Dawn of the Travel Chatbot

In the increasingly complex world of post-pandemic business travel, some travel buyers are looking to chatbots to simplify processes and communicate directly with travelers.

Chatbot technology uses natural language processing, which relies on AI-powered models to accurately understand and respond. While big business for some time has embraced this tech, it has just started to make its way into managed travel programs.

“Post-Covid, there was a shortage of expertise and lots of questions about the rules, policy, expense allowance, duty of care, a whole plethora of things as business travelers re-familiarized themselves with all the necessary information to make a return to travel,” said Adam Kerr, CEO and founder of business travel booking and planning platform Tripism, which plans to launch an intuitive chat functionality in 2024.

“The chatbot was seen as a solution for some of the more straightforward questions, freeing up resources and agency time, and therefore cost,” he said, adding that some enterprise clients have built chatbots within their own organizations.

While some corporates have developed their own closed-loop systems, others are working with tech startups or piloting new integrations to explore different use cases.

Toyota North America is looking to expand the use of an existing chatbot, currently deployed to offer internal IT support, to help improve efficiencies around traveler communication, said travel services manager Rebecca Jeffries.

“Right now, we’re in the process of filling out a very long FAQ document to feed into [the chatbot], and we hope to go live by December or January,” she said.

The auto manufacturer’s virtual agent, a system known as Agent Ask, is integrated with Microsoft Sharepoint and accessible via Teams. Once deployed as part of the travel program, Jeffries hopes the bot will automate responses to many of the policy-related inquiries posted in her Ask Travel Services channel on Teams, which she said currently has 11,000 U.S. subscribers.

“We’ve noticed in the last year that the number of questions has dramatically increased, and it takes up a lot of our time,” she said. “We’re hoping to get out of the minutiae of answering the same questions over and over and redirect people to the resources that are already available, because they can answer their own questions.”

Jeffries envisions the chatbot will send personalized Teams notifications and follow-up emails to remind travelers to renew passports or soon-to-expire credit cards. She also hopes the tool will learn how to decipher between the types of inquiries and route customer service-related questions to her travel management company, which “will all be part of the AI learning,” she said.

EY’s Chatbot Launch

Consulting giant EY noticed a similar uptick in traveler inquiries following the Covid-19 outbreak and developed a chatbot solution to streamline processes for its travel teams and call center staff.

“We transformed a static FAQ document into a user-friendly chatbot,” explained Patrícia Bircak, Latin America South travel, meetings and events manager at EY, who built the team’s first chatbot pilot using Microsoft Power Virtual Agents.

Following an initial launch in Brazil, the chatbot has since been deployed in multiple languages across several regions, including Oceania, Asia, Italy and the U.S., with content that has been customized according to local needs.

“We developed a guideline master document, and regional travel managers began their own FAQ documents to add to [their local] chatbot,” she said. 

Earlier this year, the company also partnered with an AI startup to automate responses to email-based travel inquiries.

“If the self-service chatbot can’t answer a question, [the AI] will then send the inquiry to a mailbox that’s monitored by our own Travel Service Center,” said Wendy Meier, Americas innovation manager for travel, meetings and events at EY.

Meier said the offshore service team also utilizes the bot’s natural language processing and emotion detection to help generate appropriate responses to inquiries. “Previously, we had some struggles with ambiguous questions and, at times, inconsistent responses, causing our service center to go back to the onshore team and ask” for clarifications, Meier said. 

“Through our third-party supplier, they’re now able to copy and paste the question right into the tool, which then detects the tone of the question to suggest an appropriate response. The team can then agree or disagree [with the automated response] and the bot learns as it goes.”

Inquiries are most often linked to policy, but Meier said the bot can address “anything to do with travel, meetings and events,” including making hotel recommendations. She added that many travelers “aren’t even aware” a bot is responding to their email inquiries. “It’s just helping our travel service center be more efficient,” she said. 

Adobe Pushes Compliance

Software company Adobe also is exploring use cases for AI-powered messaging across its travel program, largely to drive expense and policy compliance.

“Our expense team have been using [the technology] for about a year,” said Kathy Burdge, global travel and expense compliance manager at Adobe. “This has been quite successful, because we’re able to collect reasons for noncompliance, therefore giving us the intelligence to review policy.”

For example, Burdge said the tool has helped to provide trends around meal expenses and high-cost cities. “Rather than collecting anecdotal data, we were able to automate [this data collection] through the AI chatbot, and we started to see some trends around particular cities,” she said. “We were able to then make some intelligent decisions to adjust our high-cost city list without all of the manual efforts that we’ve historically had to do in that space.”

