Latest / Brand Insiders / Sloan Dean of AI Hospitality Group on Rebuilding Hotel Management From First Principles

Transcript

Sloan Dean of AI Hospitality Group on Rebuilding Hotel Management From First Principles · Brand Insiders

• 0:12
  SPEAKER_00: Welcome to another episode of HospitalityNets Brand Insider.
• 0:17
  SPEAKER_00: I'm joined today by Sloane Dean.
• 0:19
  SPEAKER_00: He's the founder and CEO of AI Hospitality Group.
• 0:24
  SPEAKER_00: Sloan, how are you?
• 0:25
  SPEAKER_00: I'm doing great.
• 0:26
  SPEAKER_00: Thanks for having me.
• 0:28
  SPEAKER_00: Well, it's uh great to have you on to talk about this pivotal moment in the history of hotel management, let's say, because everyone's talking about AI tools, AI workflows, AI services, uh, but it hasn't really extended that far to fully revolutionize the management side of hotels.
• 0:54
  SPEAKER_01: We're the first AI native service provider in hotels.
• 0:57
  SPEAKER_01: And so to be an AIS company, which would be a term you hear more and more about coming out of Silicon Valley, you really have to have three things.
• 1:05
  SPEAKER_01: One, that you've always been AI native, like so it's mostly startups.
• 1:11
  SPEAKER_01: Secondly, that you uh everything that you're you're doing is either trying to bring about cost savings, uh doing something faster, better, cheaper than the incumbents in the spurt the space.
• 1:25
  SPEAKER_01: And then you're an outcomes-based provider, you're not a fixed monthly fee SaaS provider, or you're not just taking a percentage of gross revenue that you're being paid on outcomes.
• 1:37
  SPEAKER_01: And so we're a profit share uh model.
• 1:40
  SPEAKER_01: That's how we get paid.
• 1:41
  SPEAKER_01: So as we make more GOP for the hotel owner, we take a percentage of that.
• 1:45
  SPEAKER_01: So it's not only are we dramatically changing how we operate, we're also dramatically changing how we get paid as well.
• 1:53
  SPEAKER_00: Hmm.
• 1:54
  SPEAKER_00: So realistically, this isn't a one-for-one other management company with some AI tools.
• 2:00
  SPEAKER_00: This is rethinking the entire business model around.
• 2:04
  SPEAKER_01: That's right.
• 2:04
  SPEAKER_01: It's getting back to first principle thinking and thinking about well, what does a human resources department look like?
• 2:10
  SPEAKER_01: Do you need a res uh human resources department for a hotel operating company?
• 2:14
  SPEAKER_01: What does accounting look like?
• 2:16
  SPEAKER_01: What, you know, rather than what most humans do is they take a current workflow or a current job and then they try to increase the productivity of it or do what I call step change.
• 2:27
  SPEAKER_01: Instead, we get back to, okay, if we were to rebuild from nothing, uh, what does a hotel operating company look like in a world of AI and in a world where you're preparing for robotics?
• 2:40
  SPEAKER_01: And that's how we've built ourselves.
• 2:42
  SPEAKER_01: So you get back to this first principle thinking that is reimaginative.
• 2:47
  SPEAKER_01: And you can't really do that if you're already operating at scale.
• 2:51
  SPEAKER_01: Because, you know, one, it's incredibly difficult to reimagine something that is big.
• 2:58
  SPEAKER_01: Secondly, uh your capitalization needs to be such that your investors understand you're spending, you may not be making money for some time, hence why I took venture capital.
• 3:09
  SPEAKER_01: And then three, uh, you really have to um go slow, solve one problem at a time, and look like a technology company reinventing a process uh before you try to scale up.
• 3:23
  SPEAKER_01: And and so doing that, um, there are some real competitive advantages to being small, organic, and venture-backed.
• 3:32
  SPEAKER_01: And that's exactly what we are.
• 3:34
  SPEAKER_00: Yeah, you're flattening the entire hierarchy of a lot of these embedded cultures and thereby deriving not only efficiencies in terms of costs, but also efficiencies in terms of how quickly you can deliver various services and operations.
• 3:50
  SPEAKER_00: Is that is that what you're saying?
• 3:52
  SPEAKER_01: Aaron Powell Yeah.
• 3:53
  SPEAKER_01: And to go further, you know, uh the coin, the term AI and S, AI native services was coined by uh the venture capital firm called Emergence out of Silicon Valley.
• 4:04
  SPEAKER_01: And they uh say, you know, it you have to be a McKenzie-like, you know, consulting operating company at the same time combined with like a technology company like Stripe.
• 4:16
  SPEAKER_01: So that's like, so you look like a software company, but you're a service provider.
• 4:22
  SPEAKER_01: And so it makes it doubly hard because you know, if you were to look at us today, half our payroll, or actually even more than that right now, is sophisticated full stack engineers that little that you typically would have at a software company.
• 4:38
  SPEAKER_01: And then we have hotel operators.
• 4:40
  SPEAKER_01: And it's the the blending of those two is where the the kind of special sauce happens.
• 4:46
  SPEAKER_01: Probably the most, the the company to study and learn from the most is probably Palantir, where you know, they actually go into their clients, do forward deployment of engineers, you know, if like the the Department of Defense, they go actually put a forward engineer in the department.
• 5:05
  SPEAKER_01: He does, they do custom solutions and then completely redo workflows.
• 5:10
  SPEAKER_01: Uh that's probably I would say the the closest parallel to what we're doing, but we're doing it for hotels.
• 5:19
  SPEAKER_00: Yeah, I I'd love to drill into that comparison.
• 5:22
  SPEAKER_00: I guess you're talking about their Gotham products that they have.
• 5:25
  SPEAKER_01: Um the fact that like with a lot of their clients, they'll actually embed a resource, you know, permanently and kind of reimagine it's nothing, you know, it's not off the shelf per se.
• 5:38
  SPEAKER_01: I think that's where people get AI wrong.
• 5:40
  SPEAKER_01: Sure, of course, like AI and simple functions are is, you know, can do tasks, you know, do a discounted cash flow or some type of financial analysis or draft an email pretty well.
• 5:53
  SPEAKER_01: But if you really want to get into full kind of workflow automation, you have to break down a process to its individual steps.
• 6:02
  SPEAKER_01: And then you have to decide when and how you inject intelligence into that process.
• 6:07
  SPEAKER_01: And then how what is the contextual data that you're feeding the intelligence for a repeatable outcome that is deterministic?
• 6:14
  SPEAKER_01: So one of the reasons you have, like, you know, uh you maybe you've used Claude or OpenAI and you go in and you prompt and you feed a similar prompt and you get a different output, is because it you're not giving it the right contextual data.
• 6:29
  SPEAKER_01: Um, and so we have no lack of data in hotels.
• 6:33
  SPEAKER_01: We just don't have it normalized and contextualized for AI to consume it.
• 6:37
  SPEAKER_01: And we've built that.
• 6:38
  SPEAKER_01: So we actually ingest data from the financial ERP, the PMS, the CRS, the RMS, the applicant tracking system, the HR platform, um, about 10 different systems.
• 6:50
  SPEAKER_01: And then we have a uh way to put that into a normalized and contextualized database, which we call a data lake, that is harnessed around the LLM.
• 7:01
  SPEAKER_01: And then that's how you build the agent substrate off of that is that you have that, all that data, all that consumer data, all that financial data, that then if you have a sales agent, you have a finance agent, you have a front desk agent, they can call on that and you structure it in a way that uh they have the information even more so than your on-prem humans have.
• 7:23
  SPEAKER_01: And so then you get very deterministic outputs that are repeatable.
• 7:28
  SPEAKER_00: So again, let's uh let's unpack that because this is huge in terms of rethinking the IT architecture and then what you can build on top of that.
• 7:37
  SPEAKER_00: So a hotel that has existing systems that they're using, PMX, CRM, CRS, you are then have a method that you've developed, proprietary or using a supplier, for extracting that data, landing it in the data lake so it's structured or unstructured as it is.
