yggy

yggy, made by Yggdrasil UG in Germany, builds knowledge graphs from existing documents, tables and systems and keeps them current as sources change. Every fact in the graph carries its source document and time, so answers made by people or AI agents can be checked. It is reachable through a web app, a CLI, an API and MCP, and can be hosted in the EU cloud, a private cloud or on-prem.

What It Is

yggy is a knowledge-graph-as-a-service platform. It reads sources such as mail, ERP exports, QMS records, PIM data, PDFs, SharePoint, OneDrive and Google Drive, extracts facts, and matches them to the entities they belong to. Matching entities from different systems, such as a part listed under two numbers and revisions, resolves into one record with the valid answer and its evidence. Unclear cases are routed to domain experts for a decision. It runs on top of existing systems rather than replacing them.

How the Graph Is Built

Extraction is guided by a versioned, RDF-based base ontology exported as OWL 2 RL, and facts that break the ontology's contract are rejected. Structured files are handled through schema-guided mappings that are applied to every row. Each fact stores its source text, valid-from and valid-to dates and a confidence value, and conflicting values on single-valued relations can be flagged for review. Facts can be exported as JSON-LD or N-Quads and the ontology as OWL, SKOS and SHACL shapes.

Developer and Agent Access

Agents such as Claude Code, Codex and Cursor connect over a remote MCP server, and the same URL works in the Claude app as a custom connector. Agents call tools such as search_entities and get_entity_context and receive facts with their sources. The CLI and API offer other ways in.

Product Lines and Use Cases

The vendor offers yggy core for developers, yggy quality for complaints and traceability, and yggy notes for personal knowledge capture. Listed use cases include customer complaints and traceability, spec changes and product compliance, fitment and master data, product knowledge, contract obligations and research data integration.