Hooked on a Feeling: Navan's AI Leans into Vibe-Based User Experience

"What we love about the way we're building AI … is it's changing the dynamic from your traditional model of [needing] to have a lot of intent or know exactly what I'm looking for into something that lets me also sort of talk about the vibe. I'm all about vibes."

Those are the words of Navan travel marketplace SVP Dane Molter, speaking Wednesday from the main stage of the travel management company's first-ever user conference, held in New York. The conference attracted more than 300 clients and well over 100 prospective clients to The Glasshouse, overlooking the Hudson River for the day-long event, which featured client panels with logo representation from the likes of Adobe, DoorDash, GE Healthcare and OpenAI, dispersed among presentations from Navan co-founders Ariel Cohen and Ilan Twig as well as senior technology and product executives.

As expected from Navan—which was early to hook its wagon to a machine learning star in 2015 and has moved quickly since 2023 into large language models and agentic AI—the artificial intelligence vibes were strong at the event. In his "Book with AI" demo, Molter showed the process of booking a business trip to Chicago to attend the Global Business Travel Association convention at McCormick Place. When the platform suggested he book a hotel nearby McCormick, Molter, a Chicago native who now lives in London, balked and told the platform that area was "boring" and was there somewhere more "hip" where he could stay. At that point, the platform brought in content from the West Loop, a neighborhood known for its restaurants, nightlife and shopping—and recommended a collection of boutique properties.

"This is really magical," he said. "I can talk about what the trip is going to feel like for me and I can actually book it. It's not just a discovery platform."

Molter later expanded upon the human-to-machine relationship in an interview with BTN, and how the large language models and AI agents now work in the background of Navan tools to calculate the context the user brings to the search—from their profile, historical searches and trips and now even more so from the richer content and sentiment details that AI agents can access and surface to the user. That's the step change from Navan's machine learning early days to the more powerful relationship-like experience the travel agency is pushing toward today.

"When you move away from forms and tables and into a context[-driven] conversational type experience where I can just tell it what I want to feel, it will completely change the dynamic of the way these travelers are experiencing that travel. That's where we think we can keep them in the platform and continue to really build that user experience in our direction," said Molter. Book with AI is not yet rolled out but is in an advanced testing stage with a number of Navan clients, according to Navan.

Program Administrators Need Vibes Too, Man

Navan never has been bashful about its focus on the individual traveler and putting that experience first. But the user conference wasn't for business travelers, and the audience of travel and finance managers, which is the core of the customer base, are looking for more intuitive ways to gain visibility into traveler activities, behavior and spend to simplify travel category management.

To that end, Navan officially introduced at the conference a new feature called Admin Companion, which functions as a kind of AI analyst for the program administrator. Admin Companion is tightly connected to Navan's analytics suite to allow travel and expense program managers to interrogate their data via natural language prompts. Like Booking with AI for traveler users, Companion currently is in beta with a handful of Navan's "best" clients, Molter said.

The demo showed it will offer standard analysis like what business groups or regions are generating the highest travel spend or where out-of-policy bookings are happening. Like the Booking with AI interface, Admin Companion users initiate search with natural language prompts and Companion returns the data. According to Molter, however, the tools go further to offer insights as to why these trends might be occurring.

"You can work with this to put hypotheses in place. [For example,] estimate why would it possibly be that in Amsterdam or in Prague, I have extremely high out-of-policy rates. [The tools] can come back with an insight that there's very little coverage here, but [the company] travels here a lot. It's a growing segment for you," he said. "And really the power of this comes with what you do with the data."

He mentioned a beta client had used the data to change their advance-booking window policy because the Admin Companion had helped them see that the in-policy range wasn't gaining them anything.

"It said, 'Hey, here's the data where I saw these bookings taking place here. You weren't actually getting any additional savings by pushing people into that 14-day window,' and it actually cites its sources back so the administrator can trust the analysis," Molter said.

Asked by BTN whether Navan is working on benchmarking tools outside of individual program data, Molter said some capabilities already were in the tool but more were on the horizon. He declined to put a timeframe around it.

"We want to make sure that whenever we expose benchmarking data … that we're not risking any other company's data exposure," he said. "So we've been running security reviews; we've been running a lot of tests on how we can expose this data without possibly exposing a customer's information. Until we're very confident that that's going to happen, the full suite of what benchmarking can do is not going to be there."

Sights Set on Enterprise

Navan has won a number of marquee logos in the past 18 months. Names like GE, Visa and Japanese banking holding company MUFG have entered its client portfolio. And according to current and prospective clients BTN has spoken to, the company is doing much of the necessary work to attract them—not only the vibe-y AI interfaces.

Molter spoke about a total rebuild of the company's policy engine for travel and expense that is happening now.

"We know we need to … give you more granular control. Things like destination [awareness] and type of travel. We know that there are places where you actually need more complex approval matrices, you need escalation paths," he said. "We're putting all of that into our new policy engine that we're rolling out through the course of this year. You're going to see these changes."

Navan is also upgrading its communications tools to understand where travelers are and improving tools to reach out to them while they are on the road.

Several buyers noted to BTN the advances the company had made on the expense side of the business and to watch Navan in that space. Navan claimed that transactions sent through its own payment workflows automatically were filed as expenses but also showed future capability for reporting out-of-pocket expenses that parsed receipt details in real time by pointing user's phone camera onto the receipt and voice prompting itemizations and allocations into the reporting tools. The company soon will roll out AI-powered expense auditing that only pushes anomalies through for human review. Buyers were more skeptical of Navan's efforts to make their integrated payment vehicles more global.

Courting more travel buyers and enterprise programs with a more mature and measured communications strategy notwithstanding, Navan hasn’t lost its characteristic brashness, either. Buyers at the conference still noted to BTN the company’s aggressive sales tactics, and their tendency to push the envelope. And now, perhaps emboldened by the fresh IPO, the agency feels ready to rumble.

Asked during a Q&A session who Navan’s competitors are in the travel space, co-founder and CTO Ilan Twigg couldn’t name one. Asked then why all Fortune 500 companies not yet cast their lots with Navan, Twig began thoughtfully enough: “It was very difficult for [my father] to digest the iPhone. So, for all of us, it was difficult to digest [that] I can talk, I can chat with a machine, and it can actually do good stuff for me. But now I see more and more people accepting it… and I do not know of any travel platform that uses AI the way we use it. And I assure you that every day marks the difference; every day you will see more and more and more [acceptance]. In time—remember the slide with Navan everywhere?—in time, everyone will use us. Your question has relevancy of maybe three months, six months.”