AI as in integral part of the design from the start
Usually, when a new tech becomes available, at first it is used as the old one was. For example, steel beams were initially used to build traditional houses, only with steel beams instead of wooden ones. It took years until architects understood their potential and started designing building impossible to do with wooden beams: skyscrapers.
To ensure we weren't building a traditional project with a little AI sprinkled on top, but a skyscraper, we worked closely with the AI team from the start of the design process.
It is not about using tech for the sake of it, but rather use the full potential of technology to solve customer problems. The scope of potential solutions grows as tech evolves.
AI personalisation from customer journey to components
There is no giant single AI, but rather a collaboration of small AI models each looking to solve an as-narrow-as-possible challenge.
At a macro level, for example, an AI can look at what product categories to show you on the homepage.
At the micro-level, other models look at what component variation to show to you: What CTA work best? are you more partial to seeing the dress worn or a pack shot? et.. Another place where a strong design system comes in handy.
A collaboration Between AI and human fashion customer knowhow
One aspect I found very interesting in this project was the collaboration between AI and human's knowledge. How they should feed off each other.
For example, if you were looking at a black shirt you, the AI would (using image recognition) show you all the similar black shirts it found – not understanding it was creating a paralysing paradox of choice.
Or how a human would have to create totally subjective data points around brands to help the AI know if you should be recommended a Diesel or A.P.C. jean if the Kitsune ones you like are sold out.
