Where Value Lives When Problems Die: The New Rules of Competition in Travel

After my first piece, the question I kept getting on my Substack was: "Okay, so what DO we compete on?" Fair question.

And I thought I had the answer worked out. Then last month, I was talking with a revenue manager at a boutique property in Austin. She pulled up her dashboard and showed me something that made me stop. "Look at this," she said, pointing to a booking at 2:47 AM. "$265. We had it listed at $380 six hours earlier." "Happens all the time now," she added. "Different travelers. Same pattern. 2-4 AM. They know exactly when to book." I started asking around. Other properties. Different cities. Same story. And that's when I realized what was actually happening.

The Pattern Nobody's Talking About

The instinct when you see this? Make your systems less predictable. Harder to game. But here's what I kept hearing from every revenue manager I talked to: "We spent years building this system. It works. It dynamically prices based on demand signals, booking velocity, competitive set, day-of-week patterns." One property in Dubai: "Our RM system is sophisticated. That's supposed to be our advantage." Except now, that sophistication is the vulnerability.

Customer AIs aren't just getting good deals, they're learning the patterns. When rates drop. What triggers panic pricing. How long a property holds before it blinks. And here's the part that kept me up at night: they're not learning alone. One AI books at 2 AM and gets $265. What if that data point feeds the network? What if next week, 50 other AIs probe at 2 AM? What if the week after, 500? The infrastructure for this coordination exists. Moltbook demonstrated agents can learn from each other. The question isn't if. It's when. The revenue managers I talked to? They can't respond by being "less predictable" without breaking the model that actually works. And if they hold firm? The AI just books with a competitor whose system it's already figured out.

Can RM systems adapt? Of course. They'll incorporate noise, detect probing, use AI to counter AI. But that's the point, it becomes an arms race where the competitive advantage shifts from operational excellence to... what exactly?

What I Didn't Expect to Find

The Austin revenue manager said something else that stuck with me about the guest who booked at 2 AM? They lost $115 in margin. But honestly? That's not what bothers her. Most disturbing was she knew that the guest got a transaction. Not an experience. Their concierge knows the city intimately. The AI optimized for price and filtered all of that out. Based on the traveler’s history, they would have loved this property. But their AI never showed them why. It just showed them the price.

And there it was. When millions of bookings become pure price optimization, purchasing experiences stops being something people care about. You get algorithmically optimal accommodations. Comfortable. Efficient. Completely forgettable. The real question isn't "How do I beat customer AIs at pricing?" It's "How do I ensure my value is visible to the AIs making decisions while creating experiences worth overriding their recommendations?"

That's a fundamentally different problem.

The Numbers That Changed My Mind

I was skeptical about how widespread this was. So I looked at the data. Consumer AI agents are proliferating. Trip planning platforms with millions of users. Price prediction tools learning supplier patterns. Booking agents that monitor and rebook automatically.

Industry data: 25% of consumers are comfortable letting AI plan their travel. 40% already use AI tools for part of trip planning. But the numbers don't show that these AIs are learning from each other. Not through explicit coordination. Through observation and pattern recognition across millions of bookings. This isn't individual travelers making individual decisions. This is collective intelligence learning to extract maximum value from supplier systems. And the entire dynamic assumes the only thing that matters is price. They're filtering out distinctiveness, meaning, experience, anything that can't be quantified and algorithmically compared.

The opportunity: Being valuable in ways AI can see, while being distinctive in ways that create override-worthy experiences. But that led me to a conversation that made everything click.

Last March, I wrote about this problem: "Exclusivity may make you invisible." Independent hotels told me "Our guests find us" and "We don't need AI; our business is built on referrals." They weren't wrong, in a world where humans searched directly. Then I talked to an owner of a property I've stayed at three times. Incredible service. “I deliberately avoided booking platforms," he told me. "Protecting margins, owning customer relationships, maintaining brand control. It made sense." He added that a guest called last week. Said her AI recommended a chain hotel instead of us for her trip. She knows us. Loves us. But her AI couldn't find us, we didn't match her preference profile.

He paused.

I asked what her preferences were. Thursday night live music. Properties that know the neighborhood. "That's literally us." The problem: that value lives in word-of-mouth stories, subjective descriptions, his relationships, formats the AI can't parse. Meanwhile, a chain hotel has local partnerships: yes in structured data fields. Live entertainment: weekly in their property management system. The AI can match these data points to the traveler's preference profile. The AI recommended the chain hotel. Not because it's a better match. Because it's the best match the AI can algorithmically verify.

The property doesn't lose on experience. It loses on algorithmic visibility and preference matching.

What This Actually Means

I've been thinking about this for months now. Talking to operators. Watching the patterns emerge. Here's what I keep coming back to: When algorithmic filtering systematically removes distinctive experiences, travel becomes transactional logistics. When travel is just logistics, people stop caring about it. You can't sustain a massive industry built on transactions nobody remembers. In an AI-mediated world, being amazing isn't enough. You have to be algorithmically visible AND experientially distinctive. Most operators are solving neither. Some are solving one. Almost nobody is solving both.

The questions I'm working through now:

What makes a property worth overriding an AI's recommendation? Smooth check-in isn't experience. It's infrastructure. Good breakfast isn't experience. It's expected operations.

How do distinctive properties become visible to AIs making decisions? If customer AIs are the ones choosing, how do they discover what makes a property special?

Where does defensible margin live when logistics commoditize? When operational excellence becomes universal and worthless, what capabilities actually matter?

I don't have all the answers yet. But the operators I'm talking to who are thinking through these questions now, the ones repositioning for a world where problems don't exist, those are the ones who'll define what comes next.

In the final piece, I'll walk through what I've learned about the hard decisions:

• How to actually build for this world

• What it means to be algorithmically compatible without sacrificing distinctiveness

• How to create override-worthy experiences

• Where to invest when operations commoditize

These patterns are emerging now. The question isn't if they become dominant, it's when, and whether operators recognize the shift before it's irreversible. Because when operational problems die, human experiences become everything.

One thing I haven't covered here is loyalty. I've been working through a pattern that completely inverts how we think about engagement and customer relationships in the AI age. If you're interested in a separate piece on this, let me know.