Artificial Intelligence could become a powerful catalyst for transforming Australia’s prefabricated construction sector, but only if the industry confronts long-standing weaknesses in data, skills and governance. That was the central message delivered by Associate Professor Dominik Holzer from the University of Melbourne at Offsite25, the annual prefabAUS conference held last year.

In a presentation titled From AI to Factory, Holzer outlined how AI has the potential to reshape prefabrication across design, manufacturing, logistics and delivery, while also warning that poor preparation could limit its impact or introduce new risks.

Holzer opened by setting out the scale of Australia’s housing challenge. National Housing Accord targets are slipping further out of reach, with dwelling approvals continuing to lag well behind required levels. At the same time, the speed of delivery has deteriorated. A decade ago, homes typically took around six quarters to move from construction start to completion. Today, that timeframe has stretched beyond nine quarters.

According to Holzer, the slowdown cannot be explained by a lack of technical capability alone. Contract structures, approval processes and regulatory complexity are placing heavy constraints on productivity. These issues are compounded by an acute labour shortage, with Australia currently facing a shortfall of approximately 83,000 tradespeople.

While construction has historically been slow to adopt digital tools, Holzer argued that AI differs fundamentally from previous technologies. Unlike Building Information Modelling, which mainly improved spatial coordination, AI can influence decision-making across finance, procurement, production planning, logistics and operations.

AI’s strength lies in its ability to identify patterns, connect fragmented information and support real-time decision-making across complex systems. This breadth of application gives it the potential to lift productivity in ways earlier digital tools could not.

Holzer outlined how AI could be deployed throughout the prefabrication workflow. In early-stage feasibility, AI systems can support market analysis, demand forecasting and site assessment, while dynamically updating financial models as conditions change.

In production environments, AI can be used to optimise schedules, reduce cycle times and automate compliance checks through integration with BIM systems. On factory floors, applications include predictive risk analysis, inventory optimisation and computer-assisted visual inspections that identify defects or deviations from design intent far more quickly than manual processes.

Logistics is another area of opportunity, with AI capable of simulating transport processes, identifying supply chain vulnerabilities and optimising delivery routes. On-site, AI-supported tools can help manage risk, plan crane movements and assist installation through augmented reality systems.

Holzer’s presentation came amid growing government support for Modern Methods of Construction. Queensland has committed to a 50 per cent MMC target for government projects as it prepares for the 2032 Olympics, while also establishing a dedicated MMC sub-group within the Queensland Building Ministerial Advisory Council.

Other states are also moving. New South Wales has launched a $10 million modular housing pilot with fast-tracked approvals. Victoria and Western Australia have each committed $50 million to housing innovation initiatives, while South Australia and Tasmania have rolled out MMC-focused social housing programs.

Byron Bay House by PreBuilt

The finance sector is beginning to adapt as well. CommBank now offers prefab-specific lending products that allow access to up to 80 per cent of the contract price before on-site installation, addressing one of the sector’s long-standing funding challenges. A Federal Treasury working group is also examining remaining barriers to MMC finance.

Despite the excitement around AI, Holzer stressed that data readiness is the single most important prerequisite for success. Many organisations rush to adopt AI tools without investing in data cleansing, tagging and standardisation, limiting the technology’s effectiveness.

AI systems require well-structured, high-quality data sets that reflect an organisation’s specific activities and history. This involves analysing past projects, standardising records and enabling systems to extract meaningful insights from historical performance. Holzer cautioned against using AI to homogenise approaches across the industry, noting that competitive advantage depends on organisation-specific knowledge.

Construction’s project-by-project nature has historically undermined knowledge sharing, even within individual firms. AI offers a way to address this weakness, but only if the underlying data foundations are sound.

Holzer also highlighted unresolved legal and governance issues. Liability remains firmly with human decision-makers, even when AI systems are involved. Organisations cannot shift responsibility to algorithms when errors occur, making human oversight essential through expert-in-the-loop systems.

Data ownership and intellectual property protection pose further risks. AI systems consume large volumes of data, increasing the potential for unintended information leakage or IP infringement. Holzer emphasised the need for strong data governance frameworks and AI-aware insurance and risk policies.

For companies looking to take first steps, Holzer pointed to relatively accessible applications such as automated quality checks, logistics optimisation and improved access to internal knowledge in real time. These use cases can deliver tangible benefits but still depend on disciplined data management.

He recommended that organisations establish dedicated AI steering groups to oversee data strategy, governance and ethical considerations. Clean, consistent and purpose-fit data remains the foundation for any successful AI deployment.

AI adoption will reshape roles across prefabrication, particularly in planning, quality assurance and materials handling. Rather than eliminating jobs, Holzer argued that AI will require new forms of collaboration between workers and intelligent systems.

He also urged caution around environmental impacts. AI-driven systems rely heavily on energy-intensive data centres, and not every task justifies their use. Responsible implementation must consider sustainability alongside productivity gains.

Holzer concluded by stressing that AI is not a simple productivity switch. It represents a broader transformation in how the construction industry captures knowledge, manages information and learns from experience.

Realising AI’s potential will require coordinated action across industry, government and education, grounded in robust data strategies and clear governance. For prefabrication to play a meaningful role in addressing Australia’s housing shortfall, AI must be approached as a long-term capability-building exercise rather than a quick technological fix.

Associate Professor Dominik Holzer conducts research into AI adoption across the architecture, engineering and construction sector and advises industry on strategic implementation through his consultancy, AEC Connect.