* Many of the products I worked on at Element AI are confidential. So I chose to disclosed the important, as a designer, the process.

Innovation Discovery Process

For the Mur.AI project I used an inspiration of the Design Sprint process from Google Venture. Using an AI style transfer algorithm, the initial request was to build a photobooth experience for the 2017 C2 Montreal exhibition that would happen in three months. So we decided to use rapid iteration process to make sure we would create a delightful experience. DS allowed us to build, test and learn every week!

Week one – we met and envisioned the users, evaluated the needs and share our vision of the end product. I had brought LEGO in order to visualized the space and the flow. Of course, some ideas were ditch right away.

At the end of week one had build the photobooth enclosed experience using wheeled whiteboard, a tablet to take the picture that would be stylized by the AI – first version would take 10-15 seconds to calculate) and displayed in on a TV screen outside the booth. We tested the experience at the Friday Fusion (beer time) with the entire company (~40 people).

Week 2 – we started by reviewing the feedback from the last demo. Nothing was really negative nor thought it was the greatest thing they had seen… and people weren’t particularly thrilled to be in the enclosure.

Someone had seen a live video stream of a stylized webcam feed, so we opened up in “diverge” mode to imagine a different experience. What if we could stylized the crowd live?!

End of week 2 we had a great demo with a live stylized webcam feed (with low performance). We realized that not only people were not intimidated to go in the box but could stay in front, with the crowd, and it would stimulate all kind of discussions.

Week 3 – we build a second version with a live feed and a style selector touchscreen. People would get in front and try different styling and take pictures with their phone!! That’s exactly the feature we were missing to generate the hype we were looking for :

People would be able to take a picture that would be shared on Twitter @ElementAIArt… Success??

See people’s reactions for yourself! There was 700 post in the first year!

That’s excluding the party pics 😛

C2 Montreal installation

Built by Robocut Studio

OK, the technology was kind-of-working but for C2 the hardware setup needed to be more robust and the structure more pretty! So I consulted with the execs to see what they like, and started to design 3d models while the dev were still optimizing (improving 20 ms here and 30 ms there).

I had nailed a few requirements:

  • Low latency, very large TV and above the crowd
  • Camera in the middle and at eye height (5′ 6″)
  • Ergonomic touch screen for standing position

I also conducted a Pre-Mortem workshop to make the installation bullet-proof (almost).

“JP led the design of the Mur.AI project with the Design Spring method from GV, making Mur.AI our first and very popular demo, still in high demand today.”

– Gabriel Duford, SVP Development & Technology and Co-founder at Element AI

The Mur.AI installation with a more portable form factor, still travels with the Event team today.