Ex-Google Flights and Kayak execs raise $1.9M to fix what AI agents are breaking in airline search

Zerolook has raised 1.5 million Swiss francs, or about $1.9 million, to attack a problem that AI agents are making worse: flight search is getting too expensive to run at scale.

The Zug-based startup does not try to answer every airfare query in real time. Instead, it uses machine learning trained on past fare, schedule and availability data to predict flight options and prices, then attaches a confidence score. That is a big shift from the live-results model used by the likes of Amadeus, Duffel and Skyscanner’s partner API.

The pressure comes from volume. OAG says searches per ticket sold have climbed from roughly 100 to 200 in the early 2000s to 10,000 to 20,000 today, and could reach 2 million by 2030 if AI agents keep multiplying searches. Ask for the cheapest Greek island in June and one request can explode into thousands of island, date and routing combinations. Airlines and travel agencies pay for those queries even when nobody books.

Zerolook’s pitch is simple: cut the computing bill without forcing the industry to rip out its current systems overnight. Simone Lini, who previously sold Waynaut to Lastminute.com and worked on Google Flights partnerships in EMEA, leads the company with George Hadjiyiannis, whose background includes MIT, Kayak, Google, Nvidia and eBay. Hadjiyiannis says the technical problem is a legacy one that has been around since the era of the 8-track tape. Slightly dramatic, sure, but the bill is real.

Playfair led the round, with Vento Ventures, TrueSight Ventures, Alpha Venture and angel investors also backing it. The cash will go toward Zerolook’s first API, a machine learning team in Zurich and pilot programs with online travel partners that have signed letters of intent. The open question is still the only one that matters: will the predictions be good enough for airlines, or just a useful patch until the bigger systems catch up?