AI may be playing a larger role in traveller research, but experts say technology capable of completing an entire travel booking without human intervention is still some way off.
While AI is being used as a search engine, complex airline distribution systems and a lack of consumer trust remain barriers to fully autonomous bookings.
Travel suppliers, OTAs and technology companies are working towards what is known as ‘agentic commerce’, where AI can complete every step of a booking, from searching and comparing options to making payment.
However, Lasse Vinther, MD of Automation Architects, believes travel’s complexity means the technology will be introduced gradually.
Opportunities for agents
Vinther believes AI’s biggest value for travel advisers will initially be seen as an assistant rather than a replacement.
“Agentic tools can automate the search-compare-enter-checkout drudgery, freeing advisers for high-value curated trips.”
Vinther notes that other benefits of agentic commerce for agents, other than advanced search-engine abilities, will eventually be its ability to aggregate seat inventory from multiple sources and channels, as well as scale servicing capabilities for disruption rebooking and customer support.
“While AI will initially be seen as a product information source, the actual conversions will be lacking due to trust and compliance when completing the actual transaction,” says Vinther.
Lack of trust
User confidence remains one of the biggest obstacles.
Vinther explains that many suppliers, OTAs and e-commerce sites are still not structured in ways that allow AI systems to easily interpret their product information, with some actively blocking AI from accessing their content.
This, along with concerns about paid partnerships with suppliers and kickbacks, makes users uncertain about whether the search engine product recommendations provided are the best possible results for their prompt.
“Ranking bias and transparency are unresolved, but it is being increasingly regulated, particularly in the EU. AI answers can narrow choice to a handful of options, raising concerns about paid placement and hidden kickbacks. Research shows the OTAs that already dominate are the ones being cited most inside large language models (LLMs),” says Vinther.
Another challenge is willingness to allow AI to complete payment transactions. While certain consumers show more trust in autonomous payments than others, the industry has only just finalised some of the very first standards for AI to manage payments, he explains, emphasising that this development still does not assist the AI agents with managing complex travel product inventories.
Although standards allowing AI systems to manage payments have recently begun emerging, most AI assistants still hand customers back to airline or OTA websites to complete transactions.
"As of mid-2026, AI assistants such as ChatGPT, Gemini and Perplexity are mainstream for planning and inspiration and dozens of OTA, airline and metasearch deals are channelling live inventory into them, but almost all the LLMs (Large Language Models) still hand the actual booking back to the supplier's or OTA's own check-out systems," says Vinther.
Airlines add complexity
Direct airline distribution channels and NDC will be the main feeders of product information to LLMs and agentic commerce.
However, these platforms are already facing many challenges with airlines’ complex rules, bundles and ancillaries that vary by market and channel and AI is struggling to track this inventory, as is, explains Vinther.
“Just some of these complexities include real-time (dynamic) pricing and availability, fare rules and ancillaries, fragmented and non-standardised NDC and GDS inventory, cancellation policies and loyalty handling.”
AI search engines will be required to consolidate all this information before agentic commerce can materialise. While some experts say this will likely take more time to achieve through improved standardisation within airline distribution, others argue that trained models could absorb NDC complexity, reducing the importance of rigid standards on the search side, but not on the servicing side, explains Vinther.
Another factor that will likely slow the process is the lack of click capacity on airlines’ direct channels, as it will eventually have to handle millions of AI bot scraping and searches simultaneously.
Regardless of the complexities, Vinther says, the emergence of agentic commerce is inevitable in the travel industry.
“Automation Architects knows, first hand from ITC clients, that AI is already being used in the back office on behalf of clients. In the next year and half, we will start seeing selective autonomous commerce in aviation, in standardised, low-complexity segments, such as point-to-point flights, and in trust-leading markets with spend caps and mandates.”
