The integration of large language models-cognitive assistants (LLM-CAs) in manufacturing within quality management systems (QMS – ISO 9001:2015) presents both transformative opportunities and significant challenges. This study systematically analysed 53 studies (2022–2024) to derive literature-based requirements for LLM-CAs in manufacturing within QMS according to the Humans, Technology, and Organization concept (HTO). Findings highlight persistent challenges, such as transparency, compliance risks, and workforce adaptation, with strong alignment between LLM-CAs and QMS principles. Requirements were categorized into HTO subsystems and their overlaps, ensuring a structured approach for seamless integration. As a result, the potential of LLM-CAs to enhance decision-making, operational efficiency, and human-centred governance within QMS is underscored. On that basis, an essential synergy between traditional practices in manufacturing such as QMS and novel technology is highlighted. This synergy represents a fundamental step to guarantee the deployment of LLM-CAs in manufacturing. Future research should explore stakeholder-driven validation, digital feedback loops, and LLM-CAs-driven continual improvement in manufacturing.