The Context
New Zealand recorded 33,530 residential consents in the year ending May 2025 — a 3.8 percent decline on the previous year, with multi-unit consents down 8.6 percent and standalone house consents up 2.4 percent. Within this context, Building Research Association of New Zealand (BRANZ) is examining how artificial intelligence could reduce the administrative burden of the consent process — a process that currently involves significant manual effort in reviewing submitted documentation, identifying missing items, and managing communications with applicants.
What Building Consent Officers Say
BRANZ’s research has directly consulted building consent officers (BCOs) about the specific workflow obstacles they face and where AI assistance would be most valuable. The priorities they have identified:
- Pre-submission verification: current pre-submission checks confirm that documents are present, not that they are complete or accurate. Missing site layouts, incomplete construction specifications, and inconsistent information between documents are common reasons for processing delays. BCOs indicate that AI could strengthen preliminary assessment by detecting absent components before formal lodgement.
- Technical processing: BCOs see value in AI tools that can rapidly cross-reference technical requirements — standards references, material compatibility protocols, and code compliance pathways — reducing the manual research time currently required for complex applications.
- Communication: BCOs seek AI support for drafting clearer Request for Information (RFI) correspondence that simplifies technical concepts for applicants who are not Building Code specialists. Poorly understood RFIs slow processing and create back-and-forth exchanges that consume officer time without advancing the consent.
Limitations and Next Steps
BRANZ’s research is explicitly exploratory at this stage — understanding the specific pain points before designing interventions. The next phase involves consultation with applicants — builders, designers, and homeowners — about the AI tools that would reduce their preparation burden and improve the quality of applications before submission. The challenge for any AI solution is that building consent applications involve significant contextual judgement: a documentation checklist can identify a missing item, but determining whether a proposed solution adequately addresses a code clause requires professional expertise that current AI cannot replicate. The realistic near-term application is augmenting BCO capacity for the systematic, repeatable elements of the workflow — not replacing the professional judgement that the consent system ultimately depends on.
