Toward an AI-native architecture practice

Most architecture practices treat AI as a rendering shortcut. We treat it as infrastructure. Over the past two years, Edition has rebuilt its internal workflows so that every stage of a project — feasibility, design studies, consent documentation, construction monitoring — runs through an AI-assisted pipeline maintained by our in-house team.

Why AI-native

The difference is structural, not cosmetic. An AI-native practice does not bolt tools onto an unchanged process; it redesigns the process around what automation does reliably, and reserves architects' time for judgement, negotiation and design intent.

The results are measurable: consent packages assembled in days rather than weeks, design options tested at a breadth no manual workflow could afford, and a documentation trail that reviews itself before council ever sees it. This journal records what we have learned building that capability inside a working New Zealand practice.