In this blog post, I’m sharing what stood out. You’ll learn my five key takeaways, the brand experiments worth watching, and the conversations that signal where our industry is really heading.
Ready, steady, data
The first thing that jumped out to me was how much effort brands have been putting into data readiness—shifting to the cloud, centralizing signals, and building more actionable data layers.
But one thing is clear: search data alone isn’t enough. As one speaker said, “If AI is going to surface true personalization, we need richer, more exhaustive descriptions of every product, amenity and experience.”
Think about it. If your search index only knows “Hilton king room,” AI can’t predict whether that room is ideal for a family celebrating a birthday or a solo business traveler in town for a meeting. It needs more context.
Key takeaway: Data isn’t just “there” yet. It needs structure, taxonomy, and a clear path into your AI models before you see returns on those big data investments.
A conversation with your AI travel agent
Next up is the shift from static search boxes to conversational and voice-driven experiences.
Several sessions hammered home that we’re moving beyond a “type keywords, then filter” approach and toward agentic AI. Soon, AI agents will act as trusted advisors and even execute tasks on the traveler’s behalf.
We learned that people check their phones 144 times a day, and they expect faster, more intuitive experiences. Travelers don’t want to scroll through dozens of results. They want a back-and-forth, human-like conversation.
1. Hilton
Hilton has gone all-in on micro-listening. Instead of waiting for a post-stay survey, they’re using in-stay feedback triggers.
When an issue occurs—such as the air conditioning not working—they initiate a workflow immediately (alerting engineering, offering a room move, etc.) before the guest checks out.
For loyalty, Hilton’s AI can forecast room availability and decide whether an upgrade is worth the gesture. A family trip has different upgrade needs than a business trip, and AI helps them guess which is which.
Bottom line: Hilton is turning raw data into real-time service recovery and loyalty moments.
2. Marriott
Marriott’s first AI-powered complimentary upgrade engine pilot didn’t perform as expected. But instead of scrapping it, they iterated—adjusting their data inputs, retraining the model, and relaunching.
The lesson? Small failures are not final. The real win is in continuous improvement. Marriott’s story is a blueprint for brands that think “pilot to production” should be a straight line. It never is.
3. Amex (Corporate Travel)
Amex is trialing agentic AI to power travel agents, giving them policy-aware chatbots and natural-language support tools.
Agents aren’t being replaced—they’re being turned into “superhuman” advisors who can quickly interpret complex corporate travel policies and book within strict guidelines.
One comment stuck out: “During testing, we realized our internal travel agents weren’t using a lot of the built-in website features. So, we built our conversational AI on top of existing servers instead of ripping everything out.”
As a result, agents feel empowered, response times have improved, and the data loop is more seamless—because the AI model already “knows” the policy, the traveler’s preferences, and the client’s cost constraints.
Key takeaway: Designing for conversation means having richer data at every step—ensuring your AI knows more than just price and availability; it knows the traveler’s intent, context, and even emotional cues.
Empowerment, not replacement
One thing all the brands and presenters were aligned on—regardless of where they sit in the industry—was the importance of shifting the conversation around AI and jobs.
If there was one rallying cry onstage, it was to stop scaring people about AI stealing jobs and start empowering them. A few speakers said it bluntly:
“Creative will be automated, but creativity never will be.”
“Using AI code doesn’t mean developers disappear. It means they become exponentially more productive.”
Key takeaway: Enablement must come first. You can’t jump to orchestration if your teams don’t even know how to experiment.
Using MCPs to speak a common language
No summit recap is complete without acknowledging the industry’s Achilles’ heel: fragmentation. One hotel executive noted they juggle over 19 systems just to run daily operations—not even counting the customer-facing front end.
So how do you connect AI into that web of legacy tech?
Enter Model Context Protocols (MCPs). The idea is to build a standard framework so AI models can plug in, request data, and generate insights—without relying on bespoke APIs.
It’s still early days, but the technology promises two things:
- Faster integration time: Skip months of point-to-point API work by delivering a protocol layer that any AI model can consume.
- Better interoperability: If your PMS, CRS, revenue management system, and loyalty platform all speak MCP, they can share data with an LLM in a much cleaner way.
Key takeaway: Fragmentation isn’t going away overnight. But MCPs (and a robust data layer) can at least give AI a clear path through the noise.
You don’t need perfection. Just start.
The phrase I found myself writing down most was: “You don’t need perfection. Just start.”
No brand has it all figured out. But the ones making the most progress:
- Run small pilots (Hilton’s in-stay recovery, Amex’s policy bot)
- Measure and iterate (Marriott’s upgrade engine)
- Invest in upfront data work so you’re not rebuilding from scratch when you scale
One speaker predicted that by 2026, we won’t be asking “What is AI?” We’ll be asking “How are you using AI?”
We’ll see a turf of B2B successes and a few B2C cautionary tales—where brands jumped in too fast without laying the groundwork.
Key takeaway: You can’t make progress with AI if you don’t start.
Signing off from the summit
This summit made one thing clear: the future of travel won’t be built overnight. But it’s being built right now—by the brands willing to test, listen, and evolve.
The recurring theme is that AI in travel is messy, evolving, and incredibly promising. But only brands that do the hard work on data, experiment without fear, and treat their teams as the secret sauce will see results.
Thanks to Skift and all the speakers for a day of unfiltered insights. If you made it to the event, drop me a line and let me know what moments struck you.
If you weren’t there, I hope this gives you a sense of where travel technology is headed. For more information, browse our Global Travel & Hospitality Outlook or book a meeting to see how Valtech can help you achieve your travel goals.
