and the AI can directly send the link, you get the commission, and now you're just making more money uh while you're sleeping. 00:42:49.280 --> 00:42:51.679 So all of those things can happen simultaneously. 00:42:51.840 --> 00:42:58.000 I will say the the upsells for the early check in and the gap nights are what I see as the biggest opportunities in the industry still. 00:42:58.159 --> 00:43:02.559 But I do think those have been around here with all of our platforms, I do believe, for quite some time. 00:43:02.639 --> 00:43:05.519 And it just becomes normal to uh to us at a certain point. 00:43:05.599 --> 00:43:09.519 But if you're leaving gap nights on the table, that's a big thing that you get to change as a host. 00:43:10.880 --> 00:43:20.960 Who else wants to have a go at it at the upsell opportunity here and and and their view on how we can uh we can automate that better also through AI? 00:43:23.679 --> 00:43:29.360 Yeah, I'd say um we, you know, the upsells, we also uh manage upsells. 00:43:29.440 --> 00:43:34.639 A big problem with coordinating upsells is knowing whether or not the, for example, the early check-in or late check-out is available. 00:43:34.880 --> 00:43:38.079 So making sure that you have a way to coordinate with the cleaners. 00:43:38.320 --> 00:43:43.519 We do it in our platform, so we have like an app that the cleaners use that gives them schedule. 00:43:43.679 --> 00:43:49.760 And we have AI automatically route which cleaner goes where to maximize, you know, minimize driver time and all that. 00:43:50.000 --> 00:43:53.199 Uh so that's uh you know, that's helpful. 00:43:53.360 --> 00:43:58.960 But any way that you can uh maximize revenue via more app sales, more direct bookings. 00:43:59.360 --> 00:43:59.599 Um 00:44:00.159 --> 00:45:04.960 those are all those are always uh up fault for for gas irritation of course Anthony you being an operator how do you go about upsell and and and capturing these opportunities with your own platform yeah uh I I would say uh I would say there's still a a lot of room uh for improvement uh in uh in our uh property management company for this um but to touch on the the point that Arlo made it's um uh I I I 100% agree the the biggest problem is not identifying if if the upsell is possible early check in late checkout there are quite some easy rules that you can follow and and know uh if it's possible or not but it's to coordinate it operationally uh especially um if you have a lot of uh short-term rental units um that are spread out single uh unit listings uh then uh knowing that if it's possible to uh facilitate a late checkout on early check-in is is the real hassle here uh operationally um so um I I think AI is is not gonna be the solution to that. 00:45:05.119 --> 00:46:27.199 I think uh a good um sort of operational platform that uh is the the layer between the operations team and uh the back office uh and that can exchange that information of course the AI can also access that information and uh in the future uh is the solution to to this problem and I I see huge value in uh in automated upsells and uh especially early check-in and and late checkout it's it's a um it's a big part of the revenue of a property management company as usually the the part of the upsells is completely for uh for them um and it's something that we heavily focus on as well so yeah Cole I can see you're back should we just check if you're um if you can come back into the conversation uh apologies for that no worries biggest upsell opportunity I think here especially in the the states I don't know if this is the same over in Europe but it's the damage waivers um yeah interesting you're raising that uh that's where everyone's making money a hundred bucks a stay you know one in four with the family go for it um and the way they're using AI to to help this is not just offer it but they'll write back a status to the PMS where it's like um damage waiver purchase and if we're we can have workflows where we're filtering on that status before they check in. 00:46:27.280 --> 00:46:34.320 And if they haven't purchased it or maybe they didn't see the message we're nudging them along trying to to make that conversion happen. 00:46:35.599 --> 00:46:52.239 Are they offering to I think yeah like in the STR industry like it's tough to have a ton of uh upsells outside of like the late checkouts, you know the the chef referral activity referrals because they're truncated by like the services you can offer without like a an on-site FMB component or SPA. 00:46:53.360 --> 00:47:06.400 So you talked about uh damage waiver what about cancellation insurance and things like that which still has is a totally an undersurfed area as well that you can potentially jump into as well. 00:47:06.880 --> 00:48:28.800 Mm-hmm Yeah people are making the most money on on the non-obvious upsells the the damage waivers and the drug insurance so let's talk about the 18 month horizon um 18 months out agentic ai books rebooks upsells resolves issue handles the whole guest journey end to end in that world does that guest ever talk to a human again and and where is the human capital within the future of operating a property management company I would love to hear from all three four of you where you probably in a totally different mindset than than people who