• 8:00
  SPEAKER_00: And then once it's in that data lake, you can then assemble that, transform it into a usable form for your own trained AI to point against, built directly off that data lake.
• 8:16
  SPEAKER_01: That's exactly right.
• 8:17
  SPEAKER_01: We call it uh AICOS or better short for the pulse.
• 8:22
  SPEAKER_01: And it's the uh contextual data lake that we've harnessed that is system or record agnostic.
• 8:28
  SPEAKER_01: So we've got strategic partnerships in financial ERP that we've not announced yet.
• 8:33
  SPEAKER_01: So we want run one financial ERP system record for the entire company because the operator controls that.
• 8:40
  SPEAKER_01: We have one strategic partnership in HR and payroll, which um because we have some very unique automation there.
• 8:46
  SPEAKER_01: So we have these systems of record that we choose and we run in every hotel.
• 8:50
  SPEAKER_01: But then when it comes to system of records like PMS, CRS, RMS, where if you want to run branded hotels, you have to run the systems that the franchise agreements dictate.
• 9:01
  SPEAKER_01: Or if you're an independent hotel, maybe you're taking a hotel and the owner has already spent a lot of money installing, you know, Muse, CloudBeds, Opera with Oracle, et cetera, and they don't want to rip out and replace, then we our uh OS ingestion works with any PMS, any CRS, any RMS.
• 9:23
  SPEAKER_01: And even if now ideally you have a really good API, and we've built dozens and dozens of really good ingestion APIs in two ways where the agents can even write back into the systems.
• 9:35
  SPEAKER_01: But the the AI is so good now that you can even build browser automation where even if it doesn't have an API or file transfer protocol, you can access the and the agent can act like a human on a browser.
• 9:49
  SPEAKER_01: Now that's not a preferred method.
• 9:51
  SPEAKER_01: You know, you want to be directly API into a system.
• 9:55
  SPEAKER_01: But yeah, we built that orchestration layer absolutely with the intent of being PMS system or record agnostic.
• 10:03
  SPEAKER_01: Because if you look at what like News and CloudBeds, who are both brilliant companies and very successful, they've built an agent substrate ecosystem off of predicated running their PMS, right?
• 10:18
  SPEAKER_01: Like so you kind of have to get all that value.
• 10:20
  SPEAKER_01: You have to have, and we the the world is a fragmented PMS world.
• 10:26
  SPEAKER_01: And then you can game theory and say, in the future, does the PMS even exist?
• 10:30
  SPEAKER_01: Right.
• 10:31
  SPEAKER_01: Like software becomes commoditized into the future.
• 10:34
  SPEAKER_01: And so we really uh the the real magic of what we've done is like the data, how we've structured it and normalized it and harnessed it.
• 10:44
  SPEAKER_01: And then the agent substrates, I would even say some of the agents we built, like we have a sales agent.
• 10:50
  SPEAKER_01: I think in the future that becomes commoditized.
• 10:52
  SPEAKER_01: You know, um, it's more around the data and how you structure that and use that powerfully.
• 10:59
  SPEAKER_01: Um, you know, that is quite, I would say, the special recipe, if you will.
• 11:05
  SPEAKER_00: Yeah, you know, it's it's always said uh data is the new oil, or don't let your data collect digital dust, which is to say we have traditionally we have all these silos of records that are not properly not properly synthesized together to derive business value in terms of analysis of who your guests are and uh real-time service delivery that's personalized.
• 11:30
  SPEAKER_00: So um I love what you said about how it this can unlock that next level of value, both in the analysis and integrating everything, but also it can help to avoid the vendor lock-in situation of uh, you know, if if all your data is in one place, it isn't siloed off and you it's usable, then you can you can rip and replace as you deem best.
• 11:58
  SPEAKER_01: That's right.
• 11:58
  SPEAKER_01: As system of records, as AI gets better and puts middleware solutions and or system of record solutions at risk, or those things evolve, and you maybe have a partnership that's not best in class, that you've got modularity in that, yes, you may have to rebuild some of the API, but frankly, that only takes a matter of days or weeks.
• 12:22
  SPEAKER_01: Um, you can kind of rip out a system of record and install that, and the AICOS orchestration layer and the agent substrates that come off of that still fully work interdependent of the system of record.
• 12:39
  SPEAKER_00: Yeah.
• 12:40
  SPEAKER_00: So I I hope that's clear to everyone listening just how clear as clear as mud, as my grandmother would often say.
• 12:50
  SPEAKER_00: So But it's uh it's it's rewriting the entire IT idea of IT uh to make it instead of uh software oriented, now it's data-oriented, truly data-oriented.
• 13:03
  SPEAKER_00: And then as you said, even the agents on top of that are gonna be uh, you know, there's gonna be a lot of stuff that's off the shelf that you can add in to feed off of the data, sales agents, marketing agents, um, et cetera, et cetera.
• 13:18
  SPEAKER_01: That's right.
• 13:19
  SPEAKER_01: And we've we've built where then you have this OS and this uh canonical data lake that then harnesses the LLMs.
• 13:28
  SPEAKER_01: And as the LLMs get better, cheaper, open source, it even becomes more powerful.
• 13:34
  SPEAKER_01: And then the agents are a substrate of that.
• 13:38
  SPEAKER_01: And you can build all these, you know, and in some cases you maybe partner on the agent side, right?
• 13:43
  SPEAKER_01: You may say, well, procurement is very complicated and it requires a lot of connectivity to different distributors.
• 13:49
  SPEAKER_01: I have a GPO agreement, then I have to stay within all these standards.
• 13:53
  SPEAKER_01: And so maybe I'm gonna go partner for the procurement agent, but I want it to have this two-way with my financial ERP and my data lake.
• 14:00
  SPEAKER_01: Well, we can do that.
• 14:02
  SPEAKER_01: And in that case, you may have, you know, an onward SAS cost with whoever that, but the challenge with if you're a current operator, a current brand, and you had do not have the technology capabilities to do what we're doing, you're going to be at the mercy of software companies who are going to go deeper into the PL value leakage, and they're going to siphon off like they, like you, you know, if you look at the hotel value chain, we've had the value leakage model away from the hotel owner for the last couple of decades.
• 14:36
  SPEAKER_01: You know, the OTAs get wealthier, the brands are doing really well, software companies continue to kind of proliferate, and even managers are very like fragmented, but you know, do fairly well as a percentage of gross revenue.
• 14:50
  SPEAKER_01: And you're you're basically outsourcing innovation to software companies.
• 14:54
  SPEAKER_01: And so what we've done is we we're vertically integrated.
• 14:58
  SPEAKER_01: Um and we think that that that allows for speed and innovation because you're the actual record of employer so that you can force how the AI is used day to day, and then you can change the labor standards of the human interacting, and you can constantly do that in an evolving way.
• 15:15
  SPEAKER_01: And it allows you to spin that flywheel faster than if you're just the employer, you're just the software provider.
• 15:21
  SPEAKER_01: And if you are just an employer, which if you look at all the third-party operators today, they're commoditized staffing companies, they're gonna be outsourcing that innovation.
• 15:30
  SPEAKER_01: Uh, even if a few of them have a few uh wedge solutions, they can't build the whole ecosystem.
• 15:38
  SPEAKER_01: And so they're gonna have to outsource that innovation, and then that value goes to a software company.
• 15:43
  SPEAKER_00: Yeah, you know, uh I want to take a step back now.
• 15:46
  SPEAKER_00: It's it's a little bit mind-blowing in terms of rewriting the actual rules of what we think of as management.
• 15:54
  SPEAKER_00: But, you know, you talked about existing management companies as outsourced staffing agencies, and this has come from the person who ran Remington Hospitality.
• 16:04
  SPEAKER_01: Yeah, I and Remington's an amazing company.
• 16:06
  SPEAKER_01: I was, I don't, I, you know, hey, I love Remington and feel uh very blessed to have been their CEO and president for eight years.
