'The best interface is no interface', is an idea that feels perfect in theory, but near impossible in practice. Coming on ten years since the book by the same name was published, we are still clicking and tapping away, entering our data over and over again.
Maybe that is all about to change?
Last night I had dinner with some brilliant product people in San Fransisco where we debated what the future of product experiences is going to look like. The conversation was driven by the lunch of Apple's Vision Pro, Humane's demo video and AI/personal assistant tech.
The idea being pitched was that with AI, etc. my computers should 'know' lots of things about me. If we take booking a flight as the example, we should expect that the computer will know where I like to sit, my passport, my credit card details, etc. and I should be able to say:
"Book me a flight from SF to NYC on Wednesday next week leaving after 10am."
And then it's done. I get presented the finished task and I go on with my life.
This is as close to the holy grail of no interface as I can think. It follows the kind of design pattern that Google moved too years ago where they assume by default that user knows what they are doing or asking for and actions instructions giving you easy edit and undo options after the fact.
The example here is in Gmail where you delete an email and you have the option to undo, but they no longer ask you if you meant to press delete.
The difference between the Gmail and flight booking example is the underlying technology. Google own the whole thing so it's easier for them to automate actions.
If I ask my operating system to book me a flight, today it is going to have to speak to Sky Scanner and/or different airlines booking systems. That's probably doable. But what is more difficult is the undo/edit functionality. Editing a flight is as difficult as editing a Tweet!
Yesterday I wanted to change a Thrifty car rental pick up time from 10:00 am to 8:00 am. I'd made the reservation with Booking.com. I thought it would be an easy change of a number in a database, but I had to go through a complex series of screens and in the end I got sent a new booking number but apparently my old booking number might still be valid so I have to bring both (?!) to the airport.
And we are back at the classic Desirability vs. Feasibility battle.
We know what we want the end state to look like, but is it something that we can actually build? And does AI help us get there?
I'd very much like everything to feel like it felt to get out of an Uber for the first time and not paying. But when we're reliant on decades of underlying, complex, legacy software, we may have to wait a while before everything feels like that.
