Kanon 2 Enricher – the first hierarchical graphitization model
58-task-head model that extracts + links entities + maps doc hierarchy—no hallucinations like LLMs.
Turn any collection of documents into a knowledge graph. Extract entities and relationships via LLM, deduplicate with your approval. Map domains, find hidden connections, spot patterns across documents — knowledge that persists and compounds, for you and your AI agents. All from the CLI.
End-to-end and local-first: point it at PDFs or docs, and it extracts entities/relations with LLMs, proposes merges for you to approve in a terminal UI, then generates an interactive browser viewer and standard graph exports. The human-in-the-loop merge workflow and support for local providers (Ollama/LiteLLM) are smart, practical choices; just remember output quality and scale will still hinge on the LLM you pick.
Researchers, data scientists, analysts, knowledge managers, and developers who need to build and explore knowledge graphs from document collections
58-task-head model that extracts + links entities + maps doc hierarchy—no hallucinations like LLMs.
Markdown-to-mind-map rendering is slick, but Obsidian Canvas already maps local notes better.
Knowledge graph generation from papers, but Elicit and Consensus already do literature synthesis.
Instant CSV-to-chart conversion without signup or complex configuration settings.
Graph-walking MCP tools beat RAG for agent memory when nobody solved coordination yet.
Per-step analytics identify exactly where docs fail, beyond simple page views.