How The Met Uses OpenAI to Bring Historical Fashion Exhibits to Life
The Metropolitan Museum of Art partnered with OpenAI to create a conversational AI experience called "Chat with Natalie" for its Sleeping Beauties fashion exhibition, letting visitors interact with a historically accurate AI portrayal of a 1930s New York socialite whose wedding dress is on display.
Challenge
Museum exhibits traditionally offer a passive viewing experience, limiting visitor engagement with historical artifacts and the stories behind them. The Met's Costume Institute sought a more immersive way for visitors to connect with the garments and figures featured in its "Sleeping Beauties: Reawakening Fashion" exhibition.
Solution
OpenAI partnered with The Met's digital team, historians, and curators to build a custom conversational AI experience called "Chat with Natalie." The system was trained on a curated dataset of primary sources — including personal letters, newspaper clippings, and historical documents — related to Natalie Potter, an early 20th-century New York socialite. Using OpenAI's most advanced language model with custom instructions, the AI authentically portrays Natalie's voice, personality, and era, allowing visitors to ask questions about her life, her 1931 wedding dress on display, and the world she lived in. The experience incorporates the same safety guardrails used in ChatGPT.
Tools & Technologies
Sign up to read complete case studies, access detailed metrics, and unlock all use cases.
Full Story
The Metropolitan Museum of Art's Costume Institute has long been celebrated for its visually stunning exhibitions, but engagement has historically been one-directional — visitors observe artifacts without direct interaction. For the landmark "Sleeping Beauties: Reawakening Fashion" exhibit, the museum wanted to push beyond static displays and invite visitors into a genuine dialogue with history.\n\nTo achieve this, The Met partnered with OpenAI to develop a first-of-its-kind interactive AI experience centered on Natalie Potter, a prominent New York socialite from the early 1900s whose 1931 wedding dress is featured in the exhibition. The challenge was to create an AI persona that felt historically grounded and respectful of the source material, rather than a generic chatbot.\n\nOpenAI worked closely with the museum's digital team, curators, and historians to assemble a rich dataset of primary sources — letters, period newspaper articles, and archival documents — all tied to Natalie Potter's life. This material was used to fine-tune a custom version of OpenAI's most capable language model with tailored instructions, enabling the AI to respond in a manner consistent with Natalie's character and her historical context.\n\nThe resulting "Chat with Natalie" experience gave exhibition visitors an active role in exploring the story behind the dress. Rather than reading a placard, guests could ask Natalie directly about her wedding, her social world, and the fashions of her time. The experience was built with the same content safety mechanisms as ChatGPT, ensuring interactions remained appropriate for all audiences.\n\nThe collaboration represents a broader vision for how AI can deepen cultural and educational experiences. By transforming a museum exhibit into a living conversation, The Met and OpenAI demonstrated that AI can serve as a bridge between historical artifacts and modern audiences — making the past more accessible, engaging, and personal.