AI Risk & Safety Assistant for Construction Sites
We developed a bespoke Voice AI Agent web application for Morgan Sindall to support on-site teams with real-time health and safety guidance.

What we delivered
- Bespoke Voice AI Agent web application
- RAG-grounded safety knowledge base (Morgan Sindall procedures & compliance docs)
- Multilingual voice interface for multicultural site teams
- Hands-free conversational interaction optimised for live site conditions
- Contextual, scenario-specific safety guidance on demand
The challenge
Construction sites are unforgiving environments. Decisions about safety, procedure, and compliance have to be made in seconds, by people whose hands are full and whose attention is split across noise, weight, and weather. Traditional safety reference, printed manuals, induction handouts, supervisor walk-arounds, cannot keep pace with the situations that actually unfold on the deck.
Morgan Sindall, one of the UK's largest construction and regeneration groups, partnered with us to put expert-level safety guidance directly into the hands of every person on site, in a format they could actually use while they worked.
The solution
We developed a bespoke Voice AI Agent web application for Morgan Sindall to support on-site teams with real-time health and safety guidance.
The agent runs in any browser, on any device a site team already carries. Workers speak to it; it understands the question, retrieves the relevant Morgan Sindall guidance, and answers in plain language in the worker's own language, within seconds. No menus, no forms, no scrolling through PDFs. The agent meets the team where the work is happening, and turns the company's entire body of safety knowledge into a conversation.
Key features
Voice-first interface. Builders and site teams interact hands-free with the AI agent, ideal for active construction environments where stopping to type, scroll, or search is impractical and unsafe.
RAG-based safety knowledge. Integrates and indexes Morgan Sindall's health & safety guidelines, procedures, and compliance documents to provide accurate, context-aware responses grounded in the company's own materials rather than generic web answers.
Multilingual support. Workers can speak in any language, enabling clear communication across multicultural teams and reducing misunderstandings on safety-critical instructions.
Real-time safety assistance. Instant answers to safety questions, risk protocols, and compliance requirements delivered directly on site, in the moment of need, without waiting on a supervisor or office.
Contextual guidance. Provides step-by-step instructions and clarifications tailored to specific scenarios and tasks, so the guidance fits what the worker is actually doing rather than offering generic protocol summaries.
Why it matters
The deployment turns the company's H&S library into an always-on safety partner that travels with the worker. Every team member, regardless of language or experience level, gets immediate access to the same expert-level guidance, and the company gets a measurable lift in compliance, fewer near-misses, and a richer signal on the kinds of questions teams actually ask in the field.
It is a model for how AI should land in industrial environments: not as a dashboard, not as a reporting layer, but as a voice in the room that knows the company's procedures inside out and can speak them back to anyone who needs them, whenever they need them.
Explore the immersive side.
Two sides of the same vision, where it all began, and the creative arm engineering next-gen XR experiences for the world's most ambitious brands.
LIVR
LIVR transforms how organisations access, share, and use knowledge, through the unique intersection of artificial intelligence and immersive technology.
LIVR Studios
We build immersive XR experiences that move the needle from enterprise training and product innovation to flagship brand activations across virtual, augmented, and mixed reality.
Have a brief like this? Let's talk.
From discovery to live deployment, we partner with teams to ship production AI that earns its keep.
Contact us