• 16:16
  SPEAKER_01: Uh, but the third-party management company landscape globally is Ainbridge any different than Highgate, any different than Pyramid, any different than Crestline?
• 16:26
  SPEAKER_01: Sure, there are some that are slightly better than others, but you know, if you talk to any institutional capital, they will tell you that entire space is commoditized.
• 16:37
  SPEAKER_01: Um and it's why you have a race to the bottom on fee because there's no true differentiation.
• 16:44
  SPEAKER_01: It's why you have ease of convenience determination because the managers have no leverage and value prop.
• 16:50
  SPEAKER_01: And it's also why, you know, you have a lot of contract churn.
• 16:55
  SPEAKER_01: You know, the average third-party management company churns anywhere from five to 12% of its portfolio annually.
• 17:00
  SPEAKER_01: And that is because there's not a lot of differentiation.
• 17:04
  SPEAKER_01: And so, and it's why you have tremendous fragmentation in the space because, you know, no one has more.
• 17:11
  SPEAKER_01: Even Ambridge at a thousand hotels doesn't even have 1% of the TAM.
• 17:15
  SPEAKER_01: So you have extreme, and that typically happens in commodity, commodities business, right?
• 17:20
  SPEAKER_01: Like where you have businesses that become a huge juggernaut or a monopoly in a business, they tend to have like a very deep competitive moat, right?
• 17:30
  SPEAKER_01: That doesn't exist in hotel management, which is good for like getting a staffing company that's priced and and you know, you're there's not big margins in that business.
• 17:43
  SPEAKER_01: And so it's maybe cost effective for hotel owners, and you have a lot of options to shop, et cetera.
• 17:49
  SPEAKER_01: It's like buying energy is like the a different energy company different than the next.
• 17:54
  SPEAKER_01: Uh, but I think it's time for a change.
• 17:56
  SPEAKER_01: And we see ourselves as, you know, we're the only AI and S option in the space, and uh, we are very differentiated and we plan to up in the industry.
• 18:08
  SPEAKER_01: And, you know, it's it's very akin to like, look, the cab industry was very commoditized, stale, had no innovation, was a bad experience.
• 18:16
  SPEAKER_01: I'm not saying that's the hotel industry because there is a lot of good hotel products.
• 18:20
  SPEAKER_00: Yeah.
• 18:21
  SPEAKER_01: But, you know, operating as operating staffing model when I got in the business 21 years ago, it's 90% of what it is today.
• 18:28
  SPEAKER_01: So there's not been any real innovation on-prem.
• 18:32
  SPEAKER_01: And we are leading the edge to be the Uber of hotels.
• 18:36
  SPEAKER_01: We really want to bring technology in a transformative and disruptive way to reimagine how hotels operate and the profitability as such, and frankly, increase, increase associate satisfaction and guest satisfaction at the same time.
• 18:50
  SPEAKER_01: It's not just a cost savings metric.
• 18:52
  SPEAKER_01: You know, if you can harness and, you know, like you said, the data, there's all kinds of things you can do to personalize the guest stay.
• 18:59
  SPEAKER_01: You know, there's also all types of things to take away mundane tasks that suck on property.
• 19:05
  SPEAKER_01: You know, nobody got into a hotel and said, I want to do accounts payable.
• 19:09
  SPEAKER_01: You know what I mean?
• 19:10
  SPEAKER_01: So, like if we can automate all that and then free up the humans to go be hosts, which is why you got into the hotel business to begin with, we can actually make the jobs more satisfying.
• 19:20
  SPEAKER_01: So I, you know, I it's not simply, hey, let's save money and cut labor and expand the margin.
• 19:26
  SPEAKER_01: Yeah, of course, that's the economic benefit.
• 19:28
  SPEAKER_01: But I think there's a lot of like inspirational things that we can do to reimagine the business.
• 19:34
  SPEAKER_00: And that's what we're trying to do.
• 19:36
  SPEAKER_00: Yeah, you know, uh, but to push back a little bit, uh, the economics are still critical here.
• 19:40
  SPEAKER_00: And oh, absolutely.
• 19:42
  SPEAKER_00: You know, uh, like uh as as I see it is we make the economics work, and then from there we innovate, we evolve, we nichefy, all that fun stuff we want to do, but we can't put the cart before the horse, if that makes sense.
• 19:57
  SPEAKER_01: We've already proven out that we can take a full service hotel and move its GOP margin 500 basis points.
• 20:04
  SPEAKER_01: And that is transform that is transformative value creation.
• 20:08
  SPEAKER_01: Apply a cap rate.
• 20:09
  SPEAKER_01: And that is why we will take off like a rocket ship.
• 20:12
  SPEAKER_01: Um that, but that is not the that is not the all end all be all of what we're doing.
• 20:18
  SPEAKER_01: But um, and what makes that really hard, I I tell the team, it's adding up dimes to dollars.
• 20:24
  SPEAKER_01: Outside of our sales agent transforming group sales in a full service hotel.
• 20:31
  SPEAKER_01: And outside of uh the orchestration layer, leading to some position savings in middle management in a full service hotel, there's really no individual thing that moves the margin 100 basis points or more.
• 20:49
  SPEAKER_01: Those two items alone, you know, if you have a hotel that a third of its revenue is grouped, bringing in the AI sales agent to actually accelerate group bookings and conversion rate.
• 20:59
  SPEAKER_01: And because the flow through on that inclemental revenue is so high, can move your margin a hundred bits or more alone.
• 21:06
  SPEAKER_01: And then the orchestration layer can lend itself to, okay, maybe I don't need the a few middle management positions, which could add up to a hundred basis points.
• 21:18
  SPEAKER_01: You will limit, but it a lot of the things are, okay, this saves 10 basis points, or this save saves 20.
• 21:24
  SPEAKER_01: So it's doing a lot of little things and the coordination of those things.
• 21:30
  SPEAKER_01: Uh, we we actually have a product roadmap to where we want uh over 75 AI agents all coordinated doing various things.
• 21:39
  SPEAKER_01: Because if you look at work at an individual hotel, like you take a middle manager, for example, like just say an AGM, they in one day, and if you watch their work, may do 30 or 40 things in the day.
• 21:55
  SPEAKER_01: And so if you want to get to a point, hey, I want part of that role to be reimagined or Transform, we have to go through and automate all these individual things because we have a lot of blended work that happens in hotels.
• 22:09
  SPEAKER_01: And of course, then you have a lot of physical work that happens that frankly you can't even touch until you get into robotics.
• 22:16
  SPEAKER_00: Yeah.
• 22:17
  SPEAKER_00: You know, the whole idea here, um, you know, we're seeing this on the ground with departments are doing uh commercial strategy instead of silos of sales, marketing, comms, revenue.
• 22:30
  SPEAKER_00: So that's happening.
• 22:31
  SPEAKER_00: But uh from what I gather, you're taking it one step further, which is to say, okay, well, that middle manager, uh, whether that's a headcount that we're trying to reduce, but realistically we're taking all the stuff, we're re-rethinking how they're working.
• 22:48
  SPEAKER_00: We're putting some AI workflows and agents behind everything.
• 22:52
  SPEAKER_00: And then that middle manager, instead of just responding to RFPs all day and emails, is actually more guest-facing for those groups and can do more virtual tours, more meetings, just upvaluing their work to help help them close more.
• 23:09
  SPEAKER_00: Is that is that what you're saying in a lot of ways?
• 23:11
  SPEAKER_01: That's exactly right.
• 23:12
  SPEAKER_01: Let me let me put it into like an actual numbers example.
• 23:17
  SPEAKER_01: So uh take a full service hotel that does about$30 million a year in revenue, has 120, 140 employees of that associate pool, about 20 are salaried managers.
• 23:33
  SPEAKER_01: Okay.
• 23:34
  SPEAKER_01: In that situation, we can operate that same hotel in an increased revenue, increased guest satisfaction level with 12 to 13 managers instead of 20 to 22.
