Knowledge management (KM) research has long recognised that organisational knowledge extends beyond codified information to include tacit, experiential, and relational forms of knowing. However, contemporary organisations increasingly rely on artificial intelligence (AI) and data-driven systems that privilege abstraction, codification, and optimisation. While these systems enhance efficiency and scalability, they risk marginalising embodied and contextual knowledge that is essential for creativity, innovation, and sustainability. This paper addresses this tension by examining how organisations can develop intelligent knowledge systems while preserving the embodied foundations of human knowing. Drawing on qualitative data from 21 semi-structured interviews with innovation leaders, managers, and artists across ten European organisations, the study investigates how arts-based initiatives (ABIs) generate and mobilise embodied knowledge in organisational contexts. ABIs, defined as structured engagements embedding artistic practices within organisational processes, provide a unique empirical setting to explore knowledge that is experiential, sensory, and difficult to codify. The findings reveal that embodied knowledge constitutes a distinct form of organisational knowledge that operates through experience rather than representation. Its integration into organisational processes depends on relational translation mechanisms, including intermediaries, boundary artefacts, and collaborative practices that bridge artistic and technical domains. At the same time, AI-enabled systems emerge as a structural pressure toward abstraction, privileging codified knowledge and reinforcing optimisation logics. This creates a tension between efficiency-driven knowledge systems and the forms of knowing required for innovation under uncertainty. The paper introduces the concept of embodied knowledge governance, which explains how organisations can recognise, integrate, and sustain experiential knowledge alongside codified and AI-mediated knowledge. This framework highlights the importance of translation, interaction, and organisational conditions that preserve contextual meaning. The findings further show that embodied knowledge plays a critical role in sustainable innovation by enabling long-term, systemic thinking and expanding value perspectives beyond short-term performance. This study contributes to KM, AI, and innovation research by advancing understanding of knowledge plurality in organisations and by highlighting governance as a key mechanism for balancing intelligent systems with human creativity and sustainability.