At Buddywise, we understand that for EHS leaders, technology must deliver verifiable impact, guaranteed scalability, and absolute reliability. Our focus is on genuine partnership, moving safety from reactive compliance to proactive prevention.
In the summer of 2025, we began our collaboration with Milliken & Industrials Limited at a UK facility, the mission was clear: Support their 100-year legacy of excellence by deploying data-driven insights to create a safer, more efficient workplace.
The Challenge: Acting on the Invisible
Across large industrial environments, the common, low-severity incidents often pose the highest systemic risk. For Milliken, like many others, the core challenge was identifying these small, daily inconsistencies before they escalated into a lost-time incident.
Their EHS teams needed to target risks that are often missed during manual rounds such as:
- Inconsistent PPE use, particularly during off-peak or night shifts.
- Obstructed pathways and materials stored unsafely in vehicle lanes.
- Near-misses involving pedestrians and heavy machinery.
- Traffic Adherence: Varying adherence to traffic rules and stop signs.
Without consistent, objective monitoring, EHS teams are playing catch-up, reacting to incidents rather than predicting and proactively preventing them. This reactive posture makes cultural change nearly impossible.
As a Milliken safety leader put it, the goal was deeper than just rules:
Customer quote
“We’ve got to drive a cultural change and understand why people do certain things. If we can get that bit right, there will be fewer of these risky situations.”
Milliken safety leader
The Solution: Reliable Data to Define the ‘Why’
We believe that true cultural change in safety requires data to pinpoint the root cause - the ‘why’ behind behaviour. Buddywise provides that data, facilitating a culture of continuous EHS improvement.
Our approach focuses on speed, AI rigor, and IT adaptability, ensuring impactful results that can be repeated across multiple sites.
A Rapid, Focused Deployment
Our deployment model is designed for multi-site repeatability, ensuring that initial success quickly scales across a customer’s site portfolio. Within 4 weeks from having an agreement in place we had connectivity established via our on-premise Buddybox solution and core risk scenarios identified.
Within the first week of going live with users, the Milliken team was already acting on new insights captured through the solution (e.g., people not following PPE policy during night shifts). From going live to end the of the initial 8 week period, new risk models were added and even developed, and new routines integrated onsite.
AI Rigor: Earning Trust Through Accuracy
In the world of computer vision False Positive Rate (FPR) is a common technological challenge to work with during the first we weeks in order to ensure high model accuracy. To cut down on false positive detections, our AI strategy is built on a dual-layer approach:
- We use robust generalized AI models for foundational accuracy in order to move fast.
- We run continuous, site-specific training cycles during and after onboarding. This process is explicitly designed to reduce the False Positive Rate (FPR), ensuring that detected risk observations are highly targeted and trusted by the site teams. Our goal is to achieve an extremely low FPR across all risk scenarios, maintaining the integrity of the data and deepening adoption.
Deep Customisation Fit For Enterprise Needs
Safety isn't one-size-fits-all, a generic solution won't work in a complex manufacturing environment. We adapted our AI models to Milliken’s specific operations in real-time, proving our agility to meet enterprise demands:
- Model Refinement: Adapted our models to include 7 new vehicle types (beam trucks, set trucks, etc.) and even a specific pink protective vest colour used on site.
- Operational Exclusions: Adjusted detections to correctly exclude operators of handheld vehicles from being categorised as pedestrians, ensuring rules adherence remained accurate to specific policies for that unique site.
- New Risk Launch: Quickly developed and launched a new risk model to detect vehicles missing designated stop signs - an insight that uncovered an unexpected high-risk behaviour at pedestrian crossings.
Seamless IT & Data Integration
Reducing friction in IT and enhancing data workflows is key. For Milliken we delivered:
- Single Sign-On (SSO): Ensuring streamlined and secure login experience from day one.
- Data Synchronisation: Enabled automated import of all EHS risk data directly from the Buddywise system into Milliken’s own PowerBI environment, empowering their internal data analysis teams.
- Live Visibility: Built an API to display safety data on the plant floor, providing real-time feedback that drives immediate behavioural change.
The Results: A New Benchmark for Proactive Safety
Our aim is to enable well-informed decisions and actions that sustain behavioural improvements. By integrating granular data into existing EHS governance, we can ensure that the cultural shift lasts while providing concrete business value such as:
- 73% Reduction in Total Risk: The total number of observed risks at the site dropped by 73% over the course of just 4 weeks.
- High-Vis Vest Compliance: Improved by 94% over the course of 8 weeks through safety talks and collaboration with Production Management and Team Leads
- Vehicle Safety: Near-hit scenarios involving forklifts reduced by 85% within 8 weeks.
- Traffic Discipline: Missed vehicle stops (running stop signs) decreased by 92% within 8 weeks. The onsite HSE teams targeted efforts, combining data with driver briefings and live screen data - reduced stop line misses by over a third in a single week.
- Housekeeping: An 89% reduction in unattended objects left in operational lanes.
- Pedestrian Safety: Vehicles entering pedestrian zones were fully stopped, with observations dropping to zero thanks to a joint effort in data analysis, safety risk review and action taking (Placement of a new physical barrier)
The success from just the initial weeks is best captured by the users themselves:
Customer quote
”I didn't realise half the problems that we had at the site, before we had this system. You assume what it's like, but then the reality can look quite different. The AI is teaching us a lot of things, and clearly capturing information that we want to see. That it’s that simple, is fantastic.”
David Bradley, Corporate EHS Manager Europe
The Takeaway: Building Sustainable Safety Excellence Pays for itself
This collaboration demonstrates that rapid deployment, deep customisation, customer-centric development, and clear data integration are not trade-offs - they are the standard for enterprise-grade EHS solutions. We don't just hand over a tool; we integrate continuous monitoring and scheduled report delivery directly into your EHS ways of work to ensure the improvements last.
For Milliken, integrating Computer Vision has acted as a support layer for their EHS teams, helping to validate their safety investments and mitigate the potential costs associated with workplace accidents. Following the success at the UK site, the system serves as a model for expansion and long term collaboration. We are currently evaluating opportunities across Milliken’s global operations throughout 2026 and beyond.
If your organisation is ready to move beyond reactive compliance and use trusted AI to drive sustainable cultural change, let's discuss how we can deliver these repeatable results across your sites.
Let's talk
Book a meeting with us for a demonstration of our AI-powered safety platform.
Book a meeting
