Sustainable design choices in early-stage real estate development are often based on fragmented knowledge and varying definitions. Teams strive for ambitious environmental goals, but the underlying decision making remains vague. This thesis investigates how a single AI agent, acting as a Digital Building Master, can structure Environmental (E) decision making in the SO and VO phases. Rather than framing environmental design choices as trade-offs between competing objectives, the research explores how integration can reveal relationships and synergies across Environmental criteria in the early design phase.
The research follows a research-through-design approach in which a prototype was developed and tested within an iterative process that ran parallel to insights from practice. The results show that the agent does not add new knowledge. The agent organises existing standards and assumptions in a way that reveals hidden connections. The agent exposes implicit assumptions and identifies tensions at an early stage. The agent links each statement to verifiable sources. This changes the structure of design discussions. Decision-making shifts from linear logic to an iterative pattern in which analysis and sketch development take place simultaneously. This increases transparency and strengthens the stability of Environmental decision making. Interpretative judgement and final design decisions remain with the designer.
The presence of the agent also changes the dynamics between designers. Discussions remain open for longer and reflection becomes more in-depth. However, the structured output format can lead to rapid agreement before all alternatives have been explored. Its use requires sufficient AI skills within the team. The agent functions strongly within clearly defined Environmental themes but has less control over issues of a qualitative or political nature. This thesis concludes that the agent acts as an epistemic catalyst. The agent increases the reflective capacity of teams. The agent distributes cognitive load in a more stable manner and supports a clear basis for Environmental decision making in the early design phase.