The key features: - Question-answer-explain format that completes the full loop from curiosity to understanding in ~10 sec each - A recommendation system that doesn't just serve existing questions, but also generates new ones based on where your interest flows - Questions generated from your curiosity get shared with others too, so we're actually capable to build toward a million whys

Although the question-answering format may remind you of similar products, what we aim for is different: - Unlike trivia, questions are designed so that even if you don't know the answer, it should feel solvable with a little more thinking. The process of answering is the learning experience. - Unlike language learning apps, you don't need to remember them afterwards or get them right next time. It's about building intuition, not drilling recall.

When I was doing my geophysics PhD, I found myself taking classes in all kinds of other fields — aerospace, AI, materials science, finance, etc. It was genuinely relaxing for me, a way to unwind from my research. And I kept thinking: why can't learning always feel both relaxing and educational? The more we explored this, the more we realized the biggest barrier to satisfying curiosity isn't personalization or fancy multimedia. It's the attention. If learning and getting feedback can be packed into ~10 sec per question — about the same time you'd spend deciding whether to scroll past a TikTok video — then curiosity may finally fit into the cracks of your day.

We're still far from our goal of building a truly scalable curiosity experience, but I've already found myself learning useful things I never would have bothered googling: why Helvetica is so prevalent, what different clothing materials actually mean, best practices for cooking steak. This kind of knowledge has a compounding effect — not because any single fact is powerful (although sometimes they are), but because together they build intuition for understanding basically anything. It's a resurrection of the spirit behind 《十万个为什么》(A Hundred Thousand Whys), a famous book that sparked curiosity for generations in China, but was never designed to scale in the questions.

One of the biggest technical challenges is the recommendation system. We use LLM in the background to dynamically generate "topic syllabi". It looks at your answer history, forms a theme, and creates short question arcs around that theme. That's when we decide whether to generate fresh questions or pull from our existing bank. Sometimes the AI gets it wrong (bad phrasing, confusing images, etc.), and when users flag those, we review and fix them.

It's free and signup isn't required to experience the main flow. Would love your feedback on how it feels and what's missing. Thanks!