The Best Interface is No Interface

Many believe the future of design is in screens. They’re wrong. Our love for the digital interface is out of control. This conversation, led by Golden Krishna, will explore a better path: NoUI. Eliminating counterintuitive input mechanisms for natural inputs inverts the contemporary focus of software design to have computers adapt for people, rather than people adapt for computers. The results can have a profound impact on your design process and our lives.

There’s so much to celebrate about technology, but there’s a terrible trend in tech today… we’re riding a wave that’s taking us away from solving peoples’ problems.

When we created computers, we made people use the CLI. But then we got the GUI. Then we got touch UI.

So now… how do you design a better (car|vending machine|toilet)? You whack an interface on it. Use a whole touchscreen device where you used to have a button.

We are surrounded by screens. We used to be surrounded by paper and dreaming of a paperless world. We should dream of a screenless world.


  1. Embrace typical processes instead of screens
    • is it really better to use a smartphone to open your car? Key vs. BMW app. A long process of opening up a phone, closing other apps, finding the right app, using a weird UI… (see slides for the full walkthrough). The real goal was the physically present human opening the door – which you can do in two steps with a physical key.
    • Meanwhile Mercedes did it differently: if your key was nearby, the car door would simply open. This also meant you couldn’t lock the keys in the car.
    • Ford realised people carrying things to the boot of a car have their hands full, so they put in a foot/kick activation to open the boot.
  2. Leverage computers instead of catering to them
    • We memorise passwords and do things the computer requires. But we do that because it’s how we treat computers, how we expect them to behave.
    • (great example of a headlamp by Petzl, which senses the difference between looking ahead and looking down at a map, when you need the map to be dimmed. No extra interface added.)
  3. Create a system that adapts to individuals
    • Nest initially looks like a device with a UI slapped on it, but it learns – so after a while you don’t need to interact with it at all.
    • Normal hospital machines are reactive – they report what has already happened. EarlySense detects normal data like heartbeat etc; but it learns your patterns and identifies changes in patterns well before things become a major problem.

There has been a lot of interesting discourse about No UI since it was published.

We are used to lots of automated things – doors, airbags, the auto transmission. It’s impressive that people accept auto transmissions, a very dangerous time and place for automation.

Some advanced No UI devices will have a standard UI as a backup.

What about the web? Services like Tripit and Pandora are trying to learn enough that you don’t need to interact with it for it to work.

How do you really do this?

Ran an exercise with a group of people who hadn’t met each other before:

  1. Embrance typical processes instead of screens → observe the job the person is doing
  2. Leverage computers instead of catering to them → work out what sensors and data sources are available
  3. Create a system that adapts to individuals

Some results:

  • One kind of crazy idea was to use a scent in the room to remind people to take their medication. Unusual, but probably still better than a UI on a computer.
  • Running shoes wear out when the shoe compresses. There are various trackers that attempt to work out if your shoes are worn out. Someone thought of having dye in the sole that turns them red when they are too worn.

You can download the worksheet from

There’s so much to celebrate about technology, but there’s a terrible trend in tech today… hopefully this has given some ideas on how to do something better.


Q: Once you move into automation – how do you handle responsibility, accountability for failures and problems? There are cases of cruise control malfunctioning and making cars accelerate, etc.

A: You have to be transparent when you are creating new technology… a lot of this relies on data tracking and understanding the users. You have to be transparent about what you are tracking, how you are doing it and give them a way to shut it off. New technology can always be scary if used the wrong way.

Q: can you mention any of the work you do at Samsung?

A: …uh, no. 🙂