• 23:47
  SPEAKER_01: So you're saving, and let's say the average full load of that position is$100 to$150,000.
• 23:54
  SPEAKER_01: So let's call it for rounding a million dollars in full payroll.
• 23:59
  SPEAKER_01: That's that's uh, but by the way, the 12 or 13, 14 you have are empowered to do even more.
• 24:07
  SPEAKER_01: And the GM at that hotel is empowered to do even more, where they are harnessed with AI and we call it 10Xing.
• 24:15
  SPEAKER_01: So the positions, and then by the way, you can pay them more because they're doing even more work and they have it can develop their career even more.
• 24:24
  SPEAKER_01: And so some of that savings you unlock, you give back to in higher wages.
• 24:29
  SPEAKER_01: So the work becomes more meaningful and you can pay them more.
• 24:32
  SPEAKER_01: That's one that's item number one.
• 24:34
  SPEAKER_01: The second is that you have departments like housekeeping, which is the largest department in most any hotel, laundry, kitchen, et cetera, where uh you can bring in AI to make those departments anywhere from five to fifteen percent more efficient per pre-robotics.
• 24:54
  SPEAKER_01: So pre-pre, you know, augmentation of physical labor, or if you're doing like an arbitrage between robotics and physical, pre-that, you can bring AI to make any department five to fifteen percent more efficient.
• 25:07
  SPEAKER_01: And so there's a that's item number two.
• 25:10
  SPEAKER_01: Assuming you're not unionized and under a collective borrowing agreement that requires all these rules sets.
• 25:16
  SPEAKER_01: So that that's a caveat.
• 25:17
  SPEAKER_01: And then the third, and this is this is a gross oversimplification, but you can bring AI to commercial strategy and expanding your revenue, expanding not only Room's revenue, but other departments, and move the top line.
• 25:32
  SPEAKER_01: And those are really the big, like that is a gross oversimplification, but that's like if you really want to kind of blow replate it down to like, okay, what are we doing?
• 25:41
  SPEAKER_01: That's where you really move the PL in those three categories.
• 25:46
  SPEAKER_00: Yeah.
• 25:46
  SPEAKER_00: So we have those three categories, and a hotel has an existing uh method of operating, uh, be it there's ownership, there's a TPO, or there's in-house uh management.
• 26:00
  SPEAKER_00: They say, Sloan, uh I love every single thing you said, 5% GOP increase uh year over year.
• 26:09
  SPEAKER_00: I can do so much with that in terms of finally paying off that loan, uh, or you know, to make my death service coverage or to implement that new PIP, that property improvement plan that I had had in my dreams.
• 26:22
  SPEAKER_00: Uh what are the first steps for getting underway?
• 26:26
  SPEAKER_01: Yeah.
• 26:26
  SPEAKER_01: So the first thing is let it reach out to, well, by the way, it's not just even 5% GOP expansion.
• 26:32
  SPEAKER_01: Let's the average full service GOP margin in the United States is 32 to 33%.
• 26:36
  SPEAKER_01: So we're moving it from that to 38, which is a percentage gain of uh, you know, over 10%.
• 26:44
  SPEAKER_01: So, you know, 10% would be 4%.
• 26:47
  SPEAKER_01: But um uh, you know, if you're interested in AIHG.com, what we do is we go through a very extensive AI underrating, AI underwriting process where we basically say, hey, it's a lot more intrusive.
• 27:04
  SPEAKER_01: Like we need your staffing model, we need the tech stack you're on, uh, we need access to some non-customary, non-traditional data.
• 27:13
  SPEAKER_01: And then we'll tell you, hey, this is a five-year performa and stair step the margin.
• 27:20
  SPEAKER_01: Now, I what I would tell you is there are some of the things you can implement day one.
• 27:24
  SPEAKER_01: Um to get destabilized expanded GOP on average is about six months.
• 27:31
  SPEAKER_01: Okay, so people always want to know how long does this take?
• 27:34
  SPEAKER_01: Okay, and because you're having some of the things require uh installation, iteration, training, iteration, training, iteration, training.
• 27:46
  SPEAKER_01: Um and it's not is that's what makes this really hard.
• 27:49
  SPEAKER_01: That's why you need to be the employer.
• 27:51
  SPEAKER_01: You can't just be a software provider where you go sell the product, they use the product incorrectly and you don't iterate on it.
• 27:57
  SPEAKER_01: One of the things that's really eye-opening is when you start doing this at scale, is that the labor standard of the humans you have evolves.
• 28:07
  SPEAKER_01: And so what most operators do is they have a labor standard for a position, they set it, they budget against it, and they measure against it.
• 28:15
  SPEAKER_01: Okay.
• 28:15
  SPEAKER_01: And then they will benchmark hotels maybe over a three-month period or six-month period or a year, they change their labor standard, try and enhance it.
• 28:23
  SPEAKER_01: Well, in this world, as you're automating work, the humans you do have, which is still your biggest expense on your PL, evolves, and you have to evolve with it.
• 28:35
  SPEAKER_01: And that is very difficult.
• 28:37
  SPEAKER_01: So, like the labor standards are a fluid thing.
• 28:40
  SPEAKER_01: Um, and I think that's something that no one has really studied it, like we have, in like how the humans in the loop, we call it, evolve along with the technology and how you utilize them.
• 28:55
  SPEAKER_00: Yeah.
• 28:55
  SPEAKER_00: Uh two things to unpack.
• 28:57
  SPEAKER_00: They're number one, human in the loop, and this whole idea of humans becoming loop engineers.
• 29:03
  SPEAKER_00: Do you have any thoughts on those two terms that we can unpack?
• 29:07
  SPEAKER_01: We well, what's funny is Eve Moore, who's our chief operating officer, who's great.
• 29:11
  SPEAKER_01: Yeah.
• 29:12
  SPEAKER_01: Uh, she's got a shirt that says human in the loop officer.
• 29:16
  SPEAKER_01: Uh, it's kind of like a running joke because we have an engineering team.
• 29:19
  SPEAKER_01: She's always giving the engineering team feedback on uh operating screens, and the agent did this, and like we need to change.
• 29:25
  SPEAKER_01: So we like, you know, those of us that are actually operators, we it's kind of an inside joke of human in the loop.
• 29:31
  SPEAKER_01: Uh, we also I like the reframe of like the humans are the the host in hospitality, right?
• 29:37
  SPEAKER_01: The thing that AI cannot do, and I frankly kind of hope it can't do ever, is have high empathy and relatability.
• 29:47
  SPEAKER_01: And like when you're showing up at the front, like do we really want a robot checking us in at the front desk?
• 29:52
  SPEAKER_01: That's I think that's a curious question to ask.
• 29:54
  SPEAKER_01: I think we will have the full capability of go straight to the room, full, you know, mobile key is still already exists, but like biometric, skip the desk if you want that.
• 30:05
  SPEAKER_01: But if you want the human interaction, let's just say you paid$1,000 a night at a lecture.
• 30:11
  SPEAKER_01: I mean, you probably kind of want somebody you had a shitty flight to be empathetic when you arrive.
• 30:17
  SPEAKER_01: And that like there's something to be said about great service.
• 30:21
  SPEAKER_01: And that's what we really want to instill too.
• 30:23
  SPEAKER_01: We don't want to lose the essence.
• 30:25
  SPEAKER_01: So I uh, but yeah, the human in the loop is kind of an inside.
• 30:29
  SPEAKER_01: And for sure, it's like your associates are helping improve the technology and iterating.
• 30:36
  SPEAKER_01: And by the way, they're benefiting by a higher wage by also being educated in AI.
• 30:41
  SPEAKER_01: And it creates this flywheel that you can't replicate if you're just a software company, and you can't replicate if you're just a commoditized staffing company.
• 30:51
  SPEAKER_00: Yeah, it's great that you drive it back to humans and this whole idea of upvaluing the work that's done.
