Jobs and Skills Australia has commenced a pilot project to build a National Stills Taxonomy (NST). A skills taxonomy is needed for Australia to support economic research, including analysis of lifelong learning initiatives and mobility within the tertiary education sector. A standardised skills nomenclature will improve linkages between employment opportunities and training outcomes by including skill concepts as part of the jobs and skills metadata ecosystem. Within the tertiary sector, a skills taxonomy will also serve harmonization initiatives between and within training and higher learning institutions and curriculum. The need for a skills-first approach to better education and industry analytics is far from a parochial concern and the NST team is benchmarking the taxonomy project with significant skills metadata projects in other countries. Engagement with other skills taxonomy projects has delivered insights into approaches for skills ontology modelling; skills extraction methodologies; and taxonomy editing workflows. The NST has already learned valuable lessons from a pre-pilot project which trialled artificial intelligence (AI) methods for extracting skills concepts from training literature. We present preliminary findings from this pre-pilot and the resulting methodology for the current pilot phase. We present our evolving approach to AI-assisted skills extraction, and implications for vocabulary modelling and construction. Work on a skill definition is also presented, and the impact that the definition has had on extraction and taxonomy methodologies discussed. We will present upcoming work and highlight challenges and opportunities for the NST in the post-pilot phase.