Transforming Workplace Safety with AI: The Real Experience at Blackmores Group

Artificial intelligence is increasingly finding its way into safety management, promising to reduce administrative burdens and improve risk control. But what does this look like in practice for a real company? In a recent webinar titled “Transforming Workplace Safety with AI: The Real Experience,” two safety leaders from Blackmores Group shared how they are leveraging AI to enhance their workplace safety programs. Andrew Wilson, Director of Work Health & Safety (WHS) at Blackmores Group, and Udd Lalich, Head of WHS at Blackmores Group, discussed their journey implementing an AI-driven safety platform (SoterAI) and the lessons they learned along the way. Below, we distill key insights from their conversation – from the initial need for AI, through challenges and integration, to the benefits, results, and future vision – all in an educational context for EHS managers.

Recognizing the Need for AI in Safety Management

Blackmores Group – a leading natural health products company with 1,200 employees across Asia-Pacific – found itself grappling with an increasingly complex safety landscape. According to Andrew Wilson, constant updates to standards and regulations were stretching his team thin: “Everything keeps changing. Standards, regulations, there’s a constant update. There’s a huge amount of lost admin time where you’re trying to do these changes and checks and balances”. The WHS team was spending countless hours manually keeping policies up-to-date and ensuring compliance across multiple jurisdictions.

Wilson noted that many existing tools addressed bits and pieces of the problem but “they don’t offer a full package” – often requiring significant manual administration or technical expertise to fill the gaps. This fragmented approach was no longer sustainable. Blackmores needed a smarter, more efficient way to manage risk and compliance at scale.

That search led them to consider an AI-driven solution. Soter’s platform, which boasts features like real-time analysis of text, images, and video, promised to “provide industry-proven corrective actions leading to a safer and more efficient workplace”. In fact, the platform claimed 99.9% accuracy in risk detection when supplied with quality data. or Wilson, adopting AI aligned perfectly with the old adage of having to “work smarter, not harder” to keep up with the pace of change. Embracing an AI tool was not just about technology for its own sake – it was a strategic decision to maintain due diligence in a fast-evolving regulatory environment while easing the administrative load on his team.

Overcoming Initial Hesitation and Challenges

Implementing AI in a safety context wasn’t without its challenges. Andrew Wilson candidly shared that he was initially skeptical about bringing AI into his field, even wondering, “Where does it end? Is this going to do us out of a job?”. This hesitation is a common concern among safety professionals who fear that intelligent systems might replace human roles. However, Wilson overcame this fear through experience, realizing that “you are always going to need skilled operators to utilise this type of equipment”. In other words, AI would be a tool to augment the safety team, not replace it. Blackmores proceeded cautiously – after first seeing the AI platform demonstrated at an industry event, the team spent 6–8 months in discussion and evaluation before feeling comfortable enough to initiate a trial. This patience ensured that by the time they adopted the tool, it had matured and everyone was on board with its potential value.

Another early hurdle was ensuring trust and accuracy. “With any kind of system, you’ve got to go through all the checks and balances,” Wilson noted, highlighting concerns around data security and reliability. Blackmores’ IT and security teams vetted the AI platform for cyber risks, as is standard practice these days. The WHS team also rigorously double-checked the AI’s outputs against known regulations and internal standards during the rollout. This validation step was critical to convince stakeholders that the information and recommendations coming from the AI could be trusted. “You’ve got to convince the business that the information we are receiving is accurate and correct,” Wilson explained – something that turned out to be “quite easy to verify” by cross-checking results, which helped build confidence in the tool.

Integration with existing processes was (and continues to be) an ongoing challenge. At first, the Soter AI system was used as a standalone tool alongside Blackmores’ traditional safety management system. According to Wilson, “we’re still integrating [SoterAI] – we probably use it as a standalone process at this point, but we are also trying to build it into more and more of our tools”. This meant workflows had to be adjusted to incorporate AI outputs manually into reports and action plans. Over time, as familiarity grew, the team started linking the AI with other systems (for example, feeding its outputs into their risk register and incident reporting workflows). While full integration is a work in progress, the interim approach allowed Blackmores to start reaping benefits without waiting for a perfect system tie-in.

From Manual Tasks to Data-Driven Insights: AI vs. Traditional Safety

Both safety leaders stressed how dramatically the AI platform has improved efficiency compared to traditional safety management methods. Judd Lalich contrasted the new approach with the old ways many EHS professionals know too well: “Traditional safety [work] is very manual, [and] admin-heavy,” he said. In the past, safety advisors would spend a lot of time “running around doing lots of admin and checking stuff,” compiling observations, filling out paperwork, and manually analyzing data. By the time a trend or hazard was identified, a lot of effort and hours had been expended.

