Hotel AI Adoption Is Being Held Hostage by the Brands

Jake Wurzak of DoveHill on why Hilton and Marriott's locked technology stacks are costing hotel owners real money - and what operators are doing about it.

 

AI adoption in hotel real estate is stalled at the operator level because Hilton, Marriott, and Hyatt control the core technology stack at every affiliated property. Jake Wurzak, founder and CEO of DoveHill Capital, which oversees 20 hotel properties across the East Coast, describes a hospitality industry where revenue management is still adjusted manually by staff making daily rate changes, guest history sits buried in systems rather than surfaced at the front desk, and back-of-house operations such as housekeeping schedules and purchasing run on processes that have not materially changed in a decade. The constraint is structural: branded hotels cannot swap in third-party AI platforms for core functions. Independent operators are already building around it.

Key Takeaways

• Brand lock-in is the primary AI bottleneck in branded hospitality. Hilton, Marriott, and Hyatt control the technology stack at affiliated hotels. Operators cannot integrate third-party AI platforms for revenue management or guest check-in without brand approval - which moves slowly.
• Revenue management is the highest-value AI target in hotel operations. DoveHill employs three staff members whose principal task is manually adjusting room rates daily. Wurzak is exploring AI agents that could perform this function simultaneously across six or seven properties, with a human reviewing recommendations rather than entering data.
• Guest personalization at luxury hotels remains shockingly manual. Even top-tier operators such as Four Seasons rely on staff to manually retrieve guest preference data at check-in. AI's near-term impact here is surfacing existing data automatically rather than collecting new data.
• The operator workaround is deploying agentic AI on top of existing systems. Rather than waiting for brands to offer their own AI tools, DoveHill is testing Claude as an agent that operates within legacy systems - replacing human data-entry tasks without requiring new system integrations.
• Back-of-house automation carries more immediate ROI than guest-facing AI. Housekeeping scheduling, purchasing, and data-linking between accounting and business intelligence platforms are all manual processes that Wurzak identifies as ready for automation today.
• AI-powered rendering is compressing renovation planning timelines. Wurzak used Gemini to generate architectural renderings from a phone photo and a text description - a process that previously cost $3,000 and took three weeks. The output was usable on the same day.
• Brands will eventually be forced to adopt AI faster than any previous technology. Wurzak's view is that cost pressure and revenue opportunity will compel Hilton and Marriott to move - but the transition has been slowed by the complexity of migrating legacy systems, a constraint that AI itself may now be removing.

Adam Gower has spent more than 30 years and $1.5 billion in commercial real estate transactions, including operating at the executive level across Asia Pacific development. His clients collectively manage over $45 billion in AUM. The friction Wurzak describes - sophisticated sponsors constrained by legacy systems and brand relationships - is a pattern Gower has watched play out across asset classes, and the hospitality sector's particular version of it is instructive for any operator evaluating where AI can actually be deployed versus where it remains aspirational.

 

The Brand Stranglehold on Hotel Technology

The hospitality industry's technology problem is not a shortage of solutions. It is a distribution problem created by three companies.

Hilton, Marriott, and Hyatt collectively control the core technology stack at branded hotels. Revenue management systems, property management systems, and guest check-in infrastructure are all supplied or mandated by the flag. An operator who wants to run an AI-powered revenue optimization platform at their Hampton Inn cannot simply install one. The brand has to offer it, approve it, or integrate it - a process that moves at the pace of a large institutional bureaucracy, not a technology deployment cycle.

Wurzak is direct about what this means in practice: "The real reason why it's very challenging is the big brands - Hilton, Marriott and Hyatt - control all of the technology in hospitality. So if you want to affiliate your hotel with one of those companies, you have to use their technology stack. And their hotel people - they're not technology people, so they're not technology first. No matter how hard they'll tell you that."

The result is that DoveHill, which manages 11 properties directly, employs three revenue management staff whose primary task is entering rate changes into computers manually - every day, for every property. This is not a staffing decision. It is a system constraint. The platforms those employees use do not offer dynamic AI-driven pricing. They offer rate buckets: $179, not $175, because the system does not support finer granularity.

Independent operators have more freedom but face a different problem. The AI and hospitality technology vendors targeting them find it difficult to build a viable business around a fragmented, undercapitalized ownership base. The result, as Wurzak describes it, is that useful technology "doesn't get pushed as much as it does in other industries into hotels, because there's this logjam with the brands." 

 

Where AI Can Actually Be Deployed in Hotel Operations Today

DoveHill's current approach separates the question of what AI could do in an ideal system from what it can do within the constraints that exist. The answer to the second question is more useful.

On the investment and asset management side, where DoveHill is not subject to brand restrictions, Wurzak has been deploying AI tools directly. He describes using Claude to review legal documents during a loan closing - reducing review time from 30 minutes to five minutes per revision cycle. He used an AI image tool to generate renovation renderings from a phone photo and a text description, cutting a process that previously cost $3,000 and took three weeks to a same-day task. He is using AI to work within Excel models, describing the output he wants and then having the tool check its own work.

The operational applications he considers most near-term are back-of-house: housekeeping scheduling, purchasing optimization, and data integration between systems that do not talk to each other cleanly. He describes a specific case where a senior employee spent six hours manually linking data between an accounting system and a business intelligence platform because the two vendors are competitors and their integration is poor. "When you're spending six hours on a project like that, your entire brain is fried. So you have no energy left for higher level strategic thinking. And this is a senior, senior senior person of mine." That data-linking task, in his assessment, is directly automatable today.

