Seasoned corporate travel veterans might be conditioned to roll their eyes when they hear about transformative technologies for the industry. Think back six years when blockchain was the industry buzzword. Or think of the more-than-decades-long slog toward New Distribution Capability adoption. It's not that these technologies don't still stand to transform the industry; it's just a long process to replace legacy technologies and mindsets.
Agentic AI, however, seems to be a different story. In the inaugural BTN Next News Desk a few weeks back, panelists said a rapid "hockey stick" adoption of agentic AI in corporate travel is realistic.
"When NDC started, in some cases airlines were starting to do NDC, but they didn't necessarily have a problem they were trying to solve; they weren't necessarily differentiating content or building better servicing," Spotnana director of content partners Seth Anagnostis said. "When it comes to AI, you're going to have the opposite thing happen. You have a set of super-concrete problems … you're trying to solve. Also, it's not terribly expensive to make these investments; you can try a lot of things to see how they work."
Travelers and travel buyers alike are ready as well. A recent Amadeus Hospitality-sponsored survey of 6,000 travelers showed more than half of business travelers would be willing to let AI make bookings for them. A recent Serko/Sabre survey of 300 travel managers showed that 90 percent already are using AI or generative AI tools within their programs, most widely to streamline the booking process.
With this rapid adoption, some industry leaders warn that buyers and suppliers who fail to adapt could find themselves out of a job—or out of business.
Building the Basics
So, what exactly is "agentic AI?" Unlike generative AI—which requires prompts to generate content or complete a task—agentic AI can make plans and complete tasks with little to no human intervention. Asking an AI engine to draw an anime version of the "Seinfeld" cast is using generative AI. Riding in a self-driving car is using agentic AI.
In the context of corporate travel, the "material investment" in that technology will be around operational and servicing efficiency, said Jay Richmond, Amadeus senior director of global business accounts and solution consulting for the Americas.
"When we think about the ability to service, particularly during times of disruption or responding to questions, the faster and cheaper that it can be done, the better experience it is for the traveler, and the better it is for the corporation's travel program," Richmond said.
Many emerging applications for corporate travel that tap agentic AI are copilots or virtual assistants designed to handle simple, repetitive tasks. For travel management companies, answering traveler emails is one area where AI is finding growing use.
That was the target for Oversee, formerly FairFly, with its AgentSee solution, launched last year. The tool can respond to simple inquiries from travelers—checking on flight status, for example—as well as some complex inquiries such as group bookings, and respond directly to them, a step beyond other solutions that either simply triage emails or compose responses that agents have to send. So far, agency users have been able to automate up to 30 percent of requests even activating only a portion of the available use cases. The company said that when all use cases are activated, that could get as high as 75 percent.
Oversee CEO and founder Aviel Siman-Tov said agencies using the tool have reported significant improvement in efficiency.
"Instead of handling a growing number of repetitive, time-consuming customer service requests—many of them are resource-intensive, non-revenue generating and prone to human error—they're now being handled by our AI agent product," Siman-Tov said. "It's a dramatic shift in focus. Our customers can provide their clients the white-glove treatment they want, freeing agents to focus on revenue generating-activities and increase [service-level agreements, such as average handling time] to impressive levels."
Traveler-facing AI assistants also are gaining traction. SkyLink, which won BTN's 2024 Innovation Faceoff, is capable of helping corporate travelers book, get answers to questions and manage disruptions within enterprise chat channels. It is capable of recognizing information in a traveler's profile and policy and can prompt suggestions, such as a hotel if a traveler books a flight without booking lodging.
Over the past several months, SkyLink has become "significantly more intelligent," CEO and cofounder Atyab Bhatti said.
"SkyLink can understand your travel patterns a lot better," he said. "Not only will it be able to serve you up compliant things within your company's policy, it will be able to do it based on what airline you like to fly at what time you like to fly. It will be able to personalize it very distinctively to your own particular preferences."
The SkyLink team also has deepened integrations, such as within Google Maps, and has added capabilities so the tool can "contextualize" travel plans. For example, it can give guidance on what travelers should pack if they're going to face cold weather, or it can give guidance on visa or other entry requirements, Bhatti said.
