Skills Intelligence and Skills-Based Organizing: A Bibliometric Mapping of Two Disconnected Literatures (2000–2026)

Skills-based organizing replaces jobs and credentials with information on skills, which artificial intelligence now extracts from job postings, résumés and human resource records at scale. Whether this information is valid for the decisions it supports depends on whether research on skills intelligence and research on human resource management inform each other. We mapped 3649 documents published between 2000 and 2026, screened with a large language model audited against a human reviewer, using co-citation, bibliographic coupling and permutation benchmarks. Managerial and computational research form separate communities: only 13% of coupling links connect them, against 46% expected by chance, and the divide is sharpest between studies of the same practice, recruitment and selection. Computational research, 71% of output in 2023–2025, is itself split between recruitment technology and labor-market analytics; it validates skill information mainly against annotated text, about one study in seven relates it to outcomes, and almost none evaluates the decisions it informs. Taxonomies such as ESCO and O*NET do not bridge the communities; work analysis does. We propose a skills intelligence value chain from data to outcomes, with five testable propositions and a research agenda for validating AI-produced skill information.