AgentLEX is a multi-agent system that discovers, extracts, and tracks regulatory requirements for any product. Given a product description (features, markets, materials), it identifies relevant EU (and optionally global) regulations, extracts structured information (IDs, clauses, dates, jurisdictions), stores them in a persistent registry, detects new or updated regulations, generates compliance summaries with mitigation measures, and continuously evaluates its own performance through quality tests.
AgentLEX aims to:
- Discover all relevant regulations for a given product in the EU (and optionally worldwide).
- Extract structured regulatory data: IDs, clauses, dates, jurisdictions, scope.
- Store results in a persistent, versioned regulatory registry.
- Detect new or updated regulations on subsequent runs.
- Generate a compliance summary and suggested mitigation measures.
- Perform continuous quality evaluation across agents and pipelines.
Regulations are numerous, evolving, and fragmented across multiple sources (EUR-Lex, ECHA, ESPR, AI Act, RoHS, etc.).
Engineers, compliance officers, and LCA practitioners increasingly need fast, explainable answers to:
- “Which regulations apply to my product in market X?”
- “What changed since last time?”
As virtual twins (e.g., 3DEXPERIENCE-based digital twins) become central to product development, their value increases when they become compliance-aware.
A multi-agent, ontology-driven system can transform raw legal text into structured, reusable, and version-controlled regulatory knowledge.