CE-RIS is developing an intelligent, agent-based research information system for interdisciplinary and transdisciplinary research in the circular economy (CE). The goal is to systematically index research information from distributed and heterogeneous sources, structure it semantically, and make it available in a way tailored to specific target groups. In doing so, the project addresses a central challenge in CE research: research results are often stored in decentralized locations, are difficult to compare across languages, and are of limited use to academia, industry, policy makers, and practitioners. At its core, CE-RIS combines ontologies, knowledge graphs, and artificial intelligence. Using classical NLP methods, large language models, and progressively expanded neurosymbolic methods, information is extracted from project reports, publications, and other sources, processed, and integrated into a CE knowledge graph. In addition, the Core Research Dataset (KDSF) is being expanded to include a transfer module and a CE-specific module in order to systematically map project results and transfer pathways. This creates a robust semantic foundation for transparent and target-group-specific queries. The technical implementation comprises three levels: data collection to connect relevant sources, integration for semantic structuring and quality assurance, and an application level with web- and chat-based access. Users can ask questions in natural language and receive project-related overviews, thematic maps, or detailed information. The system is being tested and iteratively refined in a real-world laboratory in collaboration with stakeholders from academia, business, politics, and civil society.