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

The third-party hotel management business has not fundamentally changed in 21 years. The staffing model is the same. The fee structure is the same. The organizational chart is the same. And according to Sloan Dean, who spent eight years as CEO and president of Remington Hospitality before founding AI Hospitality Group, that is precisely the problem. Not because the people running those companies are not talented, but because the model itself has become commoditized to the point where no single operator can credibly claim meaningful differentiation. "If you talk to any institutional capital," Dean says, "they will tell you that entire space is commoditized."

Dean's answer is not to build a better management company. It is to rebuild the category from scratch, using AI as the foundation rather than the add-on. AI Hospitality Group, which he describes as the first AI-native service provider in hotels, is the result: a venture-backed, engineer-heavy startup that operates hotels on a profit-share model, builds its own data infrastructure, and is currently developing a roadmap to over 75 AI agents coordinated across every function of the hotel operation. He sat down with Adam Mogelonsky to explain how it works, why the incentive structure matters as much as the technology, and what hotel operations look like in a world where full physical departments are fully automated.

What is AI Hospitality Group?

AI Hospitality Group is what the venture capital firm Emergence Capital calls an AINS company - an AI-native service provider. The term, coined in Silicon Valley, applies to companies that meet three criteria: they have always been AI native rather than retrofitting AI onto an existing model; everything they do is oriented toward doing something faster, better, or cheaper than incumbents; and they are outcomes-based rather than fixed-fee providers. "You're being paid on outcomes," Dean says. "We're a profit share model. That's how we get paid."

The business model is the most structurally significant departure from the conventional management company. AI Hospitality Group earns approximately 80% of its income through profit share only, taking a percentage of the GOP improvement it generates for hotel owners. If the company does not move profitability, it does not get paid. "We take all the risk up front," Dean says. "And that model doesn't exist anywhere else in the ecosystem." In an industry where the average EBITDA margin for a US hotel owner has contracted approximately 20% over the last eight to nine years, Dean sees the aligned incentive model as both commercially necessary and strategically differentiating. "Hotel owners have been great at giving up their onward distribution rights for decades to the detriment of their own P&L. So we align with the owner."

The closest parallel Dean reaches for is Palantir - the data analytics company that embeds forward-deployed engineers directly inside client organizations, builds custom solutions, and completely redoes workflows rather than selling off-the-shelf software. "That's probably the closest parallel to what we're doing, but we're doing it for hotels."

Why the existing model is broken

Dean's critique of the existing third-party management landscape is direct and grounded in his own experience running one of its largest players. "Is Aimbridge any different than Highgate, any different than Pyramid, any different than Crestline?" he asks. "Sure, there are some that are slightly better than others, but if you talk to any institutional capital, they will tell you that entire space is commoditized."

The symptoms of that commoditization are visible in the market data: a race to the bottom on management fees, high contract churn (the average third-party operator loses between five and twelve percent of its portfolio annually), and extreme market fragmentation, with no single operator holding even 1% of the total addressable market even at the scale of a thousand hotels. "That typically happens in commodity businesses," Dean says.

The structural reason, in his analysis, is that third-party operators are functionally staffing companies. They provide human labor at a margin. They have no particular incentive or capability to automate the work those humans do, because doing so would reduce the revenue base on which their fees are calculated. "If you are just an employer, which is what all third-party operators are today, they're gonna have to outsource that innovation and then that value goes to a software company."

What makes this moment different from previous periods of efficiency pressure is that the technology now exists to do something more fundamental than increase productivity within existing job categories. The question Dean is asking is not how to make a revenue manager 20% faster. It is whether the revenue management function, as currently structured, needs to exist at all.

The data architecture: AICOS and the canonical data lake

The technical foundation of everything AI Hospitality Group does is what Dean calls AICOS - an operating system and canonical data lake that ingests data from approximately ten different hotel systems: the financial ERP, the PMS, the CRS, the RMS, the applicant tracking system, the HR platform, and others. All of it is normalized, contextualized, and structured so that AI can consume it reliably to produce deterministic, repeatable outputs.

"One of the reasons you've used an LLM and you feed it a similar prompt and get a different output is because you're not giving it the right contextual data," Dean says. "We have no lack of data in hotels. We just don't have it normalized and contextualized for AI to consume it. And we've built that."

