Ask a question
Ask about a story, a subject, or something you’re going through. It searches short summaries of every conference talk and the stories inside them, then links you to the full talk.
Good at stories, subjects and speakers. It doesn’t know exact quotes, because we don’t copy talk text onto this site.
How does this work?
Every conference talk got a short summary, written with AI, and so did every story inside them. That’s about 22,000 little descriptions.
A computer can’t tell that “forgiving my dad” and “reconciling with a parent” mean nearly the same thing. So we turn each description into a list of 384 numbers. Think of it as an address on a giant map of ideas, where things that mean the same thing live close together. Stories about forgiveness end up in one neighborhood. Stories about pioneers crossing the plains end up in another.
When you ask a question, it gets turned into its own 384 numbers, which drops a pin on that same map. Then we look for the summaries living closest to your pin. That’s the whole trick.
The nerdy name for this is retrieval, the R in RAG. RAG is how AI chat tools look things up before they answer. We built the looking-up part and stopped there. No AI writes an answer for you, so nothing gets made up. You get the talks, our one-line summary of each, and a link to read the real thing.
The search runs on your phone or computer. The first time you ask, it downloads a small model and the map, about 30 MB, so maybe use Wi‑Fi for that first one. After that it’s quick. There’s no server on our end reading your question.