For all the hype around generative AI, adoption in travel & hospitality remains incremental and often painfully patchy. You may have seen the recent MIT study, which reports that 95% of generative AI pilots show no measurable ROI. McKinsey research indicates that while 78% of companies use GenAI, more than 80% have seen no material contribution to earnings. In our own work with hoteliers, OTAs, and hospitality technology providers, we’ve seen the same pattern – enthusiasm outpaces execution, and real progress is slowed by legacy systems, limited expertise, and uncertainty around where GenAI actually fits. Still, early wins in areas such as customer service automation, from chatbots to voice agents, demonstrate that GenAI can already deliver measurable value when applied to the right problems.
The observations below are drawn from operator conversations, survey data, and discussions held at the SKIFT Global Forum 2025. See our take on where GenAI is working today, where structural tensions lie, and what might come next.
Where GenAI Works Today
Today’s clearest wins for genAI in travel & hospitality have clustered around customer service automation areas, where reliable context already exists (e.g., PNRs, reservation data, FAQs), integrations are often tractable, and ROI is directly measurable through time savings and guest satisfaction. In our work, more than 50% of operators point towards voice agents, chatbots, or post-booking support as the first applications to deliver measurable value.
The logic here far predates generative AI. For decades, operators have leaned on IVR menus, offshore call centres, and scripted bots to triage common requests. Guest inquiries are often repetitive, time-sensitive, and high volume (“When’s check-out?”, “What amenities are in the room?”). GenAI hasn’t changed that equation, but it has made interactions feel less robotic and expanded the long tail of guest questions that operators can reliably address. Booking.com now resolves 60-70% of queries automatically. Hopper tells us, “Last year we processed 600,000 customer support calls, end-to-end with AI”, powered by HTS Assistant, its newly announced agentic AI travel agent. Canary Technologies reports that ~40% of hotel calls go unanswered, highlighting an immediate opportunity for properties able to leverage GenAI to capture those missed inquiries. Few other GenAI use cases combine clear efficiency gains, which naturally lends to why customer service has become the first proving ground.
What this also means is that the benefits from GenAI so far have been narrow, captured in cost savings, efficiency gains, and smoother service at the margins. An open question is whether GenAI can shift from back-office support to amplifying topline growth. By contrast, “traditional” AI and ML have long delivered direct, revenue-generating value in the industry in areas such as dynamic pricing and yield management. GenAI is still finding its footing, although OTAs, as we’ll discuss below, are beginning to push forward with enriched recommendations and personalization powered by GenAI.
Your Guests Don’t Want AI
As Claire Gibbons, COO at Cruisebound, puts it, “travellers don’t want AI, they want their problems solved”. They want transparent pricing, robust search, and the confidence that they’re getting the best deal. Some operators have experimented with explicit “AI entry points” in the booking flow (e.g., a button labelled “Chat with our AI assistant”), but uptake has been low unless guests are forced into it. In other words, nobody is clamouring for a standalone GenAI trip planner.
There’s also plenty of “bot scar tissue”. Travellers have lived through a decade of clumsy chatbots and half-integrated systems. Canary Technologies is one company challenging this reputation, offering an AI-powered guest management platform that covers everything from mobile check-in to guest messaging. The company’s platform tightly integrates with all major management systems. The result is simple – when a guest sends the front desk a message requesting an extra towel, the towel actually arrives.
However, amidst all the buzz around “agentic travel”, hospitality remains a human industry where the product remains “care”. In hospitality operations, you can’t just “take someone off the ground” with AI. One of our favourite recent takes on the topic comes from Mark Weinstein, Hilton’s Chief Marketing Officer: “If we can automate the parts that need to be automated to take the friction out of travel, that frees up our team members to do what they do best, which is deliver human hospitality”.
Tailored or Off the Rack?
Most operators today are testing GenAI through general-purpose tools like Microsoft Copilot, Google Gemini, or OpenAI. Our engagement with operators found that only about 25% were experimenting with hospitality-specific platforms. That raises a real question – what space is left for new, verticalized players, if horizontal incumbents can deliver “good enough”?
