Work · The story
AI Tree
Learning AI through interactive, branching visualizations.
A platform for learning and sharing knowledge about AI using interactive visualizations and a branching tree metaphor — concepts split into explorable paths instead of a flat article.
TL;DR
A learning platform where AI concepts live on an expandable tree — the structure of the knowledge becomes the interface, not a flat article.
At a glance
- Interactive branching concept tree
- Content stored in Supabase, evolves without redeploy
- Ongoing — format questions still open
- Status
- ongoing
- Period
- 2023 — present
- Role
- Builder
- Stack
- Next.js · React · TypeScript · Supabase
01
Problem
Flat articles and scrubable videos don't map to how concepts actually branch — I wanted the structure to be the interface.
Most AI learning material is either a flat long-form article or a video you scrub through. Neither maps well to how concepts actually relate — a topic branches into prerequisites, alternatives, and deeper specializations. I wanted a format where the structure of the knowledge is part of the interface.
02
Approach
A Next.js app where content lives on an expandable tree, backed by Supabase so it grows without redeploying.
Built a Next.js app where content lives on a tree you can expand and collapse. Each node is a focused concept with a short explanation and links to children that go deeper or sideways. Supabase backs the content so it can grow over time without redeploying. The interaction model borrows from file explorers and outliners — things developers already know how to navigate.
The hard part wasn't rendering a tree — it was deciding what deserves to be a node. Too granular and it's overwhelming; too coarse and it's just an article with arrows.
03
Outcome
An ongoing experiment — the open question is whether the tree metaphor holds as content scales, or whether it needs to become a graph.
An ongoing experiment in making structured knowledge feel explorable rather than imposed. It's still growing — the interesting question is whether the tree metaphor holds up as content scales, or whether it needs to become a graph.
04
What I took away
- Content structure is a UX decision, not just an information-architecture one.
- Trees are intuitive to navigate but brittle to author — the authoring experience deserves as much thought as the reading experience.
- Supabase is a great fit for content that should evolve without a CMS or redeploy loop.
Want to talk about building something like this? Let's talk or email pedrocarlos.ti@gmail.com.