Navan CEO Ariel Cohen on 40% revenue growth, AI-powered travel booking, and building proprietary models for 30% of calls

Navan: 40% Revenue Growth, AI-Powered Travel, and Proprietary Models

Navan, the corporate travel and expense platform, just closed what co-founder Ilan Twig calls an "amazing quarter" — 40% revenue growth year-over-year and 50% usage growth — despite a period that included TSA strikes, major storms across the US, and volatile gas prices. The numbers make the case that business travel is structurally resilient: people showed up anyway.

What's driving growth

Three forces are running simultaneously. First, underlying demand. Twig's core belief is that face-to-face meetings are non-negotiable for closing deals, building teams, and hiring — and the quarter's usage data supports that.

Second, AI companies are disproportionately adopting Navan. OpenAI, Anthropic, and Anysphere (Cursor) are all customers, and Twig argues Navan is one of the few B2B enterprise platforms that can walk into any company's existing AI initiative and offer a concrete, measurable return — time saved for sales teams and C-suite executives on trip changes, rebooking, and expense reconciliation.

Third, product-led growth has more than doubled year-over-year, driven by AI-enabled marketing efficiency.

“Our revenue went up by 40%, our usage by 50%, and this is in this quarter. Most of the AI companies are using us, including OpenAI and Anthropic and Cursor. 30% of our AI calls are now our own model and not the frontier models or open source — that makes us more accurate, faster, raises customer satisfaction, and obviously it's also cheaper.”

How the AI is actually built

Navan has been using machine learning since 2016 to personalize travel results and optimize spend. Twig says customers save 15% of their total travel budget by booking through the platform.

The Gen AI layer goes further. Navan's orchestration platform, called Navan Cognition, coordinates between AI agents and human agents in real time — routing support queries, managing mid-trip disruptions, and connecting to the physical world where APIs don't exist. The clearest example: when a traveler is stuck on a delayed flight and risks losing a hotel reservation, Navan's AI bot calls the hotel directly, speaks to reception, provides a virtual card to secure the room, and pushes a confirmation to the traveler's app. No human agent required, no cost.

30% of Navan's AI calls now run on its own proprietary models rather than frontier or open-source models. Twig says the shift improves accuracy, speed, and customer satisfaction, and reduces cost.

Memory and loyalty

Navan's onboarding flow automatically identifies and imports a user's loyalty program numbers and status levels across every relevant club. The platform then uses that data to personalize flight and hotel recommendations throughout the booking experience, a function Twig describes as Navan's own memory component within Cognition.

The consumer-facing product, Navan Edge, handles end-to-end trip planning — flights, hotels, restaurants, events, ground transport — with a conversational AI interface called Ava that Twig positions as a genuine travel agent replacement, not a chatbot wrapper.

Navan is also tracking real-time airport conditions, including TSA wait times, and pushing proactive alerts to travelers before and during trips. During the TSA contract dispute earlier this year, Twig says internal data showed conditions at most airports were materially better than media coverage suggested — JFK was the exception, not the rule.