The observation is now well established: professions are evolving faster than training programmes, and the skills acquired yesterday no longer guarantee employment tomorrow. In the face of this acceleration, having a common language to map what one knows how to do — and what remains to be learned — is essential. This is the challenge taken up by the Luxembourg Institute of Science and Technology (LIST) with a skills extraction tool powered by generative AI.
A universal translator for skills
Developed within the framework of the European DS4Skills project, coordinated by DIGITALEUROPE, the tool is built on ESCO, the European multilingual reference framework for skills, certifications and occupations. Its goal: to instantly analyse CVs, job postings and training programmes, and convert them into standardised, comparable skills profiles.
Employment advisers often have to review dozens of CVs written in a wide variety of formats. The tool allows them to quickly extract a structured, comparable reading of candidates' skills. Without replacing their expertise, it saves them valuable time and improves both the guidance and support they provide to candidates.
In practice, when a user uploads their CV, the tool — based on a "Context-Aware Disambiguated-LLM" approach drawing on models from OpenAI and Mistral AI, among others — automatically extracts skills, aligns them with the ESCO taxonomy and assigns each one a semantic similarity score. The user can then view the proposed term, its official definition and the source sentence from the document, while retaining the ability to validate, modify or enrich the results.
From raw data to informed decision-making
Beyond employment advisers, the tool also benefits training managers, who can better align their offerings with the current and future needs of the labour market. By improving the quality of available data and making previously heterogeneous information comparable, it becomes possible to analyse large volumes of job offers in order to identify emerging skills and adjust curricula accordingly.
A dashboard currently being rolled out will soon provide a dynamic view of related occupations, identified skill gaps and similar profiles present on the market — enabling more precise career guidance and better-calibrated training policies.
AI as a lever for equity
In a country where cross-border professional mobility is structural and where the reskilling of the workforce represents a major challenge, having a tool capable of speaking the same language as European employment platforms is a concrete advantage. The LIST tool thus aligns with the vision championed by DS4Skills: building a European skills data space that guarantees interoperability, transparency and ethics.
For the true promise of AI applied to human resources is not to replace human judgement, but to better equip it. When an adviser has, within seconds, a structured profile and a clearly identified gap relative to market needs, they can devote their expertise to what matters most: guiding, orienting and deciding.
Developed by LIST within the European DS4Skills project, in partnership with ADEM, INFPC, CNFPC, Digital Learning Hub and House of Training, this tool will be presented at the LIST stand during Nexus Luxembourg 2026 on 10 and 11 June 2026.