Agentic commerce in travel: Who wins, who loses and what to do about it

In January 2025, I published the article "Why AI agents will supercharge, not bypass, OTAs," which became the year's most popular opinion piece on PhocusWire. 

Fast forward to March 2026. I was on stage before an audience of over 250 marketing experts from the world's leading travel companies at the Phocuswright AI Marketing Summit, asking a simple question: "How many of you have tracked at least one end-to-end agentic booking so far?"

Only one hand went up.

That moment captures the tension at the heart of this conversation. AI disruption is simultaneously the most-discussed topic in travel technology and one of the least operationally real. The LinkedIn posts are everywhere. The actual bookings, almost nowhere. And yet the structural forces reshaping how travelers research, compare and buy travel are real.

The honest answer to "when will this fully arrive?" is: We don't know. The pace at which AI agents will take over travel shopping is uncertain—shaped by technology development, agentic protocols, consumer adoption and regulatory dynamics. 

But to what degree each travel player is structurally vulnerable to that disruption? That is what I tried to map, and the picture that emerges is far more nuanced—and far more actionable for travel executives.

Meet the Agentic Disruption Risk Matrix. 

It plots every major travel player category across two axes—how much agentic automation their product invites and how high their disruption risk is as a result. The output is a bubble chart where the size of each bubble reflects the breadth of players in that zone. What it shows—visually and immediately—is that the largest industry cluster is at low/medium risk, a small but significant cohort faces genuinely high exposure and a small group sits at very low risk.

Human-led: Human expertise gets more valuable, not less

Here's the first counterintuitive finding: the players at the greatest risk from agentic commerce are not the ones with the most human involvement—they're the ones with the least.

Luxury and bespoke travel specialists, adventure and expedition operators, safari camps and wilderness lodges—these players sit in what you might call the permanent human-led zone. 

Their products are co-designed with the client, their value is inseparable from the depth of human expertise behind them, and their booking process is fundamentally iterative, trust-dependent, and impossible to reduce to a query and a confirmation.

These players should not be bracing for disruption. They should be doubling down on the depth and specificity of what they know.

Agent-assisted: How the biggest cluster avoids agentic disruption

The agent-assisted zone is the most populated part of the risk landscape—and among the least threatened. This primarily comes down to three structural advantages that these players built long before agentic commerce appeared and that now function as powerful moats precisely because agents are optimization machines.

1. Loyalty programs

Hotel chains like Marriott and Hilton have built preference layers that agents are designed to honor, not bypass. When a traveler instructs their agent to prioritize Bonvoy status or Hilton Honors benefits, the chain becomes the default answer before search even begins. 

Full-service carriers benefit from the same dynamic—a business traveler whose agent is calibrated to protect British Airways Executive Club tier points is not a customer available to the lowest bidder. Global horizontal OTAs are increasingly in this conversation too: Booking.com's Genius program and Expedia's One Key create member pricing tiers and cross-platform rewards that agents can surface and optimize for, adding a loyalty dimension to what was previously a pure-search intermediary model.

Loyalty works in an agentic world because it converts consumer preference into machine-readable decision logic. The brands that built these programs are becoming the default in an era of agent-mediated decisions.

2. Unique and irreplaceable inventory

Vacation rental platforms like Airbnb hold inventory that an agent simply cannot source elsewhere. A hand-built treehouse in the Costa Rican cloud forest has no equivalent alternative. An agent asked to find something cheaper will return empty-handed. 

This dynamic is also happening at the less glamorous end of the spectrum. Platforms that have spent years digitizing fragmented, offline supply—ground transport across Southeast Asia, where thousands of independent bus and minivan operators run routes with no API and no structured pricing—now hold something genuinely scarce: structured, bookable inventory no global player can replicate. 

The same is true of platforms connecting travelers to individual guides and micro-operators running authentic local experiences: the marine biologist leading reef walks on a remote Indonesian island, the local photographer taking small groups through Oaxaca's back streets. 

Once this supply is structured and discoverable, it becomes extraordinarily valuable to agents—and only the platforms that did the hard work of building it will have it to offer.