are um uh executing hospitality um Simon I I don't know if you guys mind but I have another uh podcast webinar to jump to right after half hour here um so I would say it would be sorry did you repeat the question actually yeah I mean talking about the 18 months where is the human capital in 18 months within in full automation of you know the the agentic AI that books, resells, takes care of the desk guest comms and everything else I just want to hear where does the human capital sit within our organiz within our industry in 18 months' time. 00:48:30.079 --> 00:49:10.159 So within the short-term metal industry where there's not the on-site person, I would say the most important role in this post AI world is the conversation engineer that's reviewing what the AI could not handle, teaching it rules, adding skills, adding new connections, workflows to use AI to truly enhance this customer experience and not just automate more but deliver better hospitality thank you call appreciate it thanks for participating in STR Global Unlocked Okay Yeah please Yeah I would say the way I think about it is there's kind of three phases of like AI implementation. 00:49:10.320 --> 00:49:16.159 The first is what we've been doing the last couple of years which is having AI assist a human. 00:49:16.320 --> 00:49:36.559 It's still the human's responsibility to handle the guest communication and the business operations but the AI is kind of handling the easy stuff that the human has already set the second is where the AI completely executes the all of the workflows and kind of processes that the human has put in place. 00:49:36.719 --> 00:49:45.199 The AI is just completely the human's hands off except when needed and the AI is implementing whatever the human told them to do. 00:49:45.519 --> 00:49:49.679 And the third and we're sort of transitioning from the first phase to the second phase right now. 00:49:49.920 --> 00:50:21.920 And in the future I think you know in the next 18 months we're going to start to see this as well where instead of the human telling the AI what to do and the AI executing the human's uh rules the AI is actually coming up with its own rules and its own policies that it's coming up with based on its own predictions about how to maximize Rev car reviews all of that and so we're gonna we're gonna have AI actually be doing the decision making in the future. 00:50:22.159 --> 00:51:04.800 So Arlo can I can add a add into that so you're saying the AI will start creating its own skills if if I its own policies so like if you think of a policy as like if this happens then I should do this the AI is going to come up you know I've never seen this scenario before but I just have a like just kind of making a judgment call that operators would make the AI is going to be able to make those judgment calls uh and be use that use that autonomously is is that with a with a human in the loop that approves this workflow for example not in the future no the AI we're

gonna just we're just gonna it's gonna trust it like we trust a member like we trust a business partner. 00:51:05.199 --> 00:51:53.119 And you're gonna be trusting more and more what would you say is the um needs to happen um uh for it to to grow to that to come to that stage uh is it the DLMs that need to go uh to to become much better um or what what needs to change before it gets to that level yeah I think um if you train models um having seen lots of different scenarios uh so having seen if the guest looks like this then what should I do um that's that's what you know we're gonna be seeing that kind of thing where like the AI is actually smart enough because it's seen people like this before um and the LLM is also smart enough given enough information I've seen situations like this before I can actually know what what to do how to step in. 00:51:54.719 --> 00:51:54.960 Okay. 00:51:55.920 --> 00:51:56.880 Yeah what do you think? 00:51:57.840 --> 00:52:25.599 I also think that uh um one of the things that uh would need to be improved is the um the access to certain things for example uh the AI would need access to uh the OTAs uh to be able to deal with things like cancellations uh let's say on booking.com or or other things so that the integ the the AI platform needs to have access to all the underlying integrations and actually have full access to all of these things in order to fully automate everything. 00:52:26.239 --> 00:52:28.480 That's gonna be a big part of it as well. 00:52:29.280 --> 00:52:53.360 Yeah that's a very interesting take there um and see how that evolves Jay what is what is your opinion in that I mean that's that's something that we're all thinking about who creates the future standards and and how are they being created and by whom are they created right and and how much do we still uh be part of this hospitality experience ultimately ourselves. 00:52:53.679 --> 00:53:24.400 Yeah and I I think um you know when when people see that there's AI guest communication they think um a lot of times you know you just plug it in it does all the work for you it's automatically you know automating all of your guest messages and that's really not the case like it can do the easy stuff because it's pulling in all the information from your PMS from all of your data sources your guidebooks so it has all this stuff but until it actually is instructed on how to take on those fringe cases or direct SOPs on how to handle certain situations. 