Has Agentic AI in Travel Commerce Taken Off in 2026? Complete Survey of Agents, MCP Servers, Protocols, and Approaches

Agentic commerce has been the subject of active discussion for several years, but 2026 marked a year of explosive growth and many first real-world use cases. IDC, one of the leading market intelligence companies covering information technology, predicts that by 2030, 30 percent of travel bookings will be executed by AI agents, accelerating investment in LLM optimization and increasing direct bookings and profitability.

At AltexSoft, we’ve been developing AI agents for travel and hospitality clients and are building early MCP server implementations — a young standard, but one we're already working with in practice.

In this survey, we'll break down what agentic AI actually means for travel today, how far its capabilities really reach, what's happening under the hood, and (most practically) how suppliers, OTAs, and TMCs are already putting it to work, with real examples from the industry.

What is agentic commerce and why should travel businesses consider it

Generally speaking, agentic AI refers to AI systems that can plan and execute complex, multi-step workflows with limited human intervention.

The rapid development of agentic AI has given rise to another increasingly common concept: agentic commerce. It describes a model of online shopping and transactions in which AI agents handle some or all of the process on a person’s behalf. In travel, for example, a customer might ask an AI agent to research destinations, find and compare hotels or resorts, and eventually complete a booking.

The channel is still at an early stage, so it is reasonable to ask why travel platforms should invest in it now. There are several reasons.

Better visibility. If AI agents can’t reliably access your services and inventory, your platform risks becoming invisible to a growing share of travelers. In the US, 56 percent of leisure travelers already use LLM-powered conversational interfaces for trip planning, and the shift may go further still: 80 percent of travelers say they are willing to collaborate with an AI agent to find the best options.

Even before agents can complete a booking, simply appearing in an AI assistant’s recommendations and being linked back to your site helps keep your brand in the traveler’s consideration set.

Higher-intent traffic. According to an Adobe survey, AI assistants are becoming an increasingly important source of customers. AI-referred traffic to US travel websites jumped 194 percent year over year in May 2026.  And it’s not just about growth: Amadeus reports that 44.7 percent of AI-driven visitors reach the booking engine, compared with 25.9 percent of organic visitors.

AI-influenced users tend to spend more time on travel sites, view more content, and abandon sessions less frequently. This suggests that AI sends travelers who have already done some research and narrowed down their preferences.

Lower distribution costs. For hotels that heavily rely on OTAs for distribution, particularly independent properties, agentic commerce creates cost-saving opportunities. As mentioned before, AI assistants can send travelers directly to a hotel’s website, potentially shifting more demand into the direct channel.

At the same time, a new generation of AI-native aggregators, such as DirectBooker or Lighthouse, offers access to agentic channels via shared MCP and API infrastructure, so hotels don’t have to build AI-facing integrations themselves. These emerging intermediaries usually charge considerably lower fees than traditional OTAs.

You can read about distribution costs and hidden payments in our dedicated article.

More control over offers. Whether travelers arrive directly from an AI assistant or through an AI-native aggregator, suppliers retain more control over what is sold and how it is presented. They remain a merchant of record, with their own rates, availability, ancillaries, loyalty benefits, personalization, and upselling logic flowing into the agentic channel.

Simpler user experience. But the most straightforward reason why it’s a good idea to sell via conversational interfaces is just that users love it. “I’ve been in this industry a long time, and people usually take the easiest route to technology,” says  Alan Young, VP of Hospitality Strategy and Product Management at global enterprise software company Infor. “If asking a platform gets you the result faster than navigating screens and clicking through menus, that’s what people will do. So I don’t see agentic commerce as a bubble. I see it as the natural progression of how we’ll interact with computers.”

Praveen Kumar, Co-Founder and Chief Technology Officer at TPConnects, a modern air retailing, airline content aggregation and distribution technology company, believes that over time, the conversational approach will replace the familiar web search altogether:

"Metasearch gave travelers choice. AI will give them the right choice. Instead of scrolling through hundreds of offers from airlines and OTAs, travelers will simply describe what they need, and an AI layer on top of multi-source, multi-content inventory will narrow it down for them. In air retailing, conversation will replace search.”

Key agentic protocols to know

Agentic commerce is evolving fast. OpenAI, Google, and others are introducing new protocols for commerce and payments. While these standards are not designed specifically for travel, OTAs, TMCs, hotel chains, and other businesses are already experimenting with them to support agent-driven shopping, booking, and servicing.

Model Context Protocol (MCP), an open communication standard designed by Anthropic, is the most common way for an agent to access external data and tools. Agentic AI is typically built around one or more LLMs with function-calling capabilities, enabling agents to interact with external systems, such as APIs and databases; MCP aims to make these interactions more consistent.

A2A (Agent2Agent) protocol, released by Google with contributions from over 50 tech partners and then donated to Linux Foundation, helps independent agents (for instance, on the seller and buyer sides) communicate and coordinate actions with each other, regardless of vendor or underlying tech stack. While already in use in production, A2A collaboration across different platforms is still more of a frontier than a mainstream.

Agentic Commerce Protocol (ACP), co-developed by OpenAI and Stripe, and Google’s Universal Commerce Protocol (UCP), released by Google, both cover key steps in the shopping journey, from product discovery and cart creation to checkout. They also outline interfaces and data formats suppliers need to expose so that agents can transact with their commerce systems.

ACP supports different payment providers, yet its first real-world implementation was ChatGPT’s Instant Checkout, launched in partnership with Stripe in 2025. The approach failed to gain significant traction: Just months later, OpenAI cited its limited flexibility and pivoted toward product discovery and merchant-owned checkout.

But this stumble didn’t kill agentic commerce. Google continues to roll out UCP-based shopping in AI Mode. It works with payment service providers (PSPs) that can process Google Pay tokens. Adoption is still early, but UCP is already expanding into travel, particularly hotel booking, with partners including Amadeus, Booking.com, Expedia, Hilton, IHG, Marriott, Trip.com, and Wyndham.  

Agent Payments Protocol (AP2), also originally offered by Google and later contributed to the FIDO Alliance for open, platform-agnostic governance, can serve as a specialized payment layer within UCP or other agentic commerce frameworks.

Standards alone don’t tell us how much of the travel journey agents can actually handle today. So let’s move from infrastructure to practice.

What AI agents in travel can (or can’t) currently do?

We compared major AI travel assistants to assess what they can do, from travel search and booking to checkout and post-booking servicing.