Tell an AI you must reach New York by this evening and it books the whole trip
The assistant remembers your airline, seat, food, card, and favorite hotel, then lines up flights, cars, and calendar.
Invest Like the Best with Patrick O'Shaughnessy · Noah Shinn - Building Instinct: The Personal Agent - [Invest Like the Best, EP.493]Noah Shinn, founder of the AI personal assistant Instinct, says a traveler can record a short voice message asking to reach New York that evening and everything else is solved. He describes that kind of booking through Instinct as an even better experience than the travel site.
Older travel sites unified flights, hotels, and other services from many airlines and hotel chains into one presentation, where a person could click an option, check out immediately, and see it in email. In Shinn's example, the traveler does not reenter those details because the assistant applies what it already knows.
In his example, the work starts with location. The assistant sees the starting point, say San Francisco, and the need to reach New York.
It then applies what it already knows about the traveler. That includes the preferred airline, the class and seat type, whether it is aisle or window, the food to be delivered, and which credit card to use. It finds flight options that fit those preferences and books one.
The same memory carries forward to future bookings. Shinn says if you share a single preference about style or where you like to stay, it uses that memory to fit the next booking to what you want, and extrapolates that general taste to other bookings. As an example, he says, maybe you have stayed at the same hotel the last seven times, so selecting it is easy.
It also lines up the rest around the calendar, the cars to and from the airport and the events for the evening.
Shinn says Instinct, still invite-only with a very small user base, is approaching more than $1 billion a year in transaction volume, with 50 percent from travel alone. His question is what happens to businesses built on providing a standardized interface when a much easier standardized interface arrives.
Questions people ask
How does it learn my preferences in the first place?
You give it access to location, calendar, email, and payment methods, and it stores stated likes and past choices. Shinn says it uses that memory for future bookings and extrapolates that general taste to other things you might want to book.
What data do I have to hand over for this to work?
Shinn says users stay in control of their data and can take it back. In practice that means sharing a credit card, passwords, calendar, and location, with the tradeoff that more access makes the assistant more useful and more proactive.