HTS adds AI-powered support tool for travel sellers

HTS, Hopper’s B2B arm, has launched a generative artificial intelligence (AI) assistant to help travel sellers improve how they support customers.

Called HTS Assist, the tool is already being used by some of HTS’ banking partners including Lotte Card and SMBC, with partners reporting an average 65% drop in servicing costs as a result, the company said.

"Through Assist's innovative Al technology, we are bringing a world-class offering to market, enabling airlines and travel providers to deliver faster resolutions, improve customer satisfaction and reduce costs, all while meeting travelers where they are today," said Jo Lai, senior vice president of Al solutions and customer experience at HTS.

How HTS Assist works

Frederic Lalonde, co-founder and CEO of Hopper, said the tool can handle all changes—such as cancellations, disruptions and refunds—and has been trained on conversations the company's human agents have had with customers around the world.

“It's fully transactional. It's able to actually deal with real customer issues, and it's been specifically trained for travel,” Lalonde said. “If you were to go to ChatGPT right now in the app and try to have this conversation with it, the language model would follow you, but it would hallucinate all sorts of things... that's because it's not designed for travel, or it doesn't have the connectivity.” 

Available in 200 markets and more than 30 languages, HTS Assist can also potentially unlock new revenue streams for providers by finding upsell and cross-sell opportunities, according to the company. 

And it’s less expensive than a human customer support agent, Lalonde said.

"The average cost of resolution is $1.12, which is a fraction of what [resolution typically] costs,” he said. “It actually goes to resolution four times faster than with a human so it's incredibly efficient, and it ends up being 92% cheaper for the operator. So obviously, those are the metrics that we've seen operating in our environment, we expect the same thing to happen when we push up to customers.”

Lalonde said HTS Assist is hooked into passenger systems of airlines, car rental companies and more.

“One of the reasons this thing actually works is because it uses the totality of the HTS connectivity that we've been building up for fintech and commerce for over a decade,” Lalonde said. “We've been building this up gradually in our commerce network.”

While the tool is capable of handling tasks independently, customers will be able to reach a human agent as well. 

“There [are] a lot of reasons why certain businesses might want to say, for a very specific use case, ‘we, for compliance reasons, [need this] to be handled by a person,’” she said. “And so by design, there are two to three different models that will be used.”

Lai expects adoption to take time given the newness of the technology. 

“The idea there is not so much that the AI can't do some things, I think that's not necessarily the case,” she said. “It's so connected to all the systems that it's not only capable of retrieving information, but actually end-to-end [action]. But it's really more either an active choice on the business or where a customer needs it for accessibility or other reasons.”

The future of travel and HTS

While HTS Assist is a new development for HTS, Lalonde sees a bigger change coming down the line for commerce in general and sees HTS Assist’s capabilities as indicative of that pending shift. 

“We will not be, as a species, playing around with filters and check boxes on Booking.com like a caveman for very long,” Lalonde said. So if anything else, it is completely obvious that the service is the first place you're going to start to see innovation, because it's such a pain point. It's a more constrained environment, but I would predict that you're going to see major changes in travel commerce in the next two or three years.”

Lalonde also sees HTS leaning into AI for its financial products. 

“We've been a big data company for a long time, doing a lot of machine learning,” he said. “We are starting to play around with predictive models that are fully, fully generated.

“I don't even know what's possible there, but the idea that you would be able to offer these policies cheaply, in more places, give better outcomes to customers, clear, that's going to log and then, of course, the fundamental way we shop for travel is about to change,” he said. “So we're not showing anything right now on that, but you could imagine that it's a huge area of focus for us.”