A visual conversation tree for AI chats — see your thinking, not just your scroll.
oMyTree is a web app that turns your AI chats into a visual conversation tree.
Each message becomes a node.
Each follow-up question grows a new branch.
You can jump between branches, revisit any path, and actually see how your thinking evolves.
This is not a “knowledge tree” or curated mind map.
It’s the raw conversation, mapped into a structure that’s easier to explore than an endless chat window.
- 🌱 One question = one node
- 🌿 Follow-up = a branching path
- 🌳 Whole conversation = a living tree you can navigate
Live app: https://www.omytree.com
Start a new tree: https://www.omytree.com/app?new=1
Two years ago, when ChatGPT first launched, I was blown away – not because it could answer questions, but because it felt like we could finally talk to an AI.
But the deeper I went, the more one problem kept coming back:
Linear chats completely break once ideas get complex.
- After 10–20 messages, I’d forget why a certain line of reasoning started.
- After dozens of replies, everything became an endless scroll.
- Valuable ideas were buried somewhere in the history, effectively lost.
My brain was overloaded. The model got smarter,
but my experience of using it wasn’t getting better.
So instead of asking for “better answers”, I started asking:
Maybe what’s broken is not the model, but the interface.
oMyTree is my attempt to fix that:
by giving long AI conversations a tree structure instead of a flat chat log.
At a high level:
- You start a conversation with an AI model (or your own API key).
- Every time you ask a question, oMyTree creates a new node.
- If you ask a follow-up based on a specific message, oMyTree grows a new branch from that node.
- You can click any node to:
- see the full Q&A
- continue the conversation from there
- grow alternate branches of thought
So instead of:
“Where did we ask about X?” → scroll, scroll, scroll…
you get:
“We asked about X on that branch.” → click the node and continue from there.
Here are some concrete things oMyTree is good at:
-
Learning a new concept
Explore definitions on one branch, examples on another, and counter-arguments on a third.
You keep all branches visible instead of losing them in history. -
Researching a topic
Keep separate branches for “background reading”, “data points”, “criticism”, “implementation steps”, etc. -
Debugging & refactoring
One branch for the failing assumption, one for the fix, one for an alternative design. -
Planning & writing
Use branches for different outlines, drafts, or styles, without merging them into one messy thread.
It’s basically a map of “how you and the model got here”.
Some things that are already working today:
-
Visual conversation tree
– One node per message, one branch per follow-up.
– Click any node to focus that sub-conversation. -
Node-level follow-ups
– Ask further questions from any previous node, not just the latest reply.
– Perfect for going deeper without losing alternate paths. -
Model switching with context
– Use different models (including your own API key) in the same tree.
– Light models for quick questions, stronger ones for deep reasoning. -
Adjustable memory scope (still evolving)
– The idea is to switch between “full-tree context” and “branch-only context”.
– This is under active iteration and may change as we refine the UX. -
Export & share
– Export your tree so it’s not locked inside a black box.
– Share a tree with others so they can see the path you took. -
Reply collection
– Mark particularly valuable AI responses to revisit later.
(Some of these are still being refined – see Roadmap below.)
-
It’s not a regular chat window
– Chat UIs are great for short Q&A, but terrible for long-term reasoning.
– oMyTree makes the structure of the conversation first-class. -
It’s not a mind map tool
– You don’t manually drag bubbles and draw arrows.
– The tree grows naturally from your real conversation with the model. -
It’s not a note-taking tool (yet)
– It’s focused on live conversations and the path they take, not on long-form writing.
– But it can feed into your existing note system (via export).
- Frontend: Next.js, React, Tailwind-style minimal UI
- Backend: Node.js / Express-like API layer
- Database: PostgreSQL
- Caching / rate limiting: Redis
- LLM layer:
– Custom abstraction over providers (OpenAI, etc.)
– Support for user-provided API keys (BYOK)
The system started as a personal experiment and has evolved into a full web app with its own routing, event logging, and tree engine.
Right now, the core codebase is closed-source.
Reasons:
- I’m still actively iterating on the core ideas and UX.
- I want to stabilize the architecture and security model before exposing internals.
- Maintaining an open-source repo properly (issues, PRs, docs) is a non-trivial commitment.
This repo exists as:
- a canonical home for docs, screenshots, and the public roadmap
- a place to discuss ideas, open issues, and share feedback
- a neutral landing page I can link from Hacker News, Product Hunt, etc.
I’m open to making more of the system public over time if there’s enough interest and a clear way to do it responsibly.
Some things I’m exploring next:
- Better controls for context scope per branch
- Smarter ways to summarize branches and trees
- Richer export formats (for note apps / PKM systems)
- Collaboration modes (share a tree and continue growing it together)
- Possibly exposing an API so other tools can plug into the tree engine
If you have thoughts on what should exist here, I’d genuinely love to hear them.
I am a soon-to-be-graduated university student in Asia., building oMyTree as a solo project.
The origin story in one line:
I hit the limits of linear AI chats, got frustrated enough, and decided to build the interface I wished existed.
On the Chinese internet I wrote a longer essay about the background and philosophy behind oMyTree (in Chinese):
https://zhuanlan.zhihu.com/p/1981665938895045740
If you’re reading this from Hacker News or GitHub and you’ve ever felt “lost in the scroll” of AI chats,
oMyTree is basically me trying to fix that — first for myself, and now maybe for others too.
If you’re a developer, researcher, writer, or just someone who thinks a lot with AI:
- Does this conversation-tree UI actually help you think more clearly?
- Where does it break for you?
- What’s missing for your workflow?
You can:
- Open an issue in this repo
- Try the app: https://www.omytree.com
- Or just email me:
isbeingto@gmail.com
Thanks for reading, and for caring about better interfaces for thinking.