Lalich highlighted that AI has flipped that script. “The beauty of the tool is that you can literally take a photo or a video and it will do the analysis for you,” he explained. Instead of an advisor laboriously measuring angles in an ergonomic assessment or researching a regulation change, the AI quickly provides quantitative feedback and insights. This output then “feeds into your risk profiling or risk assessment process”, making their overall workflow far more streamlined. In short, Blackmores’ safety team is now able to work smarter, not harder – focusing their expertise where it’s most needed, while letting the AI handle data crunching and information retrieval in seconds.

By using AI, Blackmores has also reduced ambiguity in risk assessments. Lalich noted that the system’s analysis and visual reports have given the team greater confidence in understanding and communicating risks: “It’s taking away maybe some ambiguity or vagueness with respect to risk,” he said, especially in areas like ergonomics. The platform can generate color-coded, stick-figure visuals that clearly highlight high-risk postures or movements in a task. The team can “clearly see what the risk is” in these images, which makes it “extremely easy to share” the issues with others – from the workers on the floor all the way up to management. This kind of instant, graphic insight was rarely available with traditional methods. It not only speeds up hazard identification, but also helps in making the business case for interventions by providing hard evidence of the risk (e.g. heat maps of stress on a worker’s body or traffic-light style risk ratings).

Lalich gave an example that many EHS managers can relate to: justifying ergonomic improvements. In the past, it might have been challenging to get leadership to fund an adjustable work table or lifting aid without a clear quantification of the risk reduction. Now, using AI analysis, the Blackmores team can obtain objective data and visuals to support such investments. “This tool really does help give some traffic lights,” Lalich said, referring to the risk color-coding that makes the level of hazard immediately apparent. The ability to translate safety issues into easy-to-grasp visuals and metrics has been a game changer in getting buy-in for safety improvements.

Real-Life Example: AI Enhancing an Ergonomic Assessment

To illustrate how AI has transformed their approach, Judd Lalich shared a compelling real-world example from Blackmores’ operations. At one of their sites, workers were manually stacking products on pallets on the floor – a task that involves frequent bending and lifting. An adjustable lift table was available (which could raise pallets to waist height and reduce bending), but there was resistance from some workers to using it. “They sort of felt it was in the way, a bit of a pain,” Lalich explained, and as a result, the table often went unused. One worker had even started complaining of a sore back, indicating that the manual process was taking a toll.

Seeing this, Lalich thought, “this is a fantastic opportunity to use the tool” and gather evidence on the difference the lift table could make. He asked the worker to perform the task exactly as usual (building a pallet on the floor) while being filmed, and then to do a second run of the same task using the adjustable table. Importantly, Lalich did not coach the worker on better lifting techniques or posture – he wanted to capture a before-and-after comparison of normal behavior versus the engineered solution. The videos of both scenarios were then uploaded to Soter AI for analysis.

The results were illuminating. Andrew Wilson noted that the worker was not wearing any sensors or wearables; all the data came purely from video, and the person’s face was automatically blurred for privacy. The AI generated stick-figure models of the worker for each task, with joints and body parts color-coded by risk level (green, yellow, or red depending on the strain). In the first video (no lift table, lots of bending), there was a significant amount of red – particularly on the lower back – indicating high-risk postures. In the second video (using the lift table), the figure showed much less red, reflecting a safer posture. “Without going into the numbers too much, you can just see clearly that there’s more red risk on the right-hand side than the left,” Lalich described, referencing the side-by-side comparison of the two scenarios. In fact, even the “improved” scenario revealed a tweak – the worker was reaching too far across the table at one point, which the AI flagged as a remaining risk that could be fixed by simply repositioning the pallet or the person’s stance.

The immediate impact on the worker and supervisors was perhaps the most rewarding part. “The workers that we’ve used this on… they get it straight away,” Lalich said. When shown the video feedback, the worker was fascinated to see his own movements analyzed. It sparked an “aha” moment – “I didn’t realize that a minor change… in shoulder flexion could actually change that risk profile,” the worker remarked upon seeing how different body positioning reduced the red areas on the chart. This kind of instant insight is hard to achieve with a traditional safety audit or a verbal coaching session. The visual proof not only convinced the once-reluctant worker to use the lift table, but it also equipped the safety team with quantifiable evidence to reinforce the importance of ergonomic controls to management.

By leveraging AI in this way, Blackmores turned what could have been an anecdotal safety recommendation (“Bob should use the lift table because it’s better for his back”) into a data-driven case study. It resolved a local issue (back strain complaints) and helped build a broader culture of acceptance for both new equipment and new technology on the floor. As Lalich put it, the example showed how AI can “make life much easier” by analyzing a scenario and clearly demonstrating the risk reduction, which “help[s] you share issues with not just the workers themselves, but actually management”.

