Construction and infrastructure might be the most AI-underserved major industry in New Zealand. It's also one of the most promising. The work is document-heavy, compliance-intensive, and riddled with manual processes that AI handles well. Yet most construction firms haven't gone beyond a ChatGPT subscription.
What You Need to Know
- Construction generates enormous volumes of documents: resource consents, health and safety plans, inspection reports, variation orders, progress claims. Most of this is processed manually.
- Site safety is a high-value, high-stakes AI use case. Automated analysis of site photos, incident reports, and safety documentation can identify risks before they become incidents.
- Compliance checking is where AI shines in construction. Comparing project documentation against regulatory requirements, building codes, and contract terms is exactly the kind of structured analysis AI does well.
- The barriers aren't technical. They're cultural. Construction is a relationship-driven, on-site industry. Technology adoption has historically been slow. AI needs to meet the industry where it is.
7.2%
of NZ's GDP comes from the construction sector, making it one of the largest industries by economic contribution
Source: Stats NZ, National Accounts, 2024
Where AI Adds Value
Site Safety Analysis
New Zealand had 17 workplace fatalities in the construction sector between 2022 and 2024. Safety isn't abstract in this industry. It's life and death.
AI can process site photos to identify safety risks: missing PPE, unsecured scaffolding, inadequate barriers, housekeeping issues. It can analyse incident reports to identify patterns that human reviewers miss when processing hundreds of reports. It can flag overdue safety inspections and predict which sites are highest risk based on historical data.
This isn't about replacing safety officers. It's about giving them better information, faster. A safety team that reviews AI-flagged risks instead of manually inspecting everything can focus their attention where it matters most.
Compliance and Document Processing
Construction projects generate thousands of documents over their lifecycle. Resource consent applications, building consent documentation, health and safety plans, environmental impact assessments, progress reports. Each needs to comply with specific regulatory requirements.
AI can compare documents against regulatory checklists, flag gaps or inconsistencies, and track compliance status across a project portfolio. A consent application that would take hours to review against the building code can be pre-screened in minutes, with the reviewer focusing on the flagged items rather than reading every page.
Project Documentation and Reporting
Progress claims, variation assessments, and project reporting consume significant project management time. AI can draft progress reports from project data, assess variation claims against contract terms, and generate standardised reporting across multiple projects.
Procurement and Estimating
Analysing historical project data to improve cost estimation, comparing supplier quotes, and identifying procurement patterns. This requires structured historical data, which many construction firms are still building, but the potential is significant.
Why Construction Is Different
Construction isn't like financial services or healthcare, where AI adoption is already well underway. The industry has specific characteristics that shape how AI needs to be delivered:
Fragmented workforce. Large projects involve dozens of subcontractors, each with their own systems (or no systems). Any AI solution needs to work across organisational boundaries.
On-site reality. Many workers operate from mobile devices, in environments with variable connectivity. AI tools need to work in the field, not just in the office.
Project-based structure. Each project is unique, with different requirements, different teams, and different documentation. AI systems need to handle this variability.
Regulatory complexity. NZ construction regulation spans the Building Act, Health and Safety at Work Act, Resource Management Act, and numerous standards and codes. AI needs to be accurate and current across all of them.
Getting Started
For construction firms considering AI, the starting points are:
Start with document-heavy processes. Compliance checking, consent review, and reporting are high-volume, document-heavy tasks where AI adds clear value with lower risk than safety-critical applications.
Digitise first. AI needs digital data. If your safety records are in filing cabinets and your project documents are in email attachments, the first investment is in digital document management, not AI.
Pilot on a single project. Don't try to deploy AI across the firm. Pick one project, one use case, and prove the value before scaling.
Involve the site team. Technology deployed without input from the people who use it won't be used. Site managers and project managers need to be part of the design process.
The Readiness Check
If you can answer "yes" to these three questions, you're ready for AI: (1) Do we have digital documents? (2) Do we have a repeatable process we want to improve? (3) Do we have someone who will champion this on-site? If any answer is "no," start there.
The NZ construction industry is worth over $40 billion annually and employs 10% of the workforce. The efficiency gains from AI in construction aren't marginal. They're substantial: faster consent processing, fewer safety incidents, better cost control, and less time on paperwork. The firms that figure this out first will have a real competitive advantage.
Construction is where I see the biggest gap between AI potential and AI adoption in New Zealand. The question is whether the industry is ready for the technology, and I think the answer is closer to yes than most people assume.
Isaac Rolfe
Managing Director