Steve Singh's Madrona Venture Group also is a leading investor in another AI travel agent under development, Otto, founded and led by former Egencia chief product officer Michael Gulmann. Otto was designed to function like a virtual executive assistant, in which travelers can conversationally provide their preferences—travel habits, which airlines they prefer or the time of day they generally like to fly, for example—and Otto can put together travel plans tailored to those preference, Gulmann said.
For example, a traveler could tell Otto that they have a business trip in a city at a certain time and location, and Otto could provide a set of options that match those needs and the known preferences. With Spotnana available as a source of content, it can bring back the best options for the trip and book them within the tool. If the parameters of the trip change—the meeting time or location changes, for example—the traveler can tell Otto, and it can make the corresponding adjustments to travel plans.
"It's all just like an executive assistant would do," Gulmann said. "There's still a lot of capabilities we want to build in—integration with expense and other systems—but we have the world's supply for hotel and the world's supply for air."
Otto currently is in "limited beta," adding about 20 or 30 users per week who are using it to make actual bookings. A wider opening to more users is about a month or two down the road, according to Gulmann.
What's Next?
Those are just a few examples that will be expanding use of agentic AI in the corporate travel space. There are numerous other possibilities outside of travel planning and management in various stages of development. Agentic AI will help travel managers craft and fine-tune travel policies, negotiate with suppliers, manage expenses and reach sustainability goals.
"Agentic AI is: How can I put forward a goal? How can I put forward something I'm after?" Serko head of AI Andrew Revell said. "What if I put more responsibility and activity in the AI? There is simply more sophistication to it, and you can get to the point where the AI plans its way through a task and relieves you the need to plan your way through that task."
Singh, for example, sees the possibility of a world with "predictive servicing that's completely automated." If a traveler lands on time but is stuck on the tarmac without a gate at the airport for a few hours, automated AI servicing could notify the car service to come later and the hotel that the traveler wouldn't be arriving until 1 a.m., he said.
Intent, of course, can be hard to divine. If a traveler's flight is delayed by three hours, their plans might be loose enough that it doesn't matter and they want to just keep the flight, or that might be enough to make them miss their meeting completely. With such data as what is in calendars, however, AI could determine not only the impact of the delay on that meeting but also alert other attendees so that they could begin making alternate plans.
"A lot of that can be done in the agentic model," Singh said. "Will it be perfect initially? Of course not, but as more systems learn and become opening, allowing you to access data, an example like that can be completely digitized."
As those advancements occur, AI skills will become a more crucial part of the professional skill sets for travel buyers and suppliers alike. LinkedIn's annual Skills on the Rise report, published last month, listed AI literacy as the fastest-growing skill in which professionals should be investing.
Serko co-founder and CEO Darrin Grafton said that virtually 100 percent of his team has gone through training on AI. One travel buyer—and not one from a technology company—also recently told BTN her company is requiring AI training for employees to see how it could make their jobs more efficient.
Even absent formal training, the high adoption reflected in the Serko/Sabre survey shows that those who are not learning the skills risk being left behind.
"People are playing around with AI," Grafton said. "It may be unstructured at this point, but you see people looking at how it benefits their job and speeds them up."
The skills and experience of agents will become more valuable at the same time, Gulmann said.
"There's always going to be a role for highly skilled travel agents," he said. "The person who answers the phone and is really clicking through Concur or Expedia on your behalf, that is going to be replaced, but the person who is able to work magic and get you on that flight and suggest 14 different options to you because there's a massive disruption, it's going to be a while before the agentic systems can replace that."
That potential to take away repetitive tasks and free up such skilled agents in times of disruption is what's likely to lead to that "hockey stick" adoption of AI, Flight Centre Travel Group global manager of travel distribution Nicola Ping said. "Having someone sitting there and doing those [repetitive tasks] when you have lots of people who have complex scenarios who do need to talk to someone, we have to maximize the use of the human," she said. "Now, that is the bit that's really going to take off, as the technology is improving really quite quickly in this space."
For those TMCs not investing in AI, however, it's more of an existential question, Siman-Tov said. He said he recently had to correct a TMC executive who told him AI adoption was in "baby steps" mode.
"From what we are seeing, AI is already running marathons with AI-based customer service automation moving from visionary to actuality," Siman-Tov said. "If I adopt AI into my operations, I'm much more competitive, my customer service is improved and I am able to focus on more strategic tasks. Companies that do not realize they need to leverage AI as a differentiator quickly will probably not exist in five years."