Building a Window to History: Talking to Anne Frank

A few months ago, I set out to deepen my understanding of AI tooling by building a Retrieval-Augmented Generation (RAG) application using Cursor AI. As a technical product manager, I was curious about how far I could push conversational AI. My first project was a Seinfeld chatbot — trained on the full episode transcripts — that allowed users to chat with the show's characters in a believable, context-aware way. It was a playful experiment, but it taught me everything I needed to know about vector embeddings, prompt tuning, memory, and building user-friendly RAG pipelines. That foundation led me somewhere I hadn’t originally planned. On the eve of Holocaust Remembrance Day, I felt a deep urge to apply those same tools to something meaningful: helping young people engage with the voice of Anne Frank. I built annefrankai.com, a safe, multilingual chatbot based entirely on her original diaries, designed to let children around the world talk to "Anne" through historically grounded, emotionally intelligent dialogue. The next day, it went viral — crossing 7,000 chats per day, over 130 million tokens generated, and 20 million words spoken in 10 languages. That project is now being presented to Yad Vashem's leadership — and it all started by learning to prototype with the tools now available to all of us.

Speakers