This project transcribes the digitized corpus of nineteenth-century French patents (brevets d'invention) with over 2,8 million page images across over 330 thousand patents c. 1791–1901, ~9.3 TB) to extract structured inventor and invention metadata (inventor names, residences, patent titles, dates, and technical classes). The resulting database supports research in the economic history of innovation and, in particular, the gender history of invention: systematic identification of women inventors and of the social and geographic diffusion of patenting across the nineteenth century together with technical details on the inventions themselves and their long run changes, questions that cannot be answered without transcribing the corpus at full scale. The project is being funded by Jan Wallander and Tom Hedelius Foundation as well as the Crafoord Foundation. The scans are provided by the patent office of France, INPI, as part of a collaborative effort to disseminate analysable data to researchers broadly.