AI-baserat SOP- och beslutsstöd för restaurangpersonal

Sammanfattning

Background. Restaurant employees work in one of the most demanding service environments which is characterized by high workload, staff shortages and role overload. Standard operating procedures exist to ensure consistency and quality but research shows that compliance with these decreases under work pressure. Meanwhile, existing AI solutions in the restaurant industry primarily target guests and administrative tasks rather than supporting frontline staff in their daily work. Objectives. This study investigates how a large language model based system can support restaurant staff with standard operating procedures and decision support in real time, and how staff experience the usability, reliability and usefulness of such a system in practice. Methods. Using a qualitative single case study combined with design science methodology, a web application was developed and evaluated at the restaurant Boca in Spain. Data were collected through semi-structured interviews and observations with four participants across three phases. Results. The system functioned as an augmentation tool that made documented routines accessible at the moment they were needed, reducing dependence on colleagues and increasing staff autonomy. The reliability of responses was identified as the most critical factor for the system´s practical value. Key limitations included the absence of hands-free interaction, situational stress reducing usage during peakperiods and the administrative burden of maintaining system content. Conclusions. A large language model based system has potential as a meaningful support tool for restaurant staff, provided that identified technical and contextual challenges are addressed in future iterations. The study confirms augmentationas a suitable interaction model in service environments characterized by high task variation, decision-making and non-codified knowledge.