At Siemens PLM Connection 2026, which drew more than 800 attendees and featured over 90 sessions and talks, an approach for AI-based recursive clustering of technical solutions from fragmented data sources was presented. Katharina Schmucker of Siemens AG, who is pursuing her doctorate at the Institute for Information Management in Engineering under the supervision of Prof. Dr. Dr.-Ing. Dr. h. c. Jivka Ovtcharova, showed under the motto "Class trees were yesterday, clustering is today" how technical solutions can be structured in a data-driven and dynamic way, "from coarse to fine". Building on research by her Siemens colleague Dr. Jonathan Leidich, the approach uses global and local technical similarities to recursively cluster technical solutions from fragmented data sources. Unlike static, predefined classification structures, this makes it possible to uncover implicit technical relationships, distinctions, and variant structures in a data-driven way and to refine them continuously. The research goes beyond simply structuring and searching technical solutions. The clustering is part of a broader approach that both builds an evolving engineering knowledge base and develops ways to put it to practical use. To this end, agentic AI methods are to be developed that can use this knowledge reliably and at scale for specific use cases and their engineering tasks. Among other things, the methods presented can help support configure-to-order processes and variant management in a data-driven way and reduce the manual effort involved.