Todd Sullivan built a travel-support business by hand, grew it to eight figures, and learned exactly what the job costs. With Humoniq, he is handing that job to AI - and betting the whole industry follows.
Todd Sullivan has spent most of his working life on one stubborn problem: what happens after a traveler hits a snag. A missed connection. A schedule change. A refund that should be simple and never is. Most founders would call that a niche. He has treated it like a career.
Today he is the CEO and co-founder of Humoniq, a company he started in 2025 and brought through Y Combinator's Summer batch. The pitch is direct: airlines, airports, online travel agencies and corporate travel managers spend an enormous amount of money outsourcing customer support to call centers, and that money buys long hold times and high turnover. Humoniq wants to hand the work to voice and text AI agents that run around the clock - answering flight-change questions and baggage claims out front, processing exchanges and refunds in the back.
What makes the bet interesting is not the technology. It is that Sullivan already ran the thing he is now automating.
Before any of the companies, there was code. Sullivan trained as a software engineer in Australia and did engineering work in the defense sector. Then he took a turn most engineers never do: he became an investor, helping deploy roughly $55 million into Australian businesses at a private equity firm. He studied at the Australian National University, stacking degrees in software engineering, finance and management.
That combination - the engineer who can also read a balance sheet - shows up in how he talks about companies. He does not describe products in isolation. He describes markets, costs, and where value is hiding in plain sight.
His first company was Globetrooper, built in 2010 with Lauren McLeod. The idea was social: help travel-minded people find each other and plan real trips together. It launched more than a thousand journeys before they sold it in October 2011. It was small, fast, and it taught him the shape of the market he would keep circling.
Then came Flightfox in 2012 - the company that put him on the map. At a moment when everyone assumed flight search was a solved, crowded, algorithm-owned market, Sullivan pitched the opposite. Instead of a better search engine, he built a marketplace of human experts who competed to find you cheaper, smarter itineraries for a finder's fee.
It worked in a way that surprised the skeptics. In its first eight weeks of beta, Sullivan said, Flightfox pulled in more than 5,000 users and $30,000 in finder's fees. Experts searched everywhere the big engines did not bother to look - budget carriers, special offers, newsletters, frequent-flyer programs. On an average multi-city international trip, he said, they saved customers around 27%. On complicated premium-class trips, the savings sometimes ran past $5,000.
The company went through Startmate, Australia's flagship accelerator, before landing in Y Combinator's Summer 2012 batch and moving to the United States. Sullivan has been blunt about how much that early choice mattered.
Over the years, Flightfox grew up. The crowdsourced-competition model gave way in 2014 to a consultative, one-on-one service, and the customer base shifted from bargain-hunting individuals to organizations with genuinely hard travel - universities, humanitarian groups, companies moving people through complicated itineraries. Along the way it scaled to roughly $22 million in annual revenue and supported well over a hundred thousand travelers. Not every trip was ordinary. One customer, memorably, flew with three cats.
Here is the part worth sitting with. The whole premise of Flightfox was that humans could out-think the software. Thirteen years later, Sullivan started a company built on the opposite conviction: that AI is finally good enough to carry the work that used to require rooms full of people. He did not change industries. He changed his mind about the tool.
Humoniq's framing is that AI belongs on the routine, high-volume, informational work - the questions that clog a queue - while human expertise stays reserved for the genuinely complex cases. The goal Sullivan describes is to flip support from a pure cost center into something that can actually generate revenue, with deep integrations into the systems airlines and agencies already run so the agents can do real work rather than just deflect calls.
He did not build the new team from scratch, either. Humoniq reunited the core people he had already spent years in the trenches with at Flightfox, including co-founder Will Weber - a former professional soccer player in Sweden's second division who spent eight-plus years leading Flightfox's product and growth. Starting a company with people who have already survived one with you is its own kind of moat.
Ask him what he would have done differently across all of it and you get a very Sullivan answer - unromantic, earned, slightly funny in its flatness.
It is the sentence of someone who has stopped chasing novelty. The thing most founders sell - a brand new idea in a brand new space - is precisely what he has walked away from. His edge is depth. He knows what a travel call center costs because he ran one. He knows which promises are easy to make and impossible to keep. He knows the difference between a demo that dazzles and a system that survives contact with a stranded passenger at 2 a.m.
Whether Humoniq becomes the standard for how airlines and agencies handle support is an open question, and a hard one. But if you want to understand the bet, you do not start with the AI. You start with the fifteen years Sullivan spent learning, in painful detail, exactly what the job requires - and deciding to build the machine anyway.