Housekeeping represents the largest single payroll line item at most hotels. Wurzak has reviewed a robotic bathroom-cleaning demonstration and describes the operational case plainly: lower labor cost, better sanitary outcomes, reduced worker injury exposure, and lower insurance costs. The robot is AI-powered; the category label matters less than the cost structure it enables. 

 

The Agentic Workaround: Bypassing Brand Restrictions Without Replacing Systems

DoveHill's most operationally significant AI experiment is not a new platform. It is an agent running on top of existing platforms.

The concept is simple and the constraint it solves is real. A branded hotel operator cannot replace the brand's revenue management system. But the operator can put an AI agent on a computer that operates within that system the same way a human employee does - navigating the interface, reading data, and making changes. The agent does not require API access or vendor cooperation. It replaces the human interaction with the legacy system rather than the system itself.

Wurzak describes the current state of implementation: his team switched from ChatGPT to Claude roughly ten days before this recording, after reading about Claude's capabilities in agentic and computer-use contexts. The investment team is now being rolled out on Claude across the organization. His deployment philosophy is deliberately unstructured: "I don't want to put like guidelines like this is how you have to use it. I want you to figure out ways that you can use it to make your life easier."

The revenue management application he is working toward would have a single agent monitoring multiple properties simultaneously, surfacing recommended rate adjustments based on occupancy data, event calendars, competitive sets, and demand signals - and executing those adjustments across six or seven properties with a single human approval. Currently, one employee can work on one property at a time. The same headcount would cover the entire portfolio. 

 

What Hilton and Marriott Are Likely to Do - and When

Wurzak does not predict the brands will be displaced. He predicts they will be forced to accelerate.

His reasoning is cost-based. The historical reluctance to adopt new technology has not been purely organizational inertia. Legacy system migration is genuinely expensive and operationally disruptive. But AI changes the migration calculation. Where previously transitioning a large property management system required significant manual labor to move data and rewrite integrations, AI can now perform much of that work directly. Wurzak cites Oracle's shift to using AI to write software that previously required human developers - noting that the same dynamic applies to hotel system transitions.

The brands also face a competitive exposure that did not previously exist at scale. Independent lifestyle hotels and boutique properties - the segment Wurzak is moving toward with his Auberge acquisitions - are not constrained by brand technology mandates. A well-operated independent can deploy AI-powered check-in, automated guest communications, dynamic pricing, and personalization tools that a Hilton-flagged property across the street cannot match. As the guest experience gap widens, the brand value proposition erodes.

Wurzak's observation that Auberge - a smaller, more nimble luxury brand - gives him considerably more technology latitude than Hilton or Marriott reflects a more general principle: the degree of brand restriction correlates inversely with the brand's scale. Operators who want to move faster on AI have an incentive to migrate toward flags that can move with them. 

 

Frequently Asked Questions

Can branded hotel operators use AI tools that aren't provided by the brand?

It depends on the function. For core operations - revenue management, property management, guest check-in - branded hotel operators are generally required to use brand-mandated systems. Third-party AI platforms cannot typically be substituted for these functions without explicit brand approval. For peripheral functions such as purchasing, accounting, or asset-management-level analytics, operators have considerably more flexibility. Agentic AI tools that operate within existing systems rather than replacing them represent a middle path that does not require brand approval or system integration - though this approach is still early in deployment at most operator organizations.

What is the biggest immediate ROI opportunity for AI in hotel real estate?

Based on Wurzak's assessment, revenue management represents the highest-value target because it is directly tied to room rate optimization across a portfolio - and it is currently performed manually at most branded hotels. Automating data aggregation and rate-change execution across multiple properties simultaneously, with human review of recommendations rather than human data entry, could produce measurable revenue improvement with the same headcount. Back-of-house automation - housekeeping scheduling, purchasing, data integration - offers cost reduction with fewer structural barriers to deployment.

How should hotel operators approach AI adoption without a dedicated technology team?

Wurzak's approach is to give his team access to tools and explicitly not prescribe how to use them. The deployment goal is to have individuals identify their own highest-friction tasks and experiment with AI solutions for those specific problems, rather than implementing a top-down AI strategy. He distinguishes between AI as a conversational tool - the ChatGPT use case he describes as having "maxed out" - and AI as an agentic system capable of operating across software and documents. Operators building fluency with the former and progressively moving toward the latter are following the practical adoption curve he describes.

Will AI eventually replace the major hotel brand platforms?

Displacement of Hilton or Marriott as flag brands is not a near-term outcome. The more plausible scenario is that AI accelerates the migration of branded operators toward flags that impose fewer technology restrictions - smaller luxury brands, independent collections, and soft brands with less prescriptive technology mandates. The brands themselves will face pressure to adopt AI tools that reduce operating costs and improve revenue performance, particularly as independent competitors deploy capabilities that branded properties cannot match. The historical pace of technology adoption by major flags - slow by design - is likely to increase, driven by the economics rather than the aspiration.

Continue the Conversation

For CRE sponsors building an operational view of where AI applies across asset types and investment functions, the full episode with Jake Wurzak is in the links above. Related GowerCrowd content on AI deployment for real estate operators.

Sponsors who want to move beyond the individual tool stage and build systematic AI capability across their investment and operational functions can review the AI in Real Estate Accelerator Program here.

 

Adam Gower, Ph.D. is the founder of GowerCrowd and one of the most experienced practitioners in commercial real estate capital formation. With more than 30 years and $1.5 billion in transactional experience - including serving as President of a Universal Studios development division overseeing $400 million in projects across Asia Pacific - Adam now helps CRE sponsors build AI-powered systems for investor acquisition, deal management, and capital deployment. His clients collectively manage over $45 billion in assets under management.