github_copilot_thumbnail.png

GitHub Copilot

Test

GitHub Copilot

GitHub Copilot –

Project

Interactive Game

Role

UX/UI Designer

Agency

Left Field Labs

Timeline

3.5 Weeks

 
 

GitHub came to Left Field Labs to design a demo for their booth at the 2024 Microsoft Build Conference in San Francisco.

 

The
Problem

Design an interactive game to showcase the power of GitHub Copilot, their AI programming assistant. Make it fun, fast, engaging, and relatively easy to use, but also utilize an actual coding interface.


 
 
 

The Process

 

Concept – Use code to create an image that matches a source image – most similar image takes over that tile on the gameboard.

An interactive coding game where users competed on teams to take over as many tiles as possible on the game board. In order to take over a tile, the user needed to use Copilot to help them code an image that was as similar as possible to a source image. Their image would then be analyzed with AI and given a similarity percentage to the source image. If they got the high score for that tile, they claimed it for their team.

 

Low-fidelity Wireframes

 

Insights, Revisions, & Concept Development

Big Screen User Flow

Kiosk Screen User Flow

  • Fleshed out the specifics of the login flow

  • Added a team assignment screen

  • Added a selected and confirmation state for tile selection phase

  • Added links for “Instructions” and “Prompt Hints” to the gameplay screen

  • Developed the game board grid and “Homebase” locations

  • Added a “Pending” state for tiles that are being played by other players

  • Added an animation flow to show the new submissions being scored against the source image

  • Added a “Swag Progress” screen for a potential swag giveaway concept

 

High-fidelity Wireframes

 

UI – Brand Elements

 

UI – Component Library

 

UI Screens

 

Kiosk Gameplay

The user enters a username and is assigned a team color. From there, they are taken through a brief set of instructions, and then begin playing the game, where they use Copilot to generate code for an image that matches the source image. After they did their best to match the source image, their results would be sent to the big screen to be scored. If the user got the highest similarity percentage for that game tile, they would then claim that tile for their team and score a point.

 

Big Screen Submission ANimation

 

Results

5:08

Average session length

608

User image submissions
(during the 3 day event)

79%

Session completion rate

141

Tiles taken over
by another team

 

Credits

Eric Lee, CTO/Project Lead
Piper Montessori, Producer
Rocky Riccardi, Creative Director
Kris Horowitz, Senior Game Designer
Jered Cuenco, Senior Creative Technologist

 
 

Next Project →

Next Project →