The Iksan Knowledge Industry Centre is an applied case of an AI-based design process in which the city is first interpreted as data, and design is iteratively converged through the repeated coupling of an LLM and generative AI. Rather than describing the surrounding city through impressions or narrative, the project begins by translating form, regulatory logics, and access conditions into structured inputs. This urban embedding, through the implicit computation of non-linear data, reveals recurring development patterns and compound constraints, establishing a baseline against which design can operate.
​
A reference set is then assembled by selecting precedents with comparable urban conditions through morphological and operational similarity. The LLM translates the brief requirements and observed conditions into explicit, verifiable rules. Generative tools rapidly visualise options derived from these rules, and each iteration is immediately tested against evaluation criteria such as planning and code compliance, programmatic viability, operational feasibility, and the validity of urban connectivity. The review outcomes are fed back to update constraints and rules, and as this loop accumulates, the design becomes progressively more precise.
​
The significance of this process lies in using AI not as a mere instrument for image production, but as an operative methodology that systematically generates, verifies, and revises design decisions under complex urban conditions. The outcome is not architecture fixed by a singular intuition or a linear “concept-to-form” sequence, but an adaptive architectural system—converged through urban analysis and iterative validation—capable of responding to evolving requirements.

Video
https://vimeo.com/1119669601?fl=ip&fe=ec
https://vimeo.com/manage/videos/1129786739