Reliable data on the Italian Third Sector remain limited despite its socio-economic importance. This paper presents a methodology for assembling a comprehensive dataset on the sector by systematically exploiting administrative data from the Single National Register of the Third Sector (RUNTS). The proposed pipeline automates the acquisition and cleaning of data on more than 134,000 organisations through a tailored web-scraping procedure. To address common shortcomings of administrative data, such as incompleteness and self-reporting bias, a novel webaugmentation strategy is implemented, combining information extracted from organisational websites with a Generative AI–based classification system. This approach allows the identification of beneficiaries, territorial scope, and economic activities, and enables comparison between web-derived and self-declared information for the latter case. Using data for the second quarter of 2025, the paper provides an up-to-date picture of the Italian Third Sector and offers a new informational base for socio-economic research and co-programming of local policies.
