Obsidian as a Personal Knowledge Graph – Why Notes With Backlinks Change Everything

What this is about

We recently explained what a knowledge graph is and why everyone's talking about it – an enterprise topic with Neo4j, GraphRAG, and compliance. But the same principle works on a small scale too: for you, your notes, your thinking.

The tool that has done this most consistently for years: Obsidian.

What makes Obsidian different

At first glance: just another note app. On closer look:

• Local markdown files – your notes live as .md on your disk. No lock-in, no proprietary format
• Bidirectional links – [[Another note]] automatically creates a connection in both directions
• Graph view – visualizes all your knowledge as a network of nodes and edges
• Plugin ecosystem – 2,000+ community plugins, including dozens for AI workflows

Sounds unspectacular. But that's exactly what makes a knowledge graph: entities (notes) and relationships (links) instead of folder hierarchies.

From folders to graphs

The typical way knowledge is structured – whether in Notion, Google Drive, or your file system – is a hierarchy: folders, subfolders, tags. The problem: knowledge isn't hierarchical. An idea often belongs to three topics at once.

In the graph model there's no "home" for a note. Relationships emerge organically: you link what belongs together, and suddenly patterns appear that would never have surfaced in a folder.

Rule of thumb: Folders answer "where is this?". Graphs answer "what does this have to do with what?".

The Zettelkasten background

The methodology behind this isn't new – it's called Zettelkasten and was perfected by the sociologist Niklas Luhmann. Over his lifetime he produced more than 90,000 handwritten slips with cross-references, building among other things an entire social theory from them.

Obsidian is essentially Luhmann's Zettelkasten – except the cross-references are now clickable and the graph emerges automatically.

Why this is suddenly relevant again: LLMs

Two years ago Obsidian was a tool for PKM nerds (Personal Knowledge Management). Today it's the ideal data foundation for personal AI workflows:

Smart Connections

The Smart Connections plugin generates embeddings from your notes and makes your entire vault available via semantic search – locally, without sending data to the cloud.

Obsidian + Claude / GPT

Through MCP (Model Context Protocol) or custom scripts, notes can be passed directly to LLMs as context. You ask Claude or GPT something and it answers based on your notes – not on the internet. That's exactly the argument from "AI is not the bottleneck. Context is." – except here the context is structured and persistent.

To take it one step further, combine it with NotebookLM + Claude Code as an external source layer for YouTube, PDFs, and the web.

Markdown as a future-proof format

LLMs understand markdown natively. Your vault is automatically in a format that every current and future AI model can process directly. No export, no conversion.

How I use Obsidian

Three use cases that give me the biggest leverage:

1. Meeting notes with backlinks to people and projects – every meeting automatically shows up in the profile of the people involved
2. Research & writing – sources, quotes, and my own thoughts in one graph; articles like this one emerge from the connections
3. Daily notes as the entry point – everything flows into the daily note first, gets linked later, and becomes part of the graph

Obsidian vs. Notion vs. Roam

	Obsidian	Notion	Roam Research
Data	local, markdown	cloud, proprietary	cloud, proprietary
Model	graph + folders	databases + pages	outliner + graph
Backlinks	yes, native	limited	yes, native
AI integration	plugins, MCP, local	Notion AI	limited
Price	free for personal	from $10/user	$15/user
Lock-in	none	high	high

In short: Notion is better for team databases. Obsidian is better for thinking.

When Obsidian pays off

Worth it:

• You write a lot, think in structures, and want to see connections between ideas
• You want full data sovereignty – no cloud vendor that can change pricing tomorrow
• You want to use your notes as context for your own AI workflows
• You're playing the long game (5+ years) – markdown will still be around in 20 years

Not (yet) worth it:

• You need real-time team collaboration → Notion
• You want zero learning curve → Apple Notes
• You think in tables, not texts → Airtable, monday.com

From personal to enterprise graph

The jump from an Obsidian vault to an enterprise knowledge graph is smaller than it sounds. The mechanics are identical: entities, relationships, attributes. The differences are scale, governance, and the question of who writes.

Anyone using Obsidian privately intuitively understands why companies are building Neo4j or GraphRAG stacks. It's the same principle – just that instead of one human thinking, hundreds of agents are operating in parallel.

Conclusion

Obsidian isn't a hype tool. It's the most consistent implementation of an idea that has been haunting the tech world since Vannevar Bush's "Memex" (1945): knowledge should be networked, not filed away.

And with LLMs, the concept is having a second spring. Anyone who starts thinking in a graph today – personally or at enterprise scale – is building an asset that gets more valuable with every new model.

→ Download Obsidian

---

Related reading

• What is a knowledge graph – and why is everyone talking about it? – the same principle at enterprise scale
• AI is not the bottleneck. Context is. – why your knowledge is the foundation for any AI solution
• NotebookLM + Claude Code: How to give your AI access to YouTube – external source layer for your vault
• monday.com vs. Notion: Why Notion isn't a real project management tool – the counterpoint: tables instead of graph