Ariel Cohen: The Death of Salesforce; How OpenAI is Changing the Travel Industry | 20VC #975

Cohen’s first startup, StreamOnce, failed to make meaningful impact because it targeted a small, unproven problem that nobody urgently needed solved; with Navan, he focused on a massive, obvious pain point—corporate travel and expenses—where demand and budget were already clear.

Navan repeatedly rebuilt core systems—travel UX, payments, expense, and support—whenever new technology (like modern ML or OpenAI) made older approaches obsolete, even if it meant discarding years of work shortly before launch.

Cohen frames every new feature as a hypothesis with defined success metrics; if it doesn’t deliver, it gets cut—unless it’s strategically core, in which case they pivot and iterate relentlessly, as they did with Navan’s rewards program until usage finally took off.

When COVID erased all travel revenue overnight, Cohen revisited his core beliefs (travel will return, T&E is critical, people-centric software wins) and used them to justify layoffs, heavy investment into expense management, continued sales efforts, and fundraising on a “travel will come back” thesis.

By using AI and now OpenAI to automate most customer support interactions, Navan has pushed gross margins to ~75% and targets ~80–85%, showing how AI can fundamentally change the unit economics of traditionally low-margin, service-heavy categories like travel management.