A design space for AI integration in early-stage production plant planning

Early-stage production plant planning is characterised by evolving requirements, incomplete process definitions, and information distributed across documents, engineering models, and planning tools. Introducing artificial intelligence in this setting raises a fundamental question: what must be considered before a specific application, data model, or technical solution can be defined? Existing research typically starts from an already identified planning task, a selected artificial intelligence method, or an available technical infrastructure. It therefore provides limited guidance for structuring the initial need and determining how artificial intelligence should enter planning work. This paper develops a design space that addresses this early decision problem. Literature from factory planning, production planning, collaboration between humans and artificial intelligence, semantic technologies, and digital planning was compared to derive five dimensions: planning scope, information structure, authority to modify planning results, connection to planning tools and data, and functional contribution. Data quality, organisational roles, governance, and interaction are considered as additional conditions. Application to process failure mode and effects analysis shows how the appropriate configuration changes as preliminary process information develops into a maintained planning record. The design space helps planners, artificial-intelligence specialists, and system engineers specify implementation requirements before selecting a particular technical solution.