The problem
When you learn with AI, one answer raises five new questions. In a linear chat they all pile into the same context, and managing the conversation becomes harder than learning the topic.
18 of the 20 people we interviewed had felt it. 82 of 100 we observed were already using AI to study.
What we built
Every question and answer is a node in a tree. Follow-ups branch off into their own context, so you can go deep on one path without polluting another, then come back and retrace how you got there.
We shipped it on the web and as an Electron widget you call up with a shortcut over any screen. It was the first product we built as SLIT.
By the numbers
- 483
- Commits in one month
- 18 / 20
- Interviewees had the problem
- D3
- Tree visualization
Stack
- React 18
- D3.js
- Electron
- Supabase
- OpenAI