3. Bundling complexity

Global horizontal OTAs package flights, hotels, transfers and experiences into combinations that a single-service agent cannot easily replicate with the same traveler assistance and guarantees by going direct to each supplier. 

Package holiday platforms add a further layer—pre-assembled itineraries with contracted capacity that have no unbundled equivalent. The more moving parts a product has, the harder it is for an agent to disaggregate and reconstruct from raw inventory. In the agentic era, bundling is more than a commercial strategy—it’s an architectural defense.

Managed agent: the TMC paradox

The conventional reading for travel management companies (TMCs) is straightforward: routine booking and policy compliance are highly automatable, and agent AI will absorb much of that workload efficiently and cheaply. That part is correct.

But the analysis stops too soon.

When major disruptions happen, the behavior of business travelers changes. They pick up the phone. They need someone with judgment, empathy and real-time problem-solving capability to navigate a situation that doesn't fit neatly into a policy rule or an API response. 

Rebooking under pressure, coordinating duty-of-care obligations, making calls that require both factual knowledge and human reading of a situation—these are not tasks that agents handle well, and travelers know it. TMCs should lean into this—repositioning around disruption management and risk advisory.

Full autonomy: Four players who should be paying close attention

In the high-disruption end of the matrix four cases stand out—not because they're failing businesses, but because their product characteristics align almost perfectly with what an AI agent does best.

Second-tier OTAs are arguably the most structurally exposed intermediaries in the entire travel ecosystem. Their customer acquisition model is built mostly on traffic arbitrage around metasearch and Google search traffic. There is no loyalty layer, no bundling advantage, and no brand strong enough for a traveler to instruct their agent to use it. Agents bypass metasearch by design, querying supplier APIs directly or routing through better-positioned platforms.

Low-cost carriers face a mirror-image problem from the supplier side. Price is the only meaningful differentiator. There is no premium cabin to protect, no lounge network to value, no loyalty program worth encoding into an agent's decision logic. The entire LCC proposition—find the cheapest seat on the right route and book it—is a precise description of what agent AI does natively. 

Independent hotels are perhaps the most underappreciated case. Without the loyalty infrastructure of a chain and no brand that an agent's decision logic would be calibrated to favor, independent hotels competing in the commodity stay segment—particularly in business travel—are exposed. An agent booking a two-night stay for a consultant in Munich will route to the hotel establishment that delivers the most measurable value to the traveler's preference stack. 

The navigation imperative: Strategy depends on where you sit

Understanding your disruption risk is the starting point, not the destination. The more important question for any travel player is: Given where I sit on the agentic spectrum, what should I actually do?

That question is what the following chart is designed to answer. It maps the four levels of agentic adoption to the phase of disruption each represents and highlights the specific strategic focus areas that matter most. It's a practical translation of the risk landscape into a set of choices, because the right response for a luxury bespoke agency is almost the opposite of the right response for a low-cost carrier.

Players in the human-led zone—the luxury specialists, expedition operators, boutique agencies—should resist the temptation to automate for automation's sake. Their job is to deepen expertise, build client relationships and make the case for why human judgment commands a premium. 

Players in the agent-assisted zone need to win the discovery layer. That means making sure their inventory is structured, rich and visible to agent retrieval systems. It means exposing loyalty benefits via API so agents can surface them. It means differentiating not just on price and availability, but on the structured content, sustainability data and personalization signals that give agents something to work with beyond a rate.

And players staring down full autonomy face a starker choice: compete aggressively on price, API reach and reliability or start thinking hard about what adjacent value—experiences, ancillaries, loyalty, data—might restore the margin that agentic commoditization will otherwise erode.

Stop reading the hype. Start reading the map.

The agentic commerce debate will run for years. There will be more LinkedIn and X posts that look alarming and say very little. Most of it will treat the industry as a single entity facing a single threat—and most of it will therefore be useless to the executives who actually have to make decisions.

The question is not, “What will AI agents do to travel?” The more productive conversation starts by asking, “Where does my business sit in this disruption landscape, and what does that mean for the choices I need to make today?”

That question has a specific answer for every player in this ecosystem. The risk landscape is mapped. The strategic imperatives for each zone are knowable. Now is the moment to locate yourself on that map and place your bets to start building an agentic moat that will hold.