The knowledge graph for Claude and ChatGPT
A knowledge graph
your AI can read.
Link your notes and Hjarni draws the map. Every note is a dot, every wiki-link a line. You see how your knowledge connects. Claude and ChatGPT follow the same links to gather context before they answer.
Free to start. No credit card required.
The problem with folders
Folders make silos.
A meeting note sits in one folder. But it relates to a person, a decision, and a research topic that live somewhere else. A folder search never finds those connections.
Links do. When notes link to each other, your AI can trace the relationships: ask about a project and Claude follows the links to the related meetings, decisions, and sources. Context a folder search would miss.
The graph is that web of links, made visible. For you, and for your AI.
How it works
Link notes. See the map.
Wiki-links between notes build the graph. You write them, or your AI writes them for you.
Wiki-links, bidirectional
Write [[42:Note Title]] and both notes know about each other. Every note shows what it links to.
Your AI follows the trail
Through the built-in MCP server, Claude and ChatGPT see a note's links and traverse them to gather context before answering.
Hubs and orphans at a glance
The graph view makes hub topics stand out and shows orphan notes in amber, so you know what to connect next.
Not just a pretty picture
Obsidian draws a graph. Yours is readable.
A graph you can only look at is a poster. A graph your AI can traverse is memory.
Obsidian graph
Beautiful, local, and for your eyes only. No MCP, so ChatGPT and Claude never read it. Compare Hjarni vs Obsidian.
Notion / Apple Notes
Folders and pages, no graph and no link traversal for AI. Your assistant searches blind.
Hjarni
A graph you can see and an AI that follows the same links through MCP. Plain Markdown wiki-links, nothing proprietary.
I wanted a cloud knowledge base I could use as memory for my AI, in plain Markdown, with read and write access. Nothing else fit. Hjarni did. What sold me was how light and well defined the MCP is. My AI used to choke when I pasted Markdown and config files into a chat. It reads them perfectly from Hjarni. The folder-level AI instructions and the knowledge graph are the parts I did not know I needed.
Common questions
Questions about the knowledge graph
What is a knowledge graph in Hjarni?
Your notes drawn as a map. Every note is a dot, every wiki-link a line. Hub topics stand out and orphan notes show in amber. It is a picture you can read and a structure your AI can traverse.
Can my AI actually read the knowledge graph?
Yes. Through the built-in MCP server, Claude and ChatGPT see a note's links and follow them to gather context from related notes before they answer. That is the difference from a graph you can only look at.
How is this different from the Obsidian graph?
Obsidian's graph is local and for your eyes. It has no MCP, so ChatGPT and Claude cannot read it. Hjarni's graph is built from plain Markdown wiki-links and is readable by any AI you connect.
How do I build a knowledge graph from my notes?
Link related notes with wiki-links like [[42:Note Title]]. Both notes then know about each other, and the graph fills in automatically. You can write the links yourself or ask your AI to add them as it creates and updates notes.
Do I need to be technical to use it?
No. Wiki-links are plain Markdown, and your AI can add them for you. The graph view is a single page in the app. There is nothing to configure.
Connect your thinking. Once.
Start linking notes today and watch the graph fill in. Free to start.
Free to start. No credit card required.