The Office with AI: Distributed Agency and the Making of a New Sociotechnical Assemblage

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Artificial intelligence has entered office life through an unremarkable route. It drafts an email, summarises a meeting, suggests a reply, retrieves information from a document, ranks possible actions and turns conversations into tasks. Each intervention may appear small. Taken together, however, they change the organisation of a workflow. This paper argues that office AI is better understood as part of an emerging sociotechnical assemblage than as a tool added to an otherwise stable workplace. The analysis brings anthropology into conversation with science and technology studies, drawing particularly on delegation, situated human-machine relations, assemblage theory, algorithmic organising, acceleration and accountability. AI has agency in office work does not mean a language model possesses intention, consciousness or responsibility in the human sense. It indicates that AI systems participate in producing effects: they make some actions easier, visualise particular categories, suggest sequences of work, compress alternatives, redistribute attention and alter the conditions under which people exercise judgement. These changes also transform human agency. Human judgement is exercised under conditions partly produced by the technical system. The central question is how action becomes distributed between people and technical systems, and what happens when this distribution becomes ordinary.