Developed by a consortium of partners including V‑LAB (project lead), Tricore Technical Services, Correl Group, Brunel University London, TWI Ltd, and The Power Hub, SafeXtend was supported by the Innovate UK BridgeAI programme. Thanks to such support SafeXtend progressed from an early-stage concept to a validated, AI-enabled training platform. The project addresses an urgent need for scalable and skills‑targeted training that reflects real construction environments, helping workers prepare more effectively for the risks they face on site.

The challenge

Construction sites are complex, high‑risk environments where workers must quickly understand, identify and mitigate hazards that vary significantly across location, role and activity. Traditional safety training typically relies on generic guidance delivered through classroom sessions, lectures, or pre-recorded videos, which is rarely tailored to site-specific conditions. These sessions can also reduce engagement with training and hinder knowledge retention.

The existing challenges are compounded by the realities of construction work. Sites are mobile and dispersed, with limited digital connectivity — on average, only around 15% of workers have reliable internet access on site. In‑person training is also difficult to schedule around fluctuating project timelines, resulting in low attendance rates for external courses. As a result, many standard training programmes can only cover a small proportion of the skills, processes, and equipment used on real job sites, contributing to poor learning outcomes.

Until recently, technology constraints have further limited innovation. Digitising construction sites often required expensive, specialised equipment, while VR was perceived as a high-cost technology with uncertain return on investment. However, advances in photogrammetry and LiDAR scanning technology – used to create digital models of physical environments – have made high-quality 3D site documentation more accessible, which has opened up new opportunities for AI-enabled, site-specific training.

The solution

SafeXtend is an AI‑powered VR training platform that transforms 3D site scans into immersive, role-specific safety simulations and training scenarios. Rather than relying on generic environments, the platform generates personalised training modules based on realistic digital representations of active construction sites. By combining AI-driven scenario generation with real-world site digitisation, SafeXtend enables hyper-contextualised training that adapts to specific project environments, roles, and risk profiles.

The platform automatically converts raw 3D scans into interactive VR environments and uses AI to generate customised training programmes. These scenario‑based modules typically last 13-15 minutes – highly advantageous for knowledge retention – and are tailored to individual skill needs identified through embedded assessments. Workers can train using VR headsets, with optional treadmill integration to support natural movement. Remote streaming and multi-user functionality also enable collaborative learning and teamwork with increased accessibility.

With support from the BridgeAI programme, the consortium integrated advanced VR, AI and streaming capabilities into SafeXtend and validated the platform across two live construction sites. Through this, the team gained strong industry exposure at events such as UK Construction Week (UKCW), helping build momentum and visibility across the sector to enable wider adoption.

The impact

Early trials have demonstrated clear impact in terms of helping to address skills shortages, improve safety outcomes, and support the ongoing upskilling of the construction workforce. Training time was reduced by 30–40% across pilot environments, while workers showed higher levels of knowledge absorption and retention compared with conventional training methods. Most importantly, the immersive, site-specific scenarios helped teams spot hazards they might otherwise miss, contributing to measurable reductions in on-site incidents.

Described as “engaging and intuitive’, SafeXtend has sparked interest from new collaborators, including Moira One and the wider CAB Networks, who are exploring partnerships and rollout opportunities. The platform has already captured industry attention, with strong visibility at UKCW Birmingham 2024 and UKCW London 2025, alongside invitations to present at the CAB Transforming the Glazing Supply Chain with AI and VR Innovations webinar and ‘Building in the Future’ Technical Conference 2025. SafeXtend has also gained international recognition, including a presentation at a Royal Academy of Engineering–hosted event in Kenya. The platform’s recognition through the BUILD Design & Build Awards 2026 further validates its commercial and technological impact, positioning SafeXtend as a leading solution in next-generation safety training.

Finally, Safextend was featured as one of the externally communicated use cases through the AI Skills Hub UK platform, providing public visibility. This external profiling helped position the project within a wider national conversation on AI capability-building and real-world adoption. This demonstrated its relevance to broader awareness, credibility, and knowledge sharing across the UK AI skills and innovation ecosystem.

The future

SafeXtend is now positioned for market-ready deployment. The consortium is expanding use cases across additional construction sectors such as infrastructure, energy, and heavy civil engineering. It also plans to continue refining VR comfort and accessibility features and developing commercial partnerships with training providers and contractors. Future plans include international deployment across emerging markets, integration with digital twin ecosystems, and expansion into adjacent sectors such as energy, offshore, and infrastructure.

SafeXtend represents a step-change in how construction safety training is delivered. By combining AI, immersive VR, and real-world site data, we are enabling workers to train in safe digital twin virtual environments that truly reflect the on-site risks they face, improving both safety outcomes and workforce readiness.

– Dr Vladimir Vukovic, Project Lead, V-LAB