• 30:59
  SPEAKER_00: Yeah, from there, um I think that's really important.
• 31:02
  unknown: Yeah.
• 31:03
  SPEAKER_01: Because I think I have several YPO friends who own uh AI advisory businesses.
• 31:10
  SPEAKER_01: And I think what has is happening is you have like Fortune 500 businesses hiring AI experts to come in.
• 31:16
  SPEAKER_01: And they're looking at it as this arbitrage between human labor and AI.
• 31:21
  SPEAKER_01: And there's not this reimagination and this upwork happening and the inspiration around the humans that stay.
• 31:27
  SPEAKER_01: Because, you know, we're and think about hospitality.
• 31:30
  SPEAKER_01: You're in hospitality because you love people, you love travel, like you want to inspire people.
• 31:34
  SPEAKER_01: Like if all of a sudden you're a hotel operating company, you got to go, you say, to be competitive in this new world, I got to lay off half my people.
• 31:41
  SPEAKER_01: That's not really inspiring.
• 31:43
  SPEAKER_01: And by the way, we make it known we're gonna operate hotels with less people than anyone else.
• 31:48
  SPEAKER_01: I make that known.
• 31:50
  SPEAKER_01: But the inspiring thing is the humans we do have, we're gonna have it be more inspiring work, more thoughtful work.
• 31:56
  SPEAKER_01: And by the way, we're gonna train you in AI so that you're competitive in the future.
• 32:01
  SPEAKER_01: And I think you have to have that reframe because, you know, otherwise, why aren't we doing this?
• 32:05
  SPEAKER_01: You know, it's like, oh, well, we saved a bunch of money and you know, because then this the guest service is gonna suck because the humans you do have are like looking over their shoulder, well, they're gonna they're gonna eliminate my job next week.
• 32:18
  SPEAKER_01: You know, so I think psychology is a big part of this.
• 32:22
  SPEAKER_00: Yeah, it's as you said, it depends on the brand, right?
• 32:25
  SPEAKER_00: There's gonna be the luxury brands where uh what we're seeing right now is basically a one-to-one relationship at that level between host or manager to guest.
• 32:38
  SPEAKER_00: And I see that going even deeper.
• 32:40
  SPEAKER_00: And I also do see the uh, you know, check-in kiosks with biometric and you know, uh, you know, eventually passports will be just just who you are.
• 32:52
  SPEAKER_00: You don't need this this outdated booklet.
• 32:54
  SPEAKER_00: Uh, and then you're checked in and you you automatically can go to your room, whether it's on your phone, uh, check in with an iris scan, whatever.
• 33:03
  SPEAKER_00: It's it's gonna go in both directions.
• 33:05
  SPEAKER_00: And then the back of house really is where you're realizing a ton of efficiencies insofar as automating things that used to take days to do, like monthly reports or um synthesizing the data to develop paid search campaigns, uh, everything of that nature where uh there's the the true efficiencies that you're adding to the GOP.
• 33:29
  SPEAKER_01: Yeah.
• 33:29
  SPEAKER_01: And I think what we're doing, particularly in the back of house where we're automating, is applicable to all categories of hotels.
• 33:36
  SPEAKER_01: I agree with you.
• 33:37
  SPEAKER_01: I think in the luxury category in particular, that you'll have this AI adversion.
• 33:44
  SPEAKER_01: You know, it's almost like what you saw in the White Lotus episode where they go to the Thailand resort and they lock their device in and they're like, and it's I I can see that where, you know, like a four seasons or Riz Carlton, like their whole shtick is we're still 100% by humans.
• 34:01
  SPEAKER_01: Like it's the the old, you're paying for this old school, like heavy intensive.
• 34:06
  SPEAKER_01: Uh, but let's face it, nobody needs a human doing hotel accounting.
• 34:12
  SPEAKER_01: No one needs uh a human doing hotel procurement.
• 34:17
  SPEAKER_01: So there are certain functions that I think automation across the board makes sense.
• 34:21
  SPEAKER_01: But I agree with you on the service delivery.
• 34:23
  SPEAKER_01: I think there'll be some, you know, there it may be the future that there are whole categories of hotels that are heavy AI and then others aren't.
• 34:30
  SPEAKER_01: And it's like this AI on, you know, it almost becomes a division of how hotels are set up, you know, in the future.
• 34:38
  SPEAKER_00: Yeah, it it becomes a division.
• 34:40
  SPEAKER_00: It's also intellectual property for the value of that brand itself in a lot of ways.
• 34:44
  SPEAKER_00: And you're adding to that.
• 34:47
  SPEAKER_00: Yeah.
• 34:48
  SPEAKER_00: So we've talked a lot about AI, how to get started.
• 34:53
  SPEAKER_00: Uh, what I'd want to go back to is this idea of data being the foundation and how you're essentially you're preventing the vendor lock-in because now we can have all the data in one place and the AI points directly against that data.
• 35:11
  SPEAKER_00: And then that's essentially the organization of the future, where you know, there may be a day when uh we don't need a PMS.
• 35:19
  SPEAKER_00: We don't need a spa system, we don't need that because it's all built upon the data lake.
• 35:24
  SPEAKER_00: Do you see that day coming soon?
• 35:26
  SPEAKER_00: Or is that not soon?
• 35:28
  SPEAKER_01: I think I think we will be uh in a world, I think it's once we have true agentic AI.
• 35:35
  SPEAKER_01: And then probably I think if you stare step this, I think we'll actually see robotics augmenting physical work before we get into a world where these system records disappear.
• 35:46
  SPEAKER_01: Um, I think it for sure, and you're in my career, we're you know, you're in your early 40s, I'm in my mid-40s.
• 35:52
  SPEAKER_01: Uh, I think in my work career for sure, like the PMS and some of these system records go away.
• 35:58
  SPEAKER_01: Um, is it in the next five years?
• 35:59
  SPEAKER_01: No.
• 35:59
  SPEAKER_01: Is it in the next 10?
• 36:01
  SPEAKER_01: Argumentative, in the next 20 for sure, for sure.
• 36:04
  SPEAKER_01: Um, and for sure in the next 10 years, you're gonna go into hotels and full physical departments will be fully automated.
• 36:12
  SPEAKER_01: So housekeeping, laundry, kitchen.
• 36:15
  SPEAKER_01: Uh you're already seeing in other industries and in China.
• 36:18
  SPEAKER_01: Uh absolute.
• 36:20
  SPEAKER_01: Now it won't be in the next one to two years because the cost of the hardware and the arbitrage between physical labor cost and but as that hardware cost comes down and also the AI gets better and the AI becomes the brain trust of the hardware, for sure that is going to happen.
• 36:36
  SPEAKER_01: And make no mistake, in the early 2030s, we'll go into hotels in the United States and all of housekeeping will be fully automated for sure.
• 36:44
  SPEAKER_00: Yeah, I think it deserves touching upon this idea of robotics because when if we look back 10 years ago before the Chat GPT moment, everyone thought that robotics was going to come before uh before deep before deep learning AI or even AGI at this point, we're yeah, we're getting that to that point.
• 37:05
  SPEAKER_00: But now it seems the other way that AI has really gone leaps and miles into knowledge work, but the housekeeper still with their uh tactility and their dexterity seems to be something that robotics has yet to replace.
• 37:22
  SPEAKER_00: Um to put a pro prediction in your mind, um AI itself, the design of these robots, has now accelerated the manufacturing of these robots itself.
• 37:36
  SPEAKER_00: And so we are, do you see a point where we will reach the point where we have robots as housekeepers in in the near future?
• 37:46
  SPEAKER_01: Uh yes, but it's not going to be one humanoid robot for one person.
• 37:52
  SPEAKER_01: That's where everybody gets the visualization wrong.
• 37:55
  SPEAKER_01: I had a Robux uh we're in conversation with some hardware companies.
• 37:59
  SPEAKER_01: I I won't name names because we haven't signed an agreement yet.
