Knowledge graph
vs text files.
Markdown notes (the Obsidian way) and a knowledge graph are not rivals. One is a great way for a human to capture and organise thinking. The other makes that thinking queryable by a machine. The real win is combining them.
One for you,
one for the machine.
Text files are notes you write and organise by hand: an Obsidian vault, a folder of markdown. They are simple, portable, and yours, and they are a genuinely good framework for organising context while you think.
A knowledge graph is a database that stores entities and the relationships between them. It is built to be queried and reasoned over at scale, by software, not just read by a person. Same knowledge, a different shape, made for a different reader.
Side by side.
| Text files (Obsidian) | Knowledge graph | |
|---|---|---|
| What it is | Markdown notes you organise by hand | A database of entities and relationships |
| Best at | Human writing, thinking, capture | Machine querying and reasoning at scale |
| Relationships | Links you add by hand; a visual graph view | Modelled explicitly and traversable by AI |
| AI reasoning | None built in | Built for it: grounded, multi-hop answers |
| At thousands of notes | Hard to query, easy to lose the thread | Holds up; you can ask it questions |
| Best role | Your capture and organising layer | The layer that makes it queryable by AI |
A graph view is
not a graph database.
Obsidian shows a graph view, and it looks like a knowledge graph, so it is easy to assume they are the same thing. They are not. Obsidian's graph view is visual: a picture of which notes link to which, with no AI behind it. A knowledge graph is a database an AI can traverse to answer questions.
One helps you see your notes. The other lets a machine reason over them. Both are useful. They are just different jobs, and confusing them is why people expect their notes app to answer like an assistant and come away disappointed.
Stronger together
than apart.
This is the part I care about most. These two technologies are stronger combined than chosen between. Your text files are the capture and organising layer, the place you think. Feed them into a knowledge graph and that same knowledge becomes scalable and queryable: an AI can answer from it, follow the connections, and stay consistent.
You keep writing in plain files; the graph turns the pile into something you can actually ask. That combination, a human-friendly capture layer plus a machine-readable graph, is about the best context management you can give an AI. It is the same spirit as knowledge graph and RAG: not either-or, but both, each doing the job it is best at.
Common questions.
Do I have to give up my Obsidian notes?
No. Keep writing in plain files. We feed them into a knowledge graph so an AI can reason over them. Your notes stay the capture layer; the graph makes them queryable.
Isn't Obsidian's graph view already a knowledge graph?
Not in the same sense. Its graph view is a visual picture of links. A knowledge graph is a database an AI can traverse to answer questions. Different jobs, and they combine well.
Why not just point AI at my text files?
It can read them, but at scale it keyword-searches and loses the thread. A graph captures the relationships up front so answers stay grounded and consistent.
Turn your notes
into a brain.
Bring your vault to Brain Graph and watch it become queryable, or build it with me in a session.