The key design principle is system-of-record agnosticism. For certain functions, like financial ERP, HR and payroll, AI Hospitality Group runs its own strategic partnerships and chosen systems across every hotel it manages, because those are functions the operator controls. But for systems like the PMS, CRS, and RMS, where branded hotels must run whatever the franchise agreement dictates and independent hotels have often already made significant investments, AICOS is built to ingest from any system. "Even if it doesn't have an API or file transfer protocol, you can access it and the agent can act like a human on a browser. That's not a preferred method, but we've built it."

The modularity matters for a specific competitive reason: as AI improves and software companies increasingly commoditize their own functions, hotel operators who have all their data normalized in a system-of-record-agnostic layer can swap out underperforming vendors without losing their data infrastructure. "If you're at the mercy of software companies, they're going to go deeper into P&L value leakage and siphon off value away from the hotel owner," Dean says. "We've built the orchestration layer with that in mind."

The economics: 500 basis points and adding up dimes to dollars

The headline number AI Hospitality Group uses is 500 basis points of GOP margin improvement, moving the average full-service hotel in the United States from a 32 to 33% GOP margin to approximately 38%. Dean is careful about how he characterizes this. "It's adding up dimes to dollars," he says. There are two significant individual line items: an AI sales agent that transforms group sales conversion at full-service hotels where group revenue represents a third of total revenue, which can move the margin by 100 basis points or more on its own; and an orchestration layer that enables middle management reduction, potentially another 100 basis points. Everything else is incremental - ten basis points here, twenty there - accumulated across dozens of automated processes.

The staffing model example is the most concrete illustration. A full-service hotel generating approximately $30 million in annual revenue typically runs 20 to 22 salaried managers. AI Hospitality Group believes it can operate the same hotel, at increased revenue and increased guest satisfaction, with 12 to 14. At an average fully loaded compensation of $100,000 to $150,000 per position, the saving approaches $1 million annually in payroll alone. "The twelve or thirteen you do have are empowered to do even more," Dean says. "And some of those savings you give back in higher wages. The work becomes more meaningful and you can pay them more."

The time to stabilized expanded GOP from engagement is approximately six months - not because the technology takes that long to deploy, but because the iteration cycle of installation, training, and refinement across multiple agents takes time to compound. "It's not a matter of you just turn on a switch," Dean says. "Some things you can implement day one, but to get to stabilized expanded GOP takes about six months."

The three levers: management, operational efficiency, and commercial strategy

Dean reduces the AI impact on hotel operations to three categories. The first is the reimagination of the management layer: reducing middle management headcount while increasing the capability and compensation of those who remain, and "10X-ing" the GM so they can oversee a broader operation with AI-enhanced visibility. The second is operational efficiency across every department: housekeeping, laundry, kitchen, accounting, procurement, engineering - AI can make any department five to fifteen percent more efficient pre-robotics, with the caveat that unionized properties under collective bargaining agreements face structural constraints on how far this can go. The third is commercial strategy: using AI to expand not just rooms revenue but total revenue across all departments, and improving the efficiency of distribution to reduce customer acquisition costs and recapture margin from OTAs.

None of these is simple in execution. "A middle manager (an AGM)# may do thirty or forty things in a day," Dean says. "If you want part of that role to be reimagined, you have to go through and automate all those individual things, because hotels have a lot of blended work." The current AI Hospitality Group product roadmap targets over 75 agents, all coordinated within the AICOS layer, each handling a specific task or workflow that previously required human attention.

Humans in the loop: the flywheel

The most important phrase in the conversation is not one Dean uses for effect. It is one his COO Eve Moore wears on a T-shirt as an inside joke: "Human in the loop officer." The phrase captures something real about how AI Hospitality Group operates. The humans who remain in the organization are not passive beneficiaries of AI automation. They are active participants in the iteration process: identifying when an agent produces the wrong output, feeding that back to the engineering team, and improving the system. "Your associates are helping improve the technology and iterating," Dean says. "And by the way, they're benefiting by a higher wage by also being educated in AI."

This flywheel - engineer builds agent, operator uses agent, operator finds flaw, engineer fixes flaw, agent improves, operator benefits - is what Dean believes cannot be replicated by either a pure software company or a pure staffing company. A software company sells the product and moves on; a staffing company has no engineering capability to build the agents in the first place. "You can't spin that flywheel faster if you're just the employer and you're just the software provider," he says.