• 38:02
  SPEAKER_01: But uh the guy who's kind of out there, if you will, in robotics and China is doing the most progressive things, he he used this analogy.
• 38:11
  SPEAKER_01: He said, think Star Wars, not iRobot as a movie.
• 38:16
  SPEAKER_01: And so in Star Wars, all the worker robots, you have like the cart robot doing the beverage service, you got like the vacuum cleaner.
• 38:23
  SPEAKER_01: So what I envision is um you will have a few years of let's do in-room cleaning.
• 38:30
  SPEAKER_01: So you will have a purpose-built robot that does all the hard surface floor cleaning, you will have a purpose-built robot that handles trash linen management.
• 38:40
  SPEAKER_01: So the carts get heavily modified, and you strip away some of the hard physical work and where there's workers' comp issues, you know, nobody wants to handle soiled linen.
• 38:54
  SPEAKER_01: And so you kind of increase the quality of the job first, and then you hit an inflection point where all of a sudden three or four bots clean the full room.
• 39:03
  SPEAKER_01: It is not going to be figure AI's robot that costs$250,000 a day and has dexterity problems with glassware and it can't make a bed, it can strip a bed that's going to go in and one-to-one replace.
• 39:15
  SPEAKER_01: Now, in 10, 15 years, sure, maybe.
• 39:18
  SPEAKER_01: Like maybe if you wait long enough and it becomes cost effective, you can go buy that, capitalize that robot, and it the cash on cash payback is two years or less.
• 39:27
  SPEAKER_01: But we're a ways away from that.
• 39:30
  SPEAKER_01: And so it becomes it's easier to go and like take break a cleaning of a room down into like 10 different processes and then have like four or five dedicated robots to do different things.
• 39:42
  SPEAKER_01: And by the way, when you have bots doing multiple things, you can clean the room a lot faster too.
• 39:47
  SPEAKER_01: Right?
• 39:48
  SPEAKER_01: Because they're coordinating together.
• 39:49
  SPEAKER_01: It's not one humanoid.
• 39:50
  SPEAKER_01: So that's where everybody gets it wrong.
• 39:52
  SPEAKER_01: Everybody's like, oh, we're gonna show it one day and it's gonna be like Betty, the humanoid robot, cleaning my room.
• 39:56
  SPEAKER_01: No, it's not.
• 39:57
  SPEAKER_01: It's gonna look like Star Wars, and it's gonna start with innovative companies being leading edge and testing use cases.
• 40:05
  SPEAKER_01: And that's what we're doing.
• 40:07
  SPEAKER_01: We're gonna be, you know, we're not we're gonna have strategic hardware partnerships, but we're gonna be first mover in being like uh an operator at scale that's actually using robotics.
• 40:17
  SPEAKER_00: Well, you think about that.
• 40:18
  SPEAKER_00: If you're able to clean a room hyper fast and it's reliant on robot labor uh hybrid for the first part, and you combine that with all the workflow management to the back end, eventually we get to another holy grail, which is this whole idea of 24-hour room availability.
• 40:39
  SPEAKER_00: So, you know that's yeah, the whole idea like why does why is check-in always at 3 p.m.?
• 40:44
  SPEAKER_00: Why is checkout always at 11 p.m.?
• 40:45
  SPEAKER_00: Because of scheduling.
• 40:46
  SPEAKER_00: Yeah, exactly.
• 40:48
  SPEAKER_00: Human-based scheduling.
• 40:49
  SPEAKER_01: That's right.
• 40:49
  SPEAKER_01: But if a robot can work all the time in Vegas, can you turn the room twice in New York where you're running 100% occupancy?
• 40:56
  SPEAKER_01: Maybe somebody only needs the room and you don't have to call in the housekeeper.
• 40:59
  SPEAKER_01: Absolutely.
• 41:00
  SPEAKER_01: Uh I think that absolutely happens that you know it will unlock these other revenue potentials where, okay, maybe a hotel runs 130% occupancy level.
• 41:09
  SPEAKER_01: In New York, you have hotels that could run 130% if you could turn the room.
• 41:13
  SPEAKER_01: Um, you know, and then you have limitations that the PMS was structured as such to, you know, sell by 24-hour period, but that could easily change too.
• 41:21
  SPEAKER_01: Uh so absolutely, I think that happens.
• 41:24
  SPEAKER_01: It's just a matter of when, not if.
• 41:26
  SPEAKER_00: Yeah.
• 41:27
  SPEAKER_00: Uh and it it's a matter of uh, I guess making all the technology work together.
• 41:31
  SPEAKER_00: Um hardware and software.
• 41:34
  SPEAKER_01: Yeah.
• 41:34
  SPEAKER_01: One of the things that I think people don't think about is like, so what makes this really hard too, yes, the uh the AI has accelerated more, at least in the United States, more than robotics.
• 41:46
  SPEAKER_01: But uh you need, like for the robots, if they're doing more than two or three repeatable tasks, they need to have a brain.
• 41:56
  SPEAKER_01: Right.
• 41:57
  SPEAKER_01: And so the AI, if you the LLMs have become so good.
• 42:01
  SPEAKER_01: And then, but by the way, you need you need a lot, so for robots, use Waymo as an example.
• 42:08
  SPEAKER_01: It required millions and millions of hours of visual data.
• 42:13
  SPEAKER_01: But there's now new ways to train robots, but you still need contextual operating data that can tell them what to do.
• 42:24
  SPEAKER_01: Well, what is an operational OS that we've built to to to control and be able to route AI agents that don't do physical work?
• 42:35
  SPEAKER_01: What does it do really well?
• 42:36
  SPEAKER_01: It'll do the same for robotics.
• 42:39
  SPEAKER_01: So, like when you start introducing, so you that's what no one has built either, is that you actually need a brain for the robots to go do these specialized tasks.
• 42:52
  SPEAKER_01: Um so, anyways, I just call that out as something that we're thinking about too, is like, you know, you need like a a contextual data lake brain for uh hotel specific.
• 43:06
  SPEAKER_01: Or you just start changing the structural layout of hotels so that the automation is purpose-built into the hotels, which may happen in future construction too.
• 43:14
  SPEAKER_01: Like you go in instead of just a traditional laundry, the actual automation is built into the same in the kitchen, what you're doing in quick serve restaurants, that the automation is just built into the hardware.
• 43:24
  SPEAKER_01: And I think that will happen in hotels, but you got 60,000 hotels in the United States, like go tear all those down and rebuild them.
• 43:30
  SPEAKER_01: Purpose built for automation, good luck.
• 43:33
  SPEAKER_01: Uh, there's gonna have to be a lot, maybe in a hundred years, but uh, you know, anyways.
• 43:39
  SPEAKER_00: Yeah, I mean, change is slow.
• 43:41
  SPEAKER_00: Uh, you know, you have in ingrained processes and ingrained capital expenditures wherever they are, if they're, you know, the construction of how hotels are built.
• 43:51
  SPEAKER_00: And it looks like you're potentially developing a product around um redesigning hotels from the ground up so that way they're AI first as well.
• 44:01
  SPEAKER_01: Yeah, I do think we will uh, you know, I got a question asked the other day is like, will construction requirements and design protocols for hotels change?
• 44:11
  SPEAKER_01: And I think they will.
• 44:12
  SPEAKER_01: Um, like I said, I think it makes total sense that laundry, major mechanical, and um, you know, way you intake energy and build building management systems and the kitchen, that you just have purpose-built automation built into it.
• 44:27
  SPEAKER_01: And that's only two to three years away from that being at scale.
• 44:30
  SPEAKER_01: And so you'll re- you know, you'll reimagine how you build these hotels in the future and that that will change for sure.
• 44:36
  SPEAKER_01: The front of house, you know, I don't know.
• 44:39
  SPEAKER_01: Um, you know, and and you know, hopefully you don't lose some of the design.
• 44:43
  SPEAKER_01: You know, we've made a lot of leaps and bounds around design and and you know, lifestyle hotels and all this.
