This thesis examines how practising interaction designers experience working with large language models in their design work. Six qualitative semi-structured interviews with senior and mid-career designers were analysed using codebook thematic analysis (Braun & Clarke, 2021), with a theoretical frame drawn from Schön's reflective practice, Polanyi's tacit knowledge, Hutchins's distributed cognition, and Dingemanse et al.'s figure-ground reframing of the cognitive sciences. Three themes were developed in the analysis: judgement and tacit knowledge; agency and control; artefacts, documentation and role identity. Across all three themes the central finding is one of spread rather than convergence. The designers articulate six distinct positions on control, six distinct positions on professional identity, and several different framings of the human–AI relation. The variation indicates that AI use does not determine designer experience; what is experienced depends on how the designer configures the work, allocates attention across human and AI activity, and maintains the practices through which professional expertise is built. The study contributes a relational reading of AI in design practice; AI as a relational medium-partner — that names a relation the existing vocabulary of tool and colleague does not adequately capture. A more precise terminological framing is identified as a direction for further work. Implications are discussed for design education, for the reproduction of tacit professional knowledge, and for further empirical work on AI-mediated practice.