Factual Curiosity

Factual Curiosity turns any question into a personalized, source-backed lesson—built around how you learn best.

Comment

Inspiration

I wanted learning online to feel less like being handed a long article and more like having a thoughtful guide. People learn differently: some need analogies, some need concrete examples, and some want to go deeper without losing the thread. I built Factual Curiosity to turn a question into a lesson that adapts to those needs while keeping its sources visible and useful.

What it does

Factual Curiosity creates personalized, structured lessons from any topic. A learner chooses their time, goal, prior knowledge, focus areas, and learning preferences. The app then researches the topic, plans the lesson, and presents it in digestible chapters.

Each chapter includes four sources: Wikipedia for orientation and three additional publicly accessible sources for stronger support. Learners can also use Curiosity, an in-lesson chat, to ask for a simpler explanation, a deeper explanation, an analogy, or a real-world example without losing their place.

How I built it

I built the project with Codex, using a React frontend and a server-side research and lesson-generation pipeline. Gemini powers the lesson planning, writing, and research grounding. I added a source policy that filters out paywalled or inaccessible sources, keeps Wikipedia as a helpful starting point rather than the only authority, and attaches sources directly to each lesson chapter.

I also built persistent learning preferences, saved lessons and reading progress, responsive mobile layouts, and a dedicated Curiosity chat experience.

Challenges

The hardest part was making the research feel genuinely trustworthy rather than just generating an answer with a list of links. I had to handle unreliable search results, inaccessible academic-library pages, rate limits, and source variety. The solution was to separate research, planning, and writing; require inspectable public evidence; and make the lesson refuse to rely on Wikipedia alone.

Another challenge was making personalization consistent. Instead of treating analogies or examples as a one-time decoration, the lesson instructions make them recurring teaching tools wherever they help understanding.

What I learned

I learned that a useful AI learning product needs more than a strong model. It needs careful product decisions around source quality, transparency, pacing, user control, and recovery when an API or source is unavailable. Factual Curiosity is my attempt to make AI-assisted learning feel more personal, more grounded, and more trustworthy.

Creator

Made by Arman Ohanyan, a 14-year-old student from Nova Scotia, and officially entered by his parent on his behalf. This project is currently submitted using the parent's account.

Built With

Share this project:

Updates