Thirty-one years in travel: what software still misunderstands

Most travel software is built by people who have never chased a supplier for a confirmation in peak season. It shows.

Software keeps modeling travel as inventory: rooms, seats, slots, prices. But a trip is not inventory. A trip is judgment. The itinerary that is correct on paper and wrong for a 62-year-old couple. The "similar hotel" that is not similar at all. The day plan that looks efficient until someone who has actually driven that road laughs at it.

Two from memory. My first travel job, in the nineties: we booked a family on the Delhi-Agra circuit. Entry to the Taj Mahal used to be free on Fridays; that week, they made Friday closed. No internet, no way to know, and the client had no option but to skip the Taj. The itinerary was correct on paper.

Years later, at one of India's largest OTAs: a Japanese client working out of Gurgaon booked a hotel in Mauritius. The inventory came from an API with no market-specific rates, and the system happily booked him at Indian-market rates. It took a long, awkward effort to convince the hotelier. Technically right, practically wrong.

What would I have done better in those cases? I still wonder.

And the industry runs on relationships more than on systems. A DMC answers your WhatsApp at 11 pm not because an SLA says so, but because you have worked together for nine years. No database field captures that, and yet it decides whether the honeymooner gets the sea-view room.

I do not say this as a complaint about technologists. I say it as the opportunity. For the first time, the tools have gotten good enough that the operator does not need to become a programmer to build the software. AI does the typing. The operator brings the thirty-one years.

The next genuinely good travel software will be built by people who have run trips, with machines doing the parts machines are good at. That is the bet NAMA is. Ask me in a year how it went.