Architecture Computation

Architecture Computation is the seminar in which I ask what architectural practice becomes when generative AI is part of it, and I ask it by making the students found something. Every fall since 2023 each team has built a working demo app and produced a three to four minute venture film that pitches an architectural startup. The frame is deliberate: a startup has to demonstrate value to a stranger, so a proposal cannot rest on an image. It needs an app that runs, a pitch that can be followed, a regulation read correctly. The films shown here are snapshots of the apps the students developed, and they are theirs.

The frontier moves each year. In 2023 the projects extended datasets and tuned models towards just and environmentally inclusive building generation, from eye-tracking as design input to compliance review and renovation estimates, and their work was shown in an exhibition of the seminar. In 2024 large language models acted as agents that access and edit supply chains in order to scale carbon-negative building: databases of synthetic materials as carbon supplements, local craft with local materials to reduce transport, reliable cost estimation to favour repair and reuse. In 2025 architecture was treated as a processor of collective patterns, with interfaces, thresholds and superpositions that unlock plural uses and values on the suburban, single-family lots typical of Texas. In 2026 the seminar introduces Synthetic Architecture as a practice, with agentic workflows and vibe coding, through four assignments that build on one another: a speculative visual essay that projects one link of architecture’s material chain forward, a browser simulation in which AI proposals meet a physical behaviour that can push back, an instrument grounded against reality and developed into a working app, and the venture film.

The seminar bridges theory and implementation through weekly lectures, workshops and project discussions, on a cloud computing infrastructure that I customize for the course, with UT’s Discovery Center of Impact strengthening its entrepreneurial side; no prior experience with AI or programming is required. Its projects have left the classroom twice. One grew into master’s thesis research with peer-reviewed publications and became Generative Matter, the pipeline behind the ACADIA 2025 Vanguard Paper Award; another became a startup, Resia, that implements the concepts of its seminar app. The same logic carries the feasibility work I now pursue at the scale of the single lot.