Agentic AI
AI Framework Automates EU Cyber Resilience Act Compliance Documentation for SMEs
Researchers developed an agentic RAG system that automatically generates assurance cases for regulatory compliance, cutting manual effort while maintaining transparency and evidence traceability.
Meeting the EU Cyber Resilience Act's requirements poses a significant operational burden for small and medium-sized enterprises, particularly when it comes to demonstrating cybersecurity compliance and managing certification documentation. A new research effort presents an automated solution designed to ease this administrative load.
The Challenge: CRA Compliance for SMEs#
The EU Cyber Resilience Act mandates that companies prove their products meet stringent security standards. For resource-constrained organizations, assembling the necessary evidence and documentation to demonstrate compliance—known as assurance cases—requires substantial time and expertise. This gap between regulatory demands and practical capacity has motivated the search for automation approaches.
How the Framework Works#
Researchers introduced a system combining agentic Retrieval-Augmented Generation (RAG) with formal logical structures based on Claim-Argument-Evidence reasoning. The framework maps technical documentation systematically to generate comprehensive assurance cases automatically. By grounding responses in actual technical artifacts and formal logic, the approach maintains interpretability and traceable decision-making rather than treating compliance as a black box.
Real-World Validation#
The team tested their approach on Catalink's PATROLIoT wildfire monitoring system. The system generated 70 assurance cases with strong grounding—averaging 4.4 supporting artifacts per case. A Natural Language Inference evaluator achieved 0.88 accuracy in validating evidence connections, while expert assessment of plausibility scored 3.06, indicating that the automated justifications remained credible and comprehensible to human reviewers.
Practical Impact#
This work offers practitioners a scalable, transparent method for automating compliance workflows. By reducing manual burden while preserving audit trails and interpretable reasoning, the framework enables smaller organizations to meet regulatory obligations without proportionally inflating their compliance teams. The approach prioritizes both efficiency and accountability—critical factors in security-sensitive domains.