• 44:48
  SPEAKER_01: So I think the consumer will end up driving that.
• 44:51
  SPEAKER_01: But for sure, I think construction of hotels changes dramatically over the next 10, 15 years.
• 44:58
  SPEAKER_00: Yeah, one of the smaller ones that we're seeing is just the elimination of the of the physical front desk.
• 45:03
  SPEAKER_00: Right.
• 45:04
  SPEAKER_00: Uh you know, at the luxury end, you have the hosts who greet you there and then they'll check you in on an iPad uh or some other, you know, less obtrusive way.
• 45:15
  SPEAKER_00: And then from there that just becomes the next level where uh maybe there's uh, you know, an iris scan here so they can do the biometric uh right away, just let you get into your room.
• 45:27
  SPEAKER_00: Your room is either on your phone wallet or whatever piece of technology.
• 45:32
  SPEAKER_00: It's all it's it's all incredible.
• 45:35
  SPEAKER_00: But as you said, it still requires that deep integration on the back end to one central piece in order to make that happen.
• 45:43
  SPEAKER_00: Yeah.
• 45:44
  SPEAKER_00: Yeah.
• 45:44
  SPEAKER_00: No, well said.
• 45:46
  SPEAKER_00: So um it's been a delight talking with you, Sloan.
• 45:50
  SPEAKER_00: Um, looking broadly, are there any other big trends you want to touch upon that are things you're thinking about for AI Hospitality Group or are just things that get you excited?
• 46:04
  SPEAKER_01: Uh I mean, I could go on and on and on, but um, you know, I'm really excited about us now being kind of public and you know, welcome any owner listening in or interested uh about partnering with us.
• 46:20
  SPEAKER_01: Um I do think also uh we're gonna have key partnerships with hotel brands, which I'm I'm working on.
• 46:26
  SPEAKER_01: Um so you know, we're gonna manage all types of hotels, branded and unbranded.
• 46:31
  SPEAKER_01: Um and, you know, I think this uh, you know, robotics is going to come into play over the next couple of years, too.
• 46:38
  SPEAKER_01: So we continue to want to be tip of the spear there, right?
• 46:41
  SPEAKER_01: I think also distribution's changing.
• 46:44
  SPEAKER_01: Um, you know, you have Airbnb now moving in to the space in a big way.
• 46:49
  SPEAKER_01: Um, what does AI do to do distribution costs, customer acquisition costs in hotels?
• 46:54
  SPEAKER_01: I think is really exciting.
• 46:55
  SPEAKER_01: You would hope it's an opportunity for supply to take some margin back or some control back.
• 47:02
  SPEAKER_01: Um, you know, hotel owners have and have been great at giving up their onward distribution rights for decades to the detriment of their own PL.
• 47:12
  SPEAKER_01: So you have this transformational technology that is now going to change how consumers research travel and travel preferences and hopefully um, you know, supply is a four-wall business, physical places, right?
• 47:27
  SPEAKER_01: You so hopefully there's gonna be some control taken back.
• 47:30
  SPEAKER_01: I think it's just a really exciting time.
• 47:32
  SPEAKER_01: I mean, I anyone listening in, of course, that you should be a little apprehensive, right?
• 47:37
  SPEAKER_01: There is going to be job destruction, but there's gonna be job creation.
• 47:41
  SPEAKER_01: And I think in an AI world, humans are gonna travel more because we're we value experiences.
• 47:47
  SPEAKER_01: And so we happen to work in a business that I think AI is an accelerant of.
• 47:51
  SPEAKER_01: And I think that's really, really exciting.
• 47:53
  SPEAKER_01: So um obviously there's all other we could get into geopolitical, blah, blah, blah.
• 47:59
  SPEAKER_01: But uh, but I'm I'm a AI optimist and uh, you know, really hopeful for the future and it's exciting.
• 48:08
  SPEAKER_00: Yeah, it is incredibly exciting to see it.
• 48:11
  SPEAKER_00: And as you said, this whole idea of highly refined searches through AI travel search and how people are connecting with hotels and the role that your companies and others will play in filtering direct traffic to uh optimize channels, which again is going back to that that GOP bump that you can help hotels with.
• 48:35
  SPEAKER_00: And then what they do with that GOP bump is up to them to innovate their product in various ways.
• 48:40
  SPEAKER_01: Yep.
• 48:41
  SPEAKER_01: Yep.
• 48:41
  SPEAKER_01: Yeah.
• 48:42
  SPEAKER_01: And and we hope we are, you know, we've had a, we have a, at least in the United States, we have a profit sustainability problem where the average EBITDA margin for a hotel owner has contracted 20% over the last eight to nine years.
• 48:56
  SPEAKER_01: That can't keep happening, or we're not gonna have new hotel development, or a lot of hotel owners are gonna keep going out of business.
• 49:02
  SPEAKER_01: And so we do see ourselves as like, okay, let's reimagine how we operate for longer-term success, sustainability.
• 49:11
  SPEAKER_01: And I'm I'm really excited to be partnering with all kinds of owners and uh, you know, helping make the industry more sustainable and better and prepare for what is going to be the most disruptive technology that we've ever witnessed, AI.
• 49:27
  SPEAKER_00: Yeah.
• 49:28
  SPEAKER_00: And um, you know, to drive it right back to the beginning, talking about your business model.
• 49:33
  SPEAKER_00: One of the biggest issues we've had with various uh people who work for hotel owners, TPOs, designers, et cetera, is they aren't inherently aligned with EBITDA growth.
• 49:47
  SPEAKER_00: That's right.
• 49:48
  SPEAKER_00: And your business model is profit sharing.
• 49:52
  SPEAKER_00: You're wholly aligned with that EBITA growth.
• 49:55
  SPEAKER_01: 80% of our income is profit share only.
• 49:58
  SPEAKER_01: There is not another operating business, software business in the entire ecosystem that has that kind of aligned risk with the owner.
• 50:08
  SPEAKER_01: Um, so we're an outcomes-based um, and that that model doesn't exist.
• 50:13
  SPEAKER_01: And we we, as we say, uh, you know, uh, we take all the risk up front because if we don't move profitability for you, we don't get paid.
• 50:22
  SPEAKER_01: So um, you know, that and I think that nature of that in and of itself is revolutionary because we're aligning incentives with the owner, which frankly needs to be done in other categories, but um, you know, we're doing it through technology and transformation of the business process.
• 50:41
  SPEAKER_01: And and so uh it, you know, I would I could argue all day long that that one change is transformational too, but uh we want to be aligned with our partners.
• 50:51
  SPEAKER_00: Yeah, and you're you're standing behind your product because we've talked about all the the data that needs to happen in terms of storage, uh flow through, and then real-time applications.
• 51:02
  SPEAKER_00: Uh correct me if I'm wrong, but that isn't cheap.
• 51:05
  SPEAKER_01: No, no.
• 51:06
  SPEAKER_01: And and you know, it's it's one of those things that, you know, uh, yes, of course, we're saving labor costs, yes, we're moving revenue, yes, we're saving undistributed expenses, but then there's some incremental token and technology costs, right?
• 51:20
  SPEAKER_01: Tokens are not free.
• 51:21
  SPEAKER_01: Now, we one of the things we didn't even get into is the architecture with the good thing is you break down workflows in hotels, you don't need to use frontier models for a lot of things.
• 51:30
  SPEAKER_01: So, like our sales agent is often defaulting to a very cheap model that is fractions of a cent every time we call on it.
• 51:38
  SPEAKER_01: And so you have to be able to an architecture by which it is scalable.
• 51:41
  SPEAKER_01: Now, I do think token cost over time comes down.
• 51:44
  SPEAKER_01: We'll see what energy cost does long term, and open-weighted models are gaining traction, and we'll see how that's regulated.
• 51:50
  SPEAKER_01: But um, so I am uh positive on token cost coming down as you're looking at arbitrage between physical labor and token.
• 51:59
  SPEAKER_01: But yeah, it's not.