00:53:24.480 --> 00:53:26.400 How do you handle early check-in, late checkout? 00:53:26.480 --> 00:53:29.920 What if a key's lost you know things like this it's not going to be able to handle those questions. 00:53:30.000 --> 00:53:32.480 So you're gonna continue getting those escalation escalations. 00:53:32.639 --> 00:53:40.559 I think what what Arlo's uh alluding to here with with the AI kind of understanding those policies um for us we just call them SOPs. 00:53:40.719 --> 00:53:51.039 So if it's understanding oh this is how we handled it in this situation let's recommend uh that we add this to the to the SOP the knowledge base moving forward so that it's you know we don't make this mistake again. 00:53:51.199 --> 00:53:55.039 I think adding that all into the workflow is something that's happening very, very soon here. 00:53:55.440 --> 00:54:03.199 The manual approval of those SOPs in this in the recommendations of what we should do uh or what we should add to the knowledge base is already here. 00:54:03.519 --> 00:54:12.400 We're already seeing that I think the it it does take another step for it to actually implement that automatically and for us to give it the trust to just come up with its its solution. 00:54:12.559 --> 00:54:33.599 So I think it's an interesting topic um to take a step back on that though I would I would mention that um I think as far as like the human component in uh in AI and in short-term rentals um I would say that the the guests are beginning to not actually care that it's AI as much as you might think that they do. 00:54:33.840 --> 00:54:37.119 I think people are looking for prompt answers that are accurate. 00:54:37.280 --> 00:54:50.880 And if they think it's an AI, if they don't know it's an AI, really it doesn't matter as much if it's coming through text, but there is a human component when someone's picking up the phone to go and call you as a host a lot of times they're looking to talk to an actual human. 00:54:51.039 --> 00:54:56.639 And so obviously you're still going to need to have some people on your team to to deal with those situations because there's an emotional component to it. 00:54:56.800 --> 00:54:58.639 People want to make sure that they feel heard. 00:54:58.719 --> 00:55:04.800 And the only AI voice that that I've seen and I've experienced it's pretty obvious that it's AI. 00:55:04.960 --> 00:55:21.920 So I think we're as we continue to move into that direction, I think there's an opportunity for there to be replacement but I think there's always going to be a lot of people specific specifically guests that are looking for that sort of like emotional connection between the guest and the host um but I'm excited to see how that changes over time. 00:55:22.239 --> 00:55:24.480 Because I think there's a big opportunity there as well. 00:55:24.880 --> 00:56:06.639 Can you pinch yourself for a second Jay please yeah I'm glad you're saying that and I I I could I could not agree with you more and I think that's hard for everyone in in any aspect of of life right in terms of news in terms of pictures we see and social media everybody who has children in their teens would probably put um put uh social media to hell for all the right reasons that we can't even uh think between what is fake and what is real uh with the information that we're given and I think that's where I see the human capital uh coming back into our industry without a doubt for exactly that reason. 00:56:07.039 --> 00:56:10.000 And

now it's it's about us finding out these balances. 00:56:10.079 --> 00:56:42.000 And and an interesting point uh Jay that you also mentioned which is great because we talked about that at VRMA executive uh a few weeks ago you know how do we go about this uh in terms of uh AI adoption and and one thing was very clearly you know how many property management companies don't even have their SOPs today and now SOPs all of a sudden become fundamentally important in relation to your brand, the way you want to execute your business, who you want to be for, your homeowners and your guests and everything else. 00:56:42.159 --> 00:57:04.239 And now SOPs become material finally because you're forced to have SOPs and and processes and and clear structures in an organization because otherwise you're gonna turn the entire business into a uh absolute uh mess if you don't do that and to start adopting AI it definitely will not go into the right direction before you do that properly as well. 00:57:05.119 --> 00:57:27.280 So looking at time I want to wrap this up and I'm I'm gonna have some closing questions uh ready for you uh going forward and I want to start with Anthony one AI tool you wish existed in this industry but doesn't yet oh that's uh that's a good question. 