By Dmitry Paranyushkin, Nodus Labs · Updated
Most tutorials about knowledge graphs and LLMs assume you’ll set up a graph database, write Python, and build a retrieval pipeline. You don’t need any of that. With InfraNodus, a knowledge graph is one paste away: words and entities become nodes, their co-occurrences become connections, and network science does the rest — surfacing the main topics, the structural gaps between them, and the ideas hiding in the periphery (how it works).
You don’t even need an LLM to run this workflow. InfraNodus is a full visual tool that works in your browser and on mobile devices: explore the interactive graph, see the main topics and the structural gaps between them, and generate ideas right in the app. Connecting the same graphs to your LLM via MCP is a powerful second layer — not a requirement.
The shape of your text is the shape of your thinking. Once your notes are a network, you can see what dominates, what is disconnected, and where a new idea would make the most difference.
The workflow rests on that one idea. InfraNodus diagnoses the structure of your discourse (biased, focused, diversified, or dispersed) and — like thought acupuncture — steers your attention (or your LLM’s attention) to the exact points where a small intervention unlocks the most insight.
And because InfraNodus speaks MCP, the same workflow runs inside your LLM: Claude, ChatGPT, or Cursor can build graphs, read their gaps, retrieve from your knowledge base, and save new ideas back — turning a note-taking habit into a compounding AI second brain.
The Four Steps
Create your first knowledge graph from any text, notes, or a YouTube video — then connect InfraNodus to your LLM via the MCP server so every later step runs in both the visual interface and your AI chat.
Read the graph: the topical clusters your text is made of, the content gaps between them, and the insight clusters that InfraNodus proposes to bridge — the thought-acupuncture protocol.
Diagnose the state of your discourse — biased, focused, diversified, or dispersed — and steer it with four frameworks: cognitive diversity, cognitive variability, ecological thinking, and cross-contextual confluence.
Plug your graphs into your favorite LLM: GraphRAG retrieval, persistent memory, the LLM Wiki skill, and PKM workflows that generate new ideas and content from your own knowledge base.
Run the Workflow on Your Own Notes
Every step runs natively in InfraNodus — in the visual graph interface or through the MCP server inside your LLM. Start with a single paste of text and follow the steps.