AI, Apps & All-Stars: Powering Athlete-Brand Connections at Cannes

AI That Anticipates User Needs

The app's cornerstone was an AI Assistant that could answer a myriad of questions about the event (from how to access the beach to event schedules to best local restaurant picks) and tailor event recommendations to each user based on their interests. Using Stagwell's Enterprise ChatGPT as a backbone large language model, we fine-tuned our AI on a large, static dataset about Cannes and Sport Beach. We leveraged Sanity, our CMS for the event, to constantly update changing events or critical event information.

While the AI Assistant interfaced with the CMS to update its context, it also interfaced with an event recommendation engine developed in-house and hosted on Oracle Cloud Infrastructure, taking advantage of Code and Theory's partnership with Oracle. The AI Assistant "got to meet" each user during a conversational onboarding interview, and saved those learnings to suggest customized itineraries, ensuring attendees wouldn't miss relevant content. As that content shifted (speaker changes and event schedules moving), the recommendation engine and the AI Assistant could rely on their connection to the CMS to ensure their responses were explicit, accurate, and based on real-time feedback.

Powering Rapid Innovation

Code and Theory leveraged Sanity's Content Cloud to manage all event content, including sessions, speakers, event details and more. The combination of Sanity as a content provider and Vercel as a hosting platform enabled Code and Theory to rapidly build a recommendation engine and AI chat, significantly accelerating development and reducing delivery time. This approach allowed real-time data integration and scalable solutions, feeding first—and third-party data dynamically to the AI chat system.