And all of this was achieved without any invasive monitoring – just a simple video upload. Andrew Wilson added that the AI’s ability to analyze video with no wearables and still ensure privacy (via face pixelation) is a major plus for worker acceptance. People are more comfortable participating when they know the focus is on the task and not on identifying individuals. In summary, this ergonomic assessment case is a powerful demonstration of AI’s practical value: it not only identified and quantified a workplace risk, but also helped change minds and behaviors by presenting the findings in a compelling way.

Gaining Organization-Wide Buy-In for AI

After early wins in areas like ergonomics, the Blackmores safety team sought to expand the use of AI to other risk domains – but doing so required broad organizational buy-in. Introducing any new technology in an established company culture can meet resistance, so the team took proactive steps to educate and involve stakeholders. Judd Lalich recounted how Andrew Wilson organized a special due diligence session for Blackmores’ leadership, aimed at highlighting the company’s safety obligations and how AI could help meet them. This session “educated a lot of people in our business to understand exactly who’s on the line for what” in terms of legal responsibility for safety. In a candid and eye-opening way, Wilson walked the executives through the rising standards of enforcement – “the regulators are whacking out much higher fines now than they ever were”, he shared – and the potential gaps that could leave the company vulnerable. This reality check grabbed management’s attention. It wasn’t just about selling a new tech tool; it was about ensuring Blackmores stays ahead of compliance and avoids costly incidents or penalties.

According to Lalich, this approach worked. Seeing how the AI could proactively identify hazards and compliance issues made the case that it was a worthwhile investment. The result was “some real buy-in now” from senior leadership for using the platform in areas beyond ergonomics. For instance, the team is now applying the AI’s hazard identification capabilities to things like machinery and plant safety. One of Blackmores’ ongoing projects is using AI to assess risks in their conveyor systems and other equipment – an initiative that gained momentum only after management understood the serious implications of missing hazards. In Lalich’s words, they had to “pick our battles” and win one area at a time; “we’ve won the battle with the ergonomics, and the next battle” on machine safety was made easier by that initial success and leadership support.

With top-level endorsement in place, Andrew Wilson and his team are now focused on scaling AI usage across the organisation. Currently, the core WHS team (four people) has been the primary user of SoterAI, but the vision is to empower many more employees with the tool. “We want our supervisors, leading hands, frontline managers… [using it], so that way we’ll actually be able to get more and more usage out of it,” said Wilson. By training line supervisors and team leads to perform basic hazard analyses with the AI, Blackmores hopes to push safety ownership to the frontline. This also ties into their proactive safety culture: for example, WHS committee members can use AI during their monthly inspections, rather than just relying on checklists. When more people throughout the company engage with the tool, it reinforces a message that safety is everyone’s responsibility – and that everyone now has access to a powerful assistant to help them spot and address risks.

Wilson noted that as the comfort level with the technology has grown, the AI is becoming an integral part of day-to-day operations. What started as a trial is now something the team doesn’t want to work without. The ultimate goal is to embed AI into all relevant safety processes, so much so that it just becomes a seamless part of “how we do safety” at Blackmores. Getting to that point requires continued advocacy and training, but the enthusiasm is there: after seeing tangible improvements, even some initial skeptics have become champions. In short, securing organization-wide buy-in for AI came down to showing real value (through pilot successes and data), educating stakeholders about why it’s needed (due diligence and risk exposure), and then actively involving those stakeholders in the journey so they feel ownership of the solution.

Streamlining Compliance and Safety Documentation

Another significant benefit Blackmores has reaped from AI is in the realm of compliance and document management. Operating across different regions, Blackmores must comply with various regulatory regimes – from local Australian WHS laws and codes of practice, to international standards and guidelines (they align their safety management system with ISO 45001 for occupational health and safety, and need to consider SafeWork Australia requirements for other markets). Keeping all their procedures and policies up to date with the latest regulations was once a massive undertaking. Now, Andrew Wilson is leveraging AI to make this task far more efficient.

“We’re currently reviewing our WHS management system and we’re utilizing the AI to incorporate in our international markets and make certain we’re hitting the SafeWork Australia requirements,” Wilson explained.

