Scaling content design with AI
Decathlon
As content design matured at Decathlon, demand for support grew faster than the team could scale. We designed an AI assistant that extended content design expertise by generating content, providing guidance, and reinforcing governance.
Challenge
Growing demand meant content designers were spending more time supporting day-to-day requests than strategic initiatives. We needed a scalable solution that enabled product teams while maintaining Decathlon’s tone, principles, and quality standards.
Team
Product design, research, product managers, engineering and localisation
Approach
Identified opportunities for AI enablement
Analysed recurring content design requests to identify where AI could responsibly extend content design support without replacing human expertise.
Designed the AI interaction model
Developed the assistant's model, including prompt flows, contextual questioning, and guidance patterns that improved the quality and relevance of generated content.
Created scalable prompt frameworks
Developed reusable prompt frameworks, common scenarios, and embedded content principles so teams could generate consistent, on-brand content with minimal effort.
Evaluated AI output
Tested generated content against content principles and real product scenarios, refining prompts and guidance based on quality, relevance and consistency.
Established governance for responsible AI
Partnered with product teams to tailor the assistant for different domains while establishing governance principles that ensured critical user journeys continued to involve content designers.
Work

Outcome
The assistant is now used by multiple teams across Decathlon to generate content, compare alternatives, and access content guidance without relying on a content designer for every request. We gained more capacity to focus on strategic initiatives while maintaining consistency across products.
What I learned
We have entered a brave new world. Designing AI is as much about defining boundaries as it is about creating capabilities. This project reinforced that responsible AI depends on context, governance, and thoughtful interaction design - not just better prompts.