The proprietary bot, which was built internally as a closed system, currently receives data via a SFTP—secure file transfer protocol—feed from the company’s expense tool, Concur, and then automates targeted messages via email.

The AI “goes beyond some of the messaging that we have in the expense process, where people just click through, to now become a follow-up,” Burdge said, adding that the bot’s friendly tone has increased engagement.

Burdge also hopes to expand the tool’s reach in 2024 to feed in data from Navan, Adobe’s TMC, in order to identify and target travelers who might be repeatedly violating policy.

“The way we envision it working with Navan, we’re pulling all the data in through API, [and] because we have the data in our own data lake, we’re able to parse it in and use that data in our own messaging and system,” she explained.

“[It will] help us identify where the real problems are, where there may be potential fraud versus education opportunities, or perhaps policy considerations that we need to take into account,” she added.

Integration Fuels Innovation

The travel team at Microsoft in September launched a live integration between Amadeus’ Cytric booking and expense platform and the Microsoft 365 suite of tools for its U.S.-based travelers.

The integration allows travelers to search and book trips within Microsoft Teams and also share trip information with colleagues via the app, giving them the option to book the same trip.

“We’re using Teams both to chat with an agent, but also to go in and make a booking using Cytric Easy,” said Microsoft global travel director Eric Bailey. The goal, he explained, is to make the travel booking experience as efficient as possible, “not to go out to separate tools or break up the flow of your day.”

With the integration still in its early stages, Bailey acknowledged the learning curve in transitioning travelers from email inquiries and bookings to instant messaging.

“We are seeing some usage, but people are creatures of habit,” he said. “We’re really hoping that as we add more functionality down the road it will become more of the standard.”

Additional functionality could include an AI-enabled digital assistant that will make travel recommendations and on-trip itinerary suggestions.

“We want to make sure that we’re respecting the privacy of people, but also making it easy to share [travel data] with those people you want to share with,” Bailey said, adding that bridging the gap between privacy and personalization “is a challenge” in managed travel.

“We need to do a better job of understanding the profile of people,” he said. “What’s missing today is that AI can’t actually predict [your travel preferences], even though none of it is random. … The tech capability is there, but understanding how to capture that information is not there yet.”

Some tech startups are looking to address the issue of travel industry data fragmentation.

One blockchain-based solution, Travlr ID is building a network where entities – including booking tools, global distribution systems and other suppliers – that want to interact with a traveler profile can do so with the traveler’s or company’s permission, explained founder and creator Gee Mann.

“It’s a service for the traveler; not running in a central system somewhere,” he said. “Your data goes with you [across your corporate career] in your mobile wallet. In the future, this will also include an AI assistant that will be able to make personalized recommendations.”

Mann added that Travlr ID has been developed based on “live use cases” and from learnings in his adjacent business, Qurious, which provides “question-and-answer as a service” and uses generative AI to detect the sentiment behind inquiries.

“It’s more than just an FAQ bot,” Mann explained. “The AI is not just automating responses to streamline processes, but taking a proactive approach to analyze why people are asking questions… and this is where we can create a data-driven approach to travel and improve processes.”

Mann said Qurious is working with enterprise clients to build bespoke workflows for travel programs, where systems are engineered to preserve privacy. 

“You’ve got to be careful of what you feed [the bot] unless you’re hosting your own model,” he added. “You really have to do solutionizing around workflows… that’s where the real value is for AI in travel programs.”

While Toyota’s Jeffries is “really excited” to see how generative AI might evolve, she said she would rely on her TMC to deploy such a tool rather than build it internally.

EY, meanwhile, is developing a ChatGPT-solution which will enhance the travel experience of its employees by providing “much more relevant information at their fingertips”.

“It’s designed as a one-stop shop portal for pre-, during and post-travel and will include access to chat,” Meier said. “It will target specifically what the traveler needs when they need to see it, resulting in more of a push than a pull [approach].”

Adobe’s AI bot currently assesses expense data and reads emails, but Burdge hopes further integration into the travel program will include data feeds from traveler profiles.

“Where I think AI can be really powerful for us is around scalability… especially given the labor and talent shortages that we’re experiencing throughout the industry,” she said. “I don’t think AI is going to take anyone’s job away. But I think it will help us provide an experience and service to our users.”