LearnFork

LLM frontend for learning and researching

Visit Website
March 10, 2026 I made an LLM frontend for learning and researching various topics using AI

The problem:

When researching or learning about a topic using ChatGPT or Claude, and when there’s a part I don’t fully understand, I usually need to ask a follow-up question. Because the chat is linear, the context quickly gets messy or lost, and I end up repeating things like a parrot to continue the conversation about the main topic.

ChatGPT does have a branching option, but the UI isn’t great. So I decided to create something like this, mostly for myself. Since I think the tool is interesting and has some potential, I’m sharing it with you and would love to hear your feedback.

The solution:

The main thing is visual branching on a canvas. In practice, vertical messages in the same chat can be used to go deeper into a topic, while horizontal branching can be used to clarify concepts you don’t fully understand.

Canvas with multiple AI chat boxes

Context is preserved both vertically and horizontally. That means the original chat about Python won’t know that, in one branch, you asked about the speed difference between Java and Python.

The concept is BYOK (bring your own key), which means you need to create an API key with one of the supported AI providers for the app to work. The whole app is frontend-only. There are no servers, and everything is stored locally in IndexedDB and localStorage. Some providers have a free tier, so you can try it out

https://learnfork.com/docs/providers

Comment

About

LearnFork exists to solve the problem of learning or researching using linear AI chats. With infinite branching on canvas, you can explore any topic without losing context.