Scaling content design with AI
Decathlon
As the content design practice evolved at Decathlon, it became difficult for every product team to receive CD support. We built a gem for our internal teams that not only generated content but also provided guidance, evaluated options and managed governance.
Challenge
With the content design team adding more value, more teams wanted CD support but we had limited resources. As a team we wanted the content designers to be on strategic objectives rather than working on daily projects. The challenge was to create a solution for the vast scope of Decathlon while staying true to our tone, voice, and principles.
Team
Product design, research, product managers, engineering and localisation
What I did
Identified opportunities for AI to support content design
Recognised the types of content that could be be supported by and AI assistant by analysing recurring requests from product designers and managers.
Designed the AI assistant experience
Developed the structure, prompts, and guidance that powered the gem. The assistant was designed to ask for more context to product more relevant output.
Created prompt architecture, scenarios and context prompts
Worked on sample prompts for product managers and designers to use so they could just fill in details and give context about the project. Created scenarios and embedded our identity ethos in the gem
Collaboration and iteration across multiple teams promoting responsible AI usage
Worked with product teams to create personalised gems for teams that did not share same repositories or areas of work with human in the loop as our key guiding principle.
Work
Outcome
We have multiple teams at Decathlon using the gems daily to create new content, help brainstorm new options or make a decision between options that fit our tone and voice. The reduced dependency on content designers has freed time for us to focus on bigger challenges.
What I learned
We have entered a brave new world. Content design and AI seems to be a great partnership that can evolve well in the right conditions and shape the adoption of AI in organisations. It will be an iterative process to achieve the right levels in terms of outputs.