The team can upload their safety documents (policies, procedures, risk assessments, etc.) into SoterAI, which then analyzes the text against a vast database of regulations and standards. The AI highlights any gaps or differences – essentially answering, “If we follow our Australian-based procedure in another country, are we missing anything?” This automated cross-check means the team doesn’t have to manually scour foreign regulations line-by-line. Wilson expects “a massive administration benefit from the review via the uploading tools”. In practice, this could save countless hours that would otherwise be spent in compliance meetings or consulting legal experts for each region. It also gives the company confidence that nothing is falling through the cracks as they expand globally or as laws change. In one example, they found that uploading their documents helped target specific items that might have changed since the documents were originally written, prompting updates that keep their safety management system actual.

Beyond checking existing documents, the AI platform’s research and Q&A capabilities have become an everyday compliance aid. Judd Lalich mentioned that if his team needs to quickly clarify a regulatory point, they can simply ask the AI’s chat feature instead of flipping through pages of legislation. “Rather than us crawling through the regulations, we just ask a question and it says, ‘This is what’s in the regulation… and here’s the link to it.’ It makes life so much easier,” Lalich said. For example, if a new safety alert comes out or a standard gets updated, they can query the AI about what the change is and how it impacts their operations. The system will not only summarize the key information but often provide the direct reference to the clause or guidance document, saving the team time and ensuring accuracy. This use of AI as a “virtual safety librarian” has streamlined their compliance workflow: answers that might have taken days of research are now retrieved in minutes, complete with citations.

Importantly, the Blackmores team still validates critical information – the AI makes it faster to find and draft material, but human experts review and approve it. Wilson and Lalich both emphasize that SoterAI is there to support, not to make decisions in isolation. In the case of compliance, this means an experienced WHS professional will always verify that an AI-suggested policy addition is appropriate and correctly interpreted. That said, by handling the heavy lifting of document analysis and regulatory comparison, the AI has freed those professionals to focus on the higher-level strategy (like deciding how to implement a new requirement, rather than spending all their time discovering it). The platform has effectively become an extra team member that never tires of reading legislation – a welcome addition, indeed.

Automating Training Content and Saving Time

Workplace safety isn’t just about identifying risks; it’s also about educating employees and reinforcing safe practices every day. Blackmores has a robust program of toolbox talks and safety training, and here too AI has made a significant impact. Andrew Wilson described how they use Soter AI to automate the creation of safety communication materials, drastically reducing the time his team spends preparing these resources.

For context, Blackmores runs regular toolbox talks at the start of shifts – brief discussions on specific safety topics (often prompted by recent incidents, near-misses, or seasonal risks). Previously, the WHS team would pick a relevant topic each week or month, research the best practices, and manually write up a summary or presentation for supervisors to deliver. This process could take many hours for each topic, especially if the team had to gather information from various sources. Now, Wilson’s team takes advantage of the AI’s capabilities to do most of this legwork. “We started to use SoterAI for our risk assessment checks of tasks and processes, and then asking the tool to create training material and toolbox talks,” he explained. In practice, they might input the particulars of an incident or a hazard into the AI and prompt it for a toolbox talk script or safety brief. Because the AI already has context from their documents and industry best-practice content, it can produce a solid first draft almost instantly.

The efficiency gains have been remarkable. “95% of the work’s done by SoterAI. We do the screening and proof check,” said Wilson. In other words, the AI generates the bulk of the content, and the safety team members simply review it, make any necessary tweaks (for example, to adjust wording to Blackmores’ preferred terminology or to add site-specific details – a process Wilson jokingly calls “Blackmores-izing” the output), and then approve it for use. What used to take perhaps half a day of someone’s time can now be accomplished in a matter of minutes plus a quick review. Wilson noted that the AI’s assistance has “taken hours and hours of administrative tasks away” from his team, “and it’s enabled us then to become more proactive”.

Being more proactive means the team can plan their communications months ahead. Wilson shared that they maintain a safety communication calendar, and they’ve been able to prepare toolbox talk content a full quarter in advance. “Second quarter is already done [unless there’s a change]. And that was all done through SoterAI,” he said, referring to how they generated and scheduled topics for the upcoming months. If trends or risks change, they can always adjust on the fly, but otherwise the groundwork is laid. This was nearly impossible before, when writing each talk was a manual effort that often happened last-minute. Now, with the routine content creation largely handled by AI, the WHS team can dedicate more attention to analyzing incident trends or engaging with workers directly – the aspects of the job that truly require human insight and empathy.

The automated content is also consistent and comprehensive, as it draws on a wide knowledge base. Wilson noted that the AI often includes reminders or steps that a human might forget when rushing. In fact, the Soter AI tool has a feature where it can prompt the user, asking “Have you considered X? Would you like to include Y?” during the content creation process. This helps ensure that important points (like a regulatory requirement or a preventive measure) aren’t accidentally overlooked in a training piece. For Blackmores, having this safety net means the quality