The Deal AI Can Never Find

From the Travel Tech Podcast newsletter for episode 32, The Deal AI Can Never Find.

In 1990, Mark Cuban sold his first company for $6 million, went out with his friends, and while still drunk, rang American Airlines to ask whether they sold lifetime flight passes. They did. He was hungover enough the next morning to sign the paperwork, and upgraded to near unlimited first class for himself and a companion for life.

I'm telling you that story because it is the cleanest illustration I have found of what this week's guest builds for a living, minus the hangover.

Iñaki Uriz is co-founder and CEO of Caravelo, the platform behind airline subscription products at Wizz Air, JetSmart and others. Before COVID the company was testing several ways to help airlines retail better. Then the pandemic grounded its one subscription customer, and the subscription revenue kept arriving anyway. An airline making money with no aircraft in the air was reason enough to bet the company on it. Apple Podcasts, also streaming on Spotify and YouTube.

What changes once a customer commits

What I took from this conversation is how hard it is to predict what people do after they have paid up front. Caravelo's data shows patterns nobody would have put in a business case.

Subscribers use 70 to 80% of the flights they pay for. Nobody is engineering that gap, but the unused credits are revenue with no cost attached.

They fly more than they did before, for the same reason people watch more television after Netflix. That is new demand, not shifted demand.

They stop shopping. A subscriber does not compare fares because the decision is already made, so the airline takes the whole customer off the market, not just one booking.

On the last minute pass, some people pay for a year, never fly, and renew. Iñaki's words: don't ask me why.

Put together, for every $100 an airline takes in subscription revenue, $30 to $50 is money it would not otherwise have seen, and Caravelo sees that consistently across customers.

Monthly churn runs 3 to 4%. One reading is losing a third of subscribers a year. The other is keeping two thirds for a second year without spending anything to win them back.

Then there is the second wallet effect. A subscriber paid at the start of the month, so at booking time the wallet feels full and a $50 seat upgrade looks like $50 rather than $200. Ancillary attachment goes up. On the last minute pass it does not, because someone deciding tonight to fly out of Santiago tomorrow travels light.

The deal AI can't find

The second thread is where Iñaki thinks this goes. AI agents will soon run flawless, policy compliant fare comparisons across every airline. He calls it ruthless. Everything in the public market gets commoditised. A subscription sits outside that. There is no search for the agent to run, because the customer already chose. And a private deal an airline cuts with a company in exchange for subscribing is not something an agent can find.

That is why corporate is Caravelo's next move for 2027. Today a Bank of America sits down with Delta once a year and negotiates a bespoke deal. A 300 person architecture practice with sites around the country never gets that meeting. Prepackaged subscription tiers give the long tail the same trade: better price and flexibility in exchange for exclusivity, and a CFO who knows the travel spend before the year starts.

One more reason to listen. Iñaki's reflection on running a company through three years of AI: execution is no longer the scarce thing. He used to sketch a slide on paper and hand it to a designer. Now he photographs the sketch and hands it to Claude. What remains is deciding what to build and why.

Iñaki will be at World Aviation Festival in Lisbon, and so will I, in the startup village. Come and say hello.

Fly safe (or don't fly at all, if the subscription maths works out),
Alex Brooker, Host of the Travel Tech Podcast, Founder of Airside Labs