00:57:27.519 --> 00:59:26.719 Um I I think there's no tool yet um like Arlo uh I I think you explained it a little bit that you wanted to introduce is something that we are focusing on in guest we as well um that is able to um correctly identify um uh issues from guests escalated to sort of uh Monday uh equivalent boards that uh the internal teams can then um collaborate on the AI can then identify oh this shower was leaking already three months ago um this was the task that we assigned and we resolved it this way like uh an a sort of AI that operationally uh combined with the guest communication manages all of these things and and gives a grant overview this is something that uh that I would be very interested in uh at the moment as an operating company excellent thank you Anthony Arlo the most overhyped phrase in AI right now agent nobody knows what it is give us some give us some it's agentic I just it's like what is what is an AI agent like it's what is it what is it there's no definition to it people just kind of fill in whatever they want all right well love it um jay one metric every operator should track on guest comms and almost nobody is I would say guest sentiment level and I don't know that nobody is but I think that's a huge part of it if you're not tracking the actual guest sentiment level and the experience that they're having then you don't know how to improve your your uh your management of the properties how to improve your guest communication if it's just sending standard messages back and forth but you're not really getting that positive guest sentiment there's some adjustments adjustments that can be made but um if it's a good AI it's going to help you improve that guest sentiment level but it all starts with tracking it to begin with. 00:59:27.199 --> 00:59:32.239 Awesome thank you Jay Anthony do you wanted to say something? 00:59:32.639 --> 01:00:26.000 Yeah we um it's it's an important metric for for us as well we we do track it in in our platform and a lot of our clients are actually using it to automate certain things for example uh send out uh a review request uh after checkout and we see huge uh increase in uh in amount of reviews but also positive reviews because you will only ask it to people who are satisfied so I think it is a very important metric to track uh indeed Jay yep yeah we got the same thing I love that too you avoid having those negative reviews go out because you're just not mentioning to them so there's a lot of ways that you can implement that through and just going to ask them to give you maybe uh a review on a different platform uh obviously abiding by the OTA policies and things like that but figuring out who's having the best experience and really capitalizing that on that to make them like your biggest fans and uh reaching back out to them after six months, a year uh on specific events and things like that. 01:00:26.079 --> 01:00:27.599 So so much opportunity there. 01:00:28.079 --> 01:01:29.199 100% that we we even see some of our clients implementing it the other way that if clients are unsatisfied they uh they just stop the communication so they don't check out instructions check out instructions anymore or any other follow-up uh um automated messages so in the hope that uh a review doesn't come or that yeah so I I I do think uh there's a lot of value in it Arla give it a last shot agree or disagree in terms of metric uh about sentiment is that yeah not just in general what other metric is important oh uh yeah sentiment I mean the likelihood we um I have a sentiment score we also have a likelihood to leave a bad review which is kind of a little bit different from sentiment because there's certain attributes of guests where uh you know if the property maybe doesn't meet the expectation or even the weather impacts whether they're likely to leave or bad review or if they don't say anything sometimes they are likely to leave a bad review with certain requests. 01:01:29.440 --> 01:02:05.280 So we we've gotten into that uh probability of even a bad review um that's really important as well and like what what these guys both said like definitely um definitely really important perfect gentlemen I can give you all great reviews that's the conversation four people building the layer between between the operator and the guests the four very different bets on where it goes if you run an STR business and you're still doing guest comms the way you do it in 2022 this episode is your signal to look again. 01:02:05.519 --> 01:02:17.360 Arlo Anthony Colin Jay thank you so much for participating today this was a good one I'm Simon Lehman your host this is STR Global unlocked and we'll see you next time. 01:02:17.599 --> 01:02:20.159 Thank you very much for participating gentlemen. 01:02:20.480 --> 01:02:38.880 My biggest takeaway from this conversation is this AI will not replace hospitality but it will change what great hospitality looks like the operators who win will be the ones who know what to automate, what should stay human and how to use technology to create a better guest experience. 01:02:39.119 --> 01:02:48.559 And once you understand how AI is changing the guest side of hospitality the next question is what this means for the business behind it. 01:02:48.880 --> 01:02:52.880 This is exactly what I discussed with Julie Brinkman, CEO of Beyond. 01:02:53.199 --> 01:03:02.800 We talked about revenue management, profitability, market dynamics and why operators need to become more strategic about their numbers. 01:03:03.039 --> 01:03:10.960 So if you want to connect the guest experience side with the revenue and profitability side watch that episode next.