AI in AEC: collapse cognition, not geometry

There is no shortage of excitement about AI drawing buildings. Faster renderings, more variations, fluid form at the speed of prompt — all make for persuasive demonstrations (yes an LLM suggested that emdash). Yet in real estate development, where capital is actually deployed and risk is priced, the binding constraint has rarely been geometry. It is decision velocity under pressure.

In early-stage feasibility, architects are not simply shaping buildings. They are compressing zoning code, construction pricing, financing assumptions, parking requirements, entitlement exposure, and timing risk into something a client can act on. Before anyone cares what a building looks like, a developer needs clear answers to four questions. What can we legally build on this site? What will it cost? Which configuration maximises return relative to risk? And how quickly can we know that with confidence?

Today that process remains fragmented. Zoning PDFs sit in one window, spreadsheets in another, sketch models in a third. Consultant calls fill the gaps, red lines circulate, assumptions are updated, and iterations stretch across weeks. In competitive land markets, that latency carries a cost. Options expire, sellers move on, and competitors who reach conviction faster gain the advantage. If AI is going to matter in AEC, it should not begin with prettier geometry. It should begin by collapsing data latency and compressing cognition.