The implication for the humans who remain is also real. Dean is explicit that AI Hospitality Group will operate hotels with fewer people than anyone else. He makes that known to prospective partners and he does not apologize for it. But he is equally explicit that the jobs those people have should become more meaningful, better paid, and more oriented toward the guest-facing work that brought them into hospitality in the first place. "Nobody got into a hotel and said: "I want to do accounts payable." If we can automate all that and 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."

Robotics: Star Wars, not iRobot

The conversation inevitably arrives at robotics, and Dean's framing is worth quoting directly. "Think Star Wars, not iRobot." The popular imagination of hotel robotics is a humanoid robot - "Betty" - walking into a room and making a bed. That is not where the technology is heading, at least not in any commercially viable near-term timeframe. What is actually coming, in Dean's analysis, is a fleet of purpose-built single-function robots coordinated by an AI brain: one robot for hard surface floor cleaning, one for trash and linen management, one for cart handling, all working in parallel to clean a room faster than a single human housekeeper could, and eliminating the highest-injury tasks (soiled linen handling, heavy lifting) from the roles of the humans who remain.

"When you have bots doing multiple things, you can clean the room a lot faster," Dean says. "It's not one humanoid. So that's where everybody gets it wrong." The hardware costs are not yet at the point where the cash-on-cash payback for full humanoid robots makes sense. But purpose-built robots for specific physical tasks are already approaching commercial viability, and AI Hospitality Group is in active conversation with hardware partners (none named yet, pending signed agreements) to be a first-mover operator at scale.

The downstream implication Dean flags is one that most of the industry has not yet connected: if rooms can be cleaned by robots on any schedule, at any hour, the entire architecture of the hotel day collapses. Check-in at 3pm and checkout at 11am are not hotel traditions. They are scheduling constraints imposed by human labor. "If a robot can work all the time," Dean says, "can you turn the room twice in New York where you're running a hundred percent occupancy? Maybe someone only needs the room. In New York, you have hotels that could run a hundred and thirty percent occupancy if you could turn the room." The PMS, the revenue management logic, the rate structure... all of it would need to be reconsidered in a world where room availability is continuous rather than batched.

Distribution, AI travel search, and taking margin back

The final trend Dean flags is the shift in how guests discover and book hotels, and the opportunity it creates for hotel owners to recapture margin from OTAs. AI-powered travel search is changing the research and booking journey in ways that may, for the first time in decades, give independent hotels and operators a structural advantage over aggregators. "Supply is a four-wall business - physical places," Dean says. "Hopefully there's going to be some control taken back."

The mechanism is familiar from other contexts: AI search engines surface specific, high-quality answers to specific queries, and a hotel with a clearly articulated identity and a well-structured data presence has a better chance of appearing as a direct answer than an OTA listing built on aggregated volume. The same data architecture that AI Hospitality Group builds for operational purposes also positions the hotels it manages more effectively in a distribution landscape that is shifting away from search-engine-indexed aggregators and toward AI-generated recommendations. "You would hope it's an opportunity for supply to take some margin back," Dean says.

What comes next

AI Hospitality Group is currently managing a small number of properties and is actively seeking owner partnerships across branded and independent full-service hotels in the United States. The underwriting process is deliberately intensive: Dean describes it as requiring access to staffing models, technology stacks, and non-traditional data sets before the company will produce a five-year pro forma. "It's more intrusive than what most owners are used to," he says. "But we'll tell you exactly what we can do and stair-step the margin improvement."

The venture-backed structure, with its tolerance for losses during the build phase and its capitalization for engineering-heavy investment, is what makes the model possible. "You can't really do this if you're already operating at scale," Dean says. "It's incredibly difficult to reimagine something that is big. You have to go slow, solve one problem at a time, and look like a technology company reinventing a process before you try to scale up."

The company expects to have strategic partnerships with major hotel brands managing both branded and independent properties in the near term. The robotics partnerships are in development. The agent roadmap continues to grow. And the fundamental bet that the hotel management category is overdue for the same disruption that transformed taxis, retail, and media remains the animating conviction. "I think AI is an accelerant of this industry," Dean says. "In an AI world, humans are going to travel more because we value experiences. We happen to work in a business that AI makes more attractive, not less. And I think that's really, really exciting."