This paper exploits natural language processing techniques to detect explicit labour-saving goals in inventive efforts in robotics and assess their relevance for different occupational profiles and the impact on employment levels. The analysis relies on patents published by the European Patent Office between 1978 and 2019 and firm-level data from ORBIS® IP. It investigates innovative actors engaged in labour-saving technologies and their economic environment (identity, location, industry), and identifies technological fields and associated occupations which are particularly exposed to them. Labour-saving patents are concentrated in Japan, the United States, and Italy, and seem to affect low-skilled and blue-collar jobs, along with highly cognitive and specialised professions. A preliminary analysis does not find an appreciable negative effect on employment shares in OECD countries over the past decade, but further research to econometrically investigate the relationship between labour-saving technological developments and employment would be helpful.