Show HN: Kanon 2 Enricher – the first hierarchical graphitization model
3 days ago
- #LegalTech
- #AI
- #KnowledgeGraph
- Kanon 2 Enricher is the world’s first hierarchical graphitization model, transforming unstructured documents into structured knowledge graphs with sub-second latency.
- Isaacus Legal Graph Schema (ILGS) is released, a knowledge graph schema for legal documents, available under CC BY 4.0 license.
- Kanon 2 Enricher is available via Isaacus API and has been tested by notable firms like Harvey, KPMG Law, and Clifford Chance.
- An interactive 3D map of Australian High Court cases since 1903 was created using Kanon 2 Enricher and Kanon 2 Embedder.
- Kanon 2 Enricher can extract, disambiguate, and link entities, deconstruct document hierarchies, and annotate text without generative hallucinations.
- The model is computationally efficient, running locally on consumer PCs, outperforming LLMs like Gemini 3.1 Pro and GPT-5.2.
- Applications include financial forensics, legal research, and regulatory analysis, with real-world use cases from the Isaacus Beta Program.
- Kanon 2 Enricher’s architecture includes 58 task heads and 70 loss terms, enforcing strict schema constraints.
- Future plans include the Blackstone Graph, a public legal knowledge graph, and Kanon 3 Enricher with custom extraction support.
- Kadi, the first legal reasoning model, is expected by the end of the year.