One operator we spoke to leans on horizontal providers for exactly that reason. They bring speed, scalability, observability, and QA out of the box, while operators supply the travel context through their knowledge bases or prompt refinement. Others make the case for vertical tools, which promise tighter integration and faster end-to-end execution. In our discussions, operators mainly pointed to ML-driven offerings like FLYR, an AI “commercial operating system” for travel and hospitality operators that unifies offer creation, revenue optimization, and order management; Inovia portfolio company Diamo, an AI-driven revenue management system platform that automates room pricing, digital marketing revenue insights, and guest bookings; and verticalized PMS solutions like Guesty (Inovia portfolio company), which are increasingly embedding AI across their solution sets. Guesty offers Reply AI, which provides AI-generated responses for over 98% of incoming messages, and PriceOptimizer, a tool that leverages AI to help property owners set competitive rates across channels like Airbnb and Booking.com.
However, the reality is that most hospitality stacks are still “patches and bridges”, so layering in GenAI, vertical or not, can add cost and fragility faster than it adds value. 50% of operators we engaged pointed to legacy systems as their biggest blockers.
For operators, we don’t have a universal playbook of what you should do – build or buy, tailored or off-the-rack. The better question is, which parts of your customer experience need to feel uniquely yours? If GenAI and AI are central to how you differentiate your guest experience, you’ll need to build deeper and own more of the stack. If it’s not, then there may be value in keeping AI as a rented layer in the background.
For startups, you need to know which camp you’re selling into. Most operators don’t need another conduit for ChatGPT (they’re already using it!). The bar is simple: prove you can deliver outcomes that horizontals can’t.
The Booking Agent Named ChatGPT
The long-running tension between OTAs and suppliers isn’t new. OTAs optimize for acquisition and conversion, while hotels fight to drive direct bookings, repeat bookings, and avoid commissions that often run 10%+. What GenAI introduces is a potential reset in how the booking journey unfolds. ChatGPT is already emerging as the new “search engine” for travel, and is increasingly inching towards becoming the channel where bookings are actually completed.
For OTAs, that’s naturally an existential risk – will they be pushed into the background as little more than inventory pipes? This dynamic doesn’t just endanger OTAs, but will impact entire categories of travel content creation businesses, blogs, affiliate marketing, and discovery platforms.
Against this backdrop, travel & hospitality operators are already gaming out the scenarios. Some operators expect major hotel brands will try to plug directly into LLM platforms to claw back distribution. However, today’s models still tend to keep users “on campus” (i.e., within their own ecosystems), limiting the immediate risk of OTA disintermediation, at least for now. Meanwhile, OTAs are racing to make the “GenAI experience” their own, leaning on hyper-personalized guest experiences. Priceline is investing in LLM-powered discovery that combines first- and third-party data, including guests’ preferences, to outperform standard filtering. Super.com has systematically adopted AI across its operations, with the Company reporting that “no two users will see the same app thanks to AI personalization”. Hopper offers a B2B pricing engine that uses a customer’s full itinerary context to predict the likelihood of add-on purchases, helping operators maximize ancillary revenue.
In Conclusion
The story of GenAI in travel and hospitality so far is one of narrow but tangible progress. Automating call volumes, resolving name corrections and catching a missed booking are real wins, but they live in the background. They save staff time and protect revenue, but they don’t yet change why a guest books with you, or what makes them come back. Contrast that with sectors where GenAI is already central, such as software engineering, where nearly two-thirds of developers rely on AI coding assistants like Windsurf and Cursor in their core workflows.
Most of the industry is still in testing mode, feeling out where GenAI fits for operators; that usually means starting with general-purpose tools like OpenAI or Gemini, while only a minority are experimenting with vertical platforms. The trade-off is straightforward; horizontals move quickly and are easy to adopt, but they rarely close the loop across PMS, CRS, or staffing systems. Vertical tools promise tighter integration, but in practice often add fragility and cost to stacks already held together by “patches and bridges”.
At the same time, a bigger question looms upstream in distribution. Whoever owns the “first hello” with a traveller owns the funnel. For decades, that entry point was search engines, OTAs, or brand sites. GenAI introduces a new contender – assistants that can hold a conversation, remember context, and narrow choices to a single recommendation.
Through all of this, the essence of the industry doesn’t change. Hospitality’s core product is still care. Guests remember whether things worked, whether they felt looked after, and whether the experience was worth repeating. AI’s role is not to replace that, but to ease the friction in delivering that – a human, memorable guest experience.