Sanofi × Area 23 — AI-Generated Experiential Content · 2024
Smoldering AI
Led a five-person team to build a 1,000-image library for an interactive ECTRIMS conference booth, where doctors' reflections on patients' fading memories were visualized dissolving into smoke. Silver, 2025 MM+M Awards.
Challenge
Sanofi and Area 23 needed a booth experience for the ECTRIMS conference built around a kiosk: doctors would reflect on memories their patients had lost to their condition, and each memory would be visualized on screen as it dissolved into smoke. That meant sourcing roughly 1,000 unique images that could plausibly represent almost any memory a doctor typed in, while staying visually and emotionally consistent as one aesthetic, a volume and range no traditional photo or illustration pipeline could hit on a conference timeline.
Approach
I led a five-person core team, with additional support pulled in from other departments, and built the production model from the ground up. I trained a custom AI model tuned to the smoky, dissolving look the concept demanded, then developed a library of prompt templates that could reliably reproduce that aesthetic across an enormous range of subject matter, from specific objects to abstract, memory-like forms. To run production at scale, I built a tracking system (a live catalog of thumbnails and keywords) and broke the full image list into daily assignments across the team, benchmarking a target of 50 finished images per person per day based on my own early testing. We generated roughly 3,500 images and hand-curated the strongest 1,000, retouching by hand wherever the AI output needed a final push.
Impact
The finished library powered the Sanofi booth experience at ECTRIMS and won Silver at the 2025 MM+M Awards. Beyond the honor, it proved out a production model (custom-trained AI plus a disciplined pipeline) that could deliver image libraries at a volume and speed no traditional process could match, without losing the emotional register the creative concept depended on.