Making Rooms 3.0

2024

.

11

.

29

/

Log

7

This week, we at Things.inc released Rooms 3.0, with a new Actions editor, and I wanted to give some context about how this came to be.

When we first started Rooms, we debated whether to create a visual editor or let people directly write code to bring their rooms to life. We ultimately decided on code in order to ensure that the ceiling of possibilities was truly limitless. This worked well initially, as Rooms was desktop only and we had a limited number of functions.

Over time, we added wayyyy more functions and, after launching Rooms on iOS late last year, the majority of usage shifted to mobile. As it turns out, people don’t like to code on mobile — and I don’t blame them. Aside from typing being more challenging with tiny keys, the keyboard takes up nearly half the screen, which doesn’t leave room for your code and a preview.

In light of this, we considered using AI to help users code — and even had a working prototype that used ChatGPT to generate code from simple prompts.

Unfortunately, this approach had some problems:

  1. Despite lots of coaxing, the AI models we tried consistently hallucinated capabilities that simply do not exist.
  2. It’s difficult to know what is and isn’t possible, leading to people making prompts for behaviors that the platform doesn’t support.

So about a month ago, we embarked on a path to let users bring their rooms to life, utilizing all the coding functions we support under the hood, but without requiring users to write any code.

We set out with the goals of 1) minimizing the need for keyboard input and 2) showing a live preview as quickly as possible. We figured that most people would rather move sliders around until their intended movement/action felt right rather than try to figure out the meaning of and reasonable values for Amplitude, Frequency, XYZ offsets, etc.

An early sketch from Bruno Oliveira

We also took this opportunity to group functions into categories, which we decorated with emoji to ensure anyone could understand, regardless of age or language. Here’s where we started:

An early wireframe showing categories and graphic representations of actions

And here’s where we ended:

The result felt surprisingly good! Here is one of the first rooms created with Actions from my co-founder Bruno Oliveira:

Within a week, our teammate Melissa (aka viviennevioletta) created a room that demonstrated the storytelling potential of the new Actions editor:

Fast forward to today, and Rooms 3.0 is now live. You can try it on iPhone, iPad, or web. I hope you enjoy!