FCM Takes First Steps for TMC Chatbots

business manager travelling
Travel News 21 Feb 2017

More than 30,000 branded chatbots, apps that use natural language to interact with users, launched in 2016, according to the publication VentureBeat. That includes one in corporate travel: FCM Travel Solutions' SmartAssistant for Mobile, or Sam. "Sam interprets and preempts requests and actions," said Marcus Eklund, global general manager for the travel management company.

Among the tasks Sam can perform is prompting travelers at appropriate moments to book ground transportation transfers within policy—or prompting them to change their travel arrangements if it detects a disruption to an arranged trip. According to FCM, Sam also deploys artificial intelligence to understand travelers' personal preferences better and refine the suggestions it makes based on its deepening knowledge.

In July, Sam became available on the iOS platform via the Apple app store in the U.S. Now Android and Apple watch owners can use it, too. FCM will release Sam in Europe in August or September and in Asia in November. FCM predicts that by 2020 "text or voice-based interfaces will be the new norm for communicating with the Web," but the revolution is not happening overnight. In reality, Sam, which was developed in Boston and Barcelona, is still being beta tested with two clients in the U.S. And while Sam generates all notifications to travelers automatically, only 20 percent of chat responses to travelers are generated automatically. FCM employees handle the rest.

Eklund believes the automatic chat responses will rise to 80 percent as Sam learns. How far will chatbots go in relieving human TMC employees of the task of engaging with travelers? "It will never be 100 percent; that's for sure," said Eklund. "But we hope to automate the 20 percent of questions which take up 80 percent of our people's time, such as, 'How much would it cost to change my flight?'" What won't be automated will be answering questions like, "Should I fly to India before I fly to Vietnam?" That's the kind of cost-conscious itinerary construction for which FCM currently retains consultants.

At a London briefing, FCM offered tangible examples of what Sam can do, though the demo was a PowerPoint presentation, not live. Drawing both on the user profile of a traveler Sam knows is a soccer fan and on the traveler's passenger name record, Sam recommends an evening soccer match occurring during the traveler's trip to Paris the following week. Then Sam offers to book and pay for the ticket using the traveler's payment info.

A few hours before the flight, Sam asks the traveler if he or she needs transportation to the airport. If the traveler says yes, Sam asks whether from the office or home. Because Sam knows these locations, it calculates the estimated costs and times to the airport of different options before connecting the traveler to an approved ground transportation company.

Toward the end of the trip, Sam notices that traffic to the airport at Paris is heavy and alerts the traveler either to leave 30 minutes earlier or to book a later flight. If the traveler chooses the latter, Sam searches for new flights, but seeing that these fall outside policy, offers a Eurostar train ride instead. The traveler opts to take a later flight, and Sam, aware that the traveler has two hours to kill at the airport, suggests this might be a good time to manage expense receipts.

Some of these two-way communications have been available since ConTgo, now owned by Concur, launched a decade ago. The difference today is that there is even more data, and AI, to act on it. "The new AI technology is a self-learning mechanism that understands the patterns of what travelers like and links them with other patterns," said Eklund. "Today, there is so much data that we can't build rules around it all. AI allows the computer to start making correlations by itself, although there are always framework rules to govern behavior. Until now, corporate travel has been notoriously poor at leveraging information."

Eklund believes traveler feedback from mobile surveys will personalize chatbots' service further. Social media comments can do the same. "We don't do it today and we are very aware of data privacy restrictions, but we are starting to play around to see what information it will bring," he said.