• 52:01
  SPEAKER_01: I mean, our it might one of our big one of my biggest costs is token costs on my PL, because you know, we we have a lean mean engineering team, but they're spending thousands and thousands of dollars on tokens.
• 52:12
  SPEAKER_01: So uh yeah, there's a real cost there for sure.
• 52:16
  SPEAKER_00: Yeah, but then again, you know, the incentives are aligned.
• 52:19
  SPEAKER_00: So it's not like you're just paying those tokens forward.
• 52:21
  SPEAKER_00: It's yeah, you know, you're if you're not producing, then that's right, you're absorbing those costs and that's right.
• 52:27
  SPEAKER_00: You yeah, you you say thousands of dollars.
• 52:30
  SPEAKER_00: That as you as you attain scale, that will quickly become a six-figure line item.
• 52:35
  SPEAKER_01: Oh, seven figures and not the too distant future.
• 52:39
  SPEAKER_00: Yeah.
• 52:40
  SPEAKER_00: So yeah, I guess uh that's the uh anthropic IPO we're all looking forward to.
• 52:44
  SPEAKER_00: Yeah.
• 52:45
  SPEAKER_00: Yeah.
• 52:45
  SPEAKER_01: Yeah.
• 52:45
  SPEAKER_01: We have partnerships with Anthropic and Open AI.
• 52:48
  SPEAKER_01: Uh yeah, maybe I should.
• 52:50
  SPEAKER_01: I I am actually a shareholder in Anthropic.
• 52:53
  SPEAKER_01: I was into their uh raise, but um totally unrelated to this business.
• 52:57
  SPEAKER_01: But yeah, maybe I should have bought more.
• 53:00
  SPEAKER_01: So yeah, I do think token cost comes down because I think a lot of the open weight models, and there's going to be regulatory there, there needs to be some regulation around.
• 53:12
  SPEAKER_01: And you know, and I do think that unless energy cost explodes, the, you know, just like with any technology, it gets cheaper better over time.
• 53:20
  SPEAKER_01: And so I think that that naturally is part of the flywheel too, that even if your your token costs are using more, you know, you don't you don't need to be using frontier models, and frankly, the models get cheaper over time.
• 53:32
  SPEAKER_00: I do Yeah, but it's it's it it's as you said, like a sales RFP repl reply.
• 53:38
  SPEAKER_00: Once that's trained and it's and it's it's perfected to 90% accuracy, you can arbitrage which model you're using to save on tokens.
• 53:49
  SPEAKER_00: And that's exactly right.
• 53:51
  SPEAKER_00: You you have the human in the loop to you know manage.
• 53:54
  SPEAKER_01: OS actually self-elects which when it calls on the agent, what LLM to use, and cost is part of the intake of that.
• 54:03
  unknown: Okay.
• 54:16
  SPEAKER_01: So because if you feed it all the data, right?
• 54:20
  SPEAKER_01: Um you have all this data that is is you're powering it by, it has tons of context.
• 54:26
  SPEAKER_01: And then you're injecting the intelligence in a very specific way.
• 54:29
  SPEAKER_01: And you don't need like Fable from Claude to run it, et cetera.
• 54:32
  SPEAKER_01: You can run all these lower cost models.
• 54:34
  SPEAKER_01: So that's the beauty of it.
• 54:36
  SPEAKER_00: Yeah, because you've already structured it in a way where you're essentially just putting it through to do a repeatable task that um, you know, some team member would be trained to do that's right that they didn't like to do in the first place.
• 54:48
  SPEAKER_00: That's right.
• 54:49
  SPEAKER_01: Doing accounts payable, you do not need, you know, high AI.
• 54:54
  SPEAKER_01: Um, you know, there's a few functions that, you know, when you're doing maybe really sophisticated financial analysis or very sophisticated revenue management stuff, but absent a few, yeah.
• 55:05
  SPEAKER_01: That's that's the beauty of this, that if you architect the data, the workflow, and the processes correct, that you know, you don't you don't have to have massive token cost.
• 55:15
  SPEAKER_00: Yeah.
• 55:16
  SPEAKER_00: I mean, it's it's great to hear it coming from you because that's something that I look at every day for my own personal use.
• 55:23
  SPEAKER_00: And you know, it's it's something I think we're all looking at, going like, hey, you know, I tried out uh Opus on this thing and now I've used up my uh my token allowance uh in in in you know two turns, or you know, it told me that it was using all these two all these tool calls and I'm done, right?
• 55:42
  SPEAKER_00: Yeah, yeah.
• 55:43
  SPEAKER_00: So yeah.
• 55:44
  SPEAKER_00: So it's it's something that you're you've found a way to streamline so that way you have some direct tools into your platform, your operating system that the hoteliers don't necessarily need to think about that.
• 55:58
  SPEAKER_00: It's gonna be as perfect as possible.
• 56:00
  SPEAKER_00: And this is trained, it's ready to go.
• 56:03
  SPEAKER_00: Uh there's immediate time to value in terms of realizing all these functions.
• 56:07
  SPEAKER_01: Yeah.
• 56:07
  SPEAKER_01: Yeah.
• 56:08
  SPEAKER_01: And I think if you talk to any very sophisticated software company, too, they're building everything with optionality on the model that you use to always take into a token uh cost.
• 56:21
  SPEAKER_01: So I would argue that's not unique.
• 56:23
  SPEAKER_01: Uh, but obviously us being the employer where we're arbitraging between labor and token, we're always thinking about that, maybe more intently, because the software company is just thinking about it on their onward token cost and their software margin.
• 56:37
  SPEAKER_01: We're thinking about it as like we're the operator sharing in this profit with the owner.
• 56:42
  SPEAKER_01: And if token cost gets out of control, we make no money, you know, versus like the software provider, they'll pass it on or something else.
• 56:50
  SPEAKER_01: So it's it's it's just a different angle.
• 56:52
  SPEAKER_01: But anyways, I I digress.
• 56:55
  SPEAKER_01: I you get me going on this token cost stuff, and you just won't get me off the horse, you know?
• 57:00
  SPEAKER_00: No, but it's it's so important because it's something you have to manage on your end.
• 57:04
  SPEAKER_00: And because your incentives are aligned with the owners, uh, it's something that you're not just gonna go, oh, by the way, your software subscription on renewal is going up by 25%.
• 57:16
  SPEAKER_00: It's right, you know, no, no, we're we've figured out a process to streamline it.
• 57:21
  SPEAKER_00: And by the way, we've added these 12 new agents this month that can further enhance your operations.
• 57:28
  SPEAKER_00: That's something like that.
• 57:29
  SPEAKER_00: That's right.
• 57:30
  SPEAKER_01: And the agents have costs, that's the token cost.
• 57:32
  SPEAKER_01: And we need to make sure that the those agents are fractions of cents of actual humans.
• 57:38
  SPEAKER_01: Um, and so that's that's exactly what we're doing.
• 57:42
  SPEAKER_00: Yeah, and it's fantastic to hear.
• 57:44
  SPEAKER_00: Um Sloane, is there anything else that you want to mention of all the things we've talked about, all the future stuff?
• 57:52
  SPEAKER_00: I don't know.
• 57:52
  SPEAKER_00: Anything else you want to mention?
• 57:54
  SPEAKER_01: I don't think so.
• 57:54
  SPEAKER_01: I I think uh, you know, I'd welcome uh any feedback from your listeners.
• 57:58
  SPEAKER_01: And you can find me on LinkedIn, uh, reach out, and thanks for having me.
• 58:03
  SPEAKER_00: Awesome.
• 58:04
  SPEAKER_00: Well, it's great to have you on the show and just a delightful power hour to learn about what you're doing and how it's changing and revolution really revolutionizing hospitality.
• 58:14
  SPEAKER_01: Hey, thank you.
• 58:14
  SPEAKER_01: Enjoy your birthday.
• 58:16
  SPEAKER_00: Thanks.
• 58:17
  SPEAKER_00: Thanks, Sloan.