South KoreaPublic sector2024 – ongoing

Architecture & Urban Research Institute

Archilaw V2 building-code interpretation pilot

AURI reported 78% agreement between Archilaw V2 and officials' definitive answers

In progressCompany's own claimOfficial ↗

Problem

Complex building regulations made legal interpretation time-consuming and contributed to permitting delays.

Approach

AURI released Archilaw V2 in 2024 using GPT-4o and retrieval-augmented generation. It incorporated previous inquiries and authoritative legal interpretations. Experts and public officials evaluated its answers.

Results

78%agreement with officials' definitive answers in AURI's evaluationCompany's own claim[1]

60.4%surveyed experts and public officials satisfied with system-generated answers, per AURICompany's own claim[1]

Takeaway

The evaluation covered questions with definitive official answers. Conversation context and reliable incorporation of legal amendments remained limitations.

Sources

AI-Driven Transformation of Building Administration: Achievements of the AI Building Code System and Future Policy Directions
Architecture & Urban Research Institute24 Dec 2025Company publication

Source summaries

[1]Architecture & Urban Research Institute24 Dec 2025

AURI described the tested system and its remaining limitations. It reported 78% agreement with definitive official answers and 60.4% user satisfaction.

Full details are in the original sources.

Added 1 Nov 2026 · Updated 1 Nov 2026