The world is moving steadily towards AI driven design, removing people from the equation. We decided to challenge this to show how computers and AI will evolve and rather than AI being a threat to design, it can create new tools that help human designers exponentially increase their output and precisions. Our own R&D showed the potential for AI + Humans to produce amazing artwork through the use of semantic maps and visual guidelines. The inherent design language of a recognised and iconic design brand like Braun was the perfect opportunity to explore if AI could be “taught” an aesthetic in order to assist a designer in working in that style. With the help and Adobe's support we built and 'taught' an AI extension for Illustrator the rules of the Braun design system in order to make that design language accessible and intuitive for human designers.

Designers can look at design and know it’s braun “Braun" but computers will look at the same and see only shape and compositions. We are years away from computers that think creatively for themselves but with the right guidance we discovered deep learning algorithms can be trained to mimic a broad set of design rules while imagining new shapes and novel compositions. Braun's minimalist style, so transformative for industrial design over the last century, was an ideal candidate to teach a computer a language.

Our intent? We wanted to teach our AI how to speak the very nuanced and precise Braun language - to understand the specifics of Braun form and composition and aesthetics that a human designer may take years to master. The ambition is to create an extension to the Adobe Suite that empowers designers around the world to do what they do best, design.

Our intent? We wanted to teach our AI how to speak the very nuanced and precise Braun language - to understand the specifics of Braun form and composition and aesthetics that a human designer may take years to master. The ambition is to create an extension to the Adobe Suite that empowers designers around the world to do what they do best, design - not to create a plethora of formats while struggling to maintain visual guidelines.

Braun's vast catalogue of work was refined down to core geometric shapes and compositions. This created a language encompassing the totality of the Braun design ethos. A generative adversarial network - two neural networks competing - were trained on this language and a series of tests were run to create novel shapes. The stunning success of this initial run led to a more sophisticated approach designed to create new layouts in separate stages, generating millions of new elements and compositions. An extension was developed for Illustrator to read any layout, temporarily converting it to a simplified format, and match it with appropriate suggestions from DesignLanguage.ai. A designer could then apply the suggestion to their page if they preferred it. With Adobe's support we are now moving into an executional phase of the process using AI and traditional computer vision techniques to filter the new elements being created and in informing the design of the initial language itself.

Following robust testing across a multitude of experiments, formats, inputs and outputs the expectation is to roll the Adobe extension out to Designers around the world as well as Braun's network of designers as another tool in their Adobe quiver of creative assets. This is expected to result in an increased efficiency in production and development of campaign assets - freeing the Designers time to focus back up the chain on concepts and work that adds real value to the Brand and its bottom line. In time, the AI will be taught new languages for different brands, with a plan that in future, all Brands will have their own bespoke language included within its lexicon creating efficiencies at scale, and providing designers with a powerful new set of tools.