The Trillion-Dollar Leak: AUT and Soterra Partnering to End Construction Rework
In the construction industry, some of the most expensive problems are the most familiar: failed inspections, repeated defects, rework, programme delays, and lessons learned too late.
Across New Zealand and Australia, around one in eight hours is lost to rework, while globally, rework is estimated to cost the construction sector up to US$1 trillion each year. For construction-tech startup Soterra, those numbers reflect a reality founders Maree Lamositele and Adam Domok have seen first-hand across residential and commercial projects.
“After years in construction quality, we kept seeing the same issues repeat from one project to the next,” says Adam. “The knowledge was there, buried in inspection reports and site notes, but it wasn’t being turned into guidance teams could use before problems happened. That was the moment we realised there had to be a smarter way to help companies learn from their own history.”
That idea is now the focus of a new collaboration between AUT and Soterra, supported by a Tīmatanga Initiate Grant from the AUT Innovation Fund. Together, the team is developing a smart construction platform that acts as a private AI co-pilot for building firms.
Turning Hindsight into Foresight
Led by Associate Professor Farhaan Mirza, Director of Computer and Information Sciences Research Centre (CISRC), the project uses artificial intelligence to learn from historical inspection data. By identifying repeat defects, rework drivers, time-to-close trends, and inspection-readiness patterns, the platform aims to turn old project data into company-specific intelligence that can guide future work.
“We are applying advanced machine learning to an under-digitised sector to create measurable productivity gains,” says Associate Professor Mirza. “Instead of just filing away inspection reports, this technology turns that data into actionable, company-specific intelligence. Receiving the Tīmatanga Grant has allowed us to build a working prototype and test the concept in a real-world environment.”
Industry Experience Meets Academic Expertise
A major strength of the project is its deep connection to the realities of construction sites. Soterra’s founders bring 20 years of combined experience in construction management and quality assurance, along with access to anonymised datasets and pilot opportunities through construction industry partners.
While AUT leads AI & software development, Soterra is contributing domain expertise, data preparation, scenario testing, and continuous validation to ensure the prototype reflects how construction companies actually work.
The project is already underway and is set to deliver a commercially viable prototype that Soterra plans to integrate into its product roadmap. It is a powerful example of how pairing university data science expertise with commercial know-how can help build a smarter, more resilient future for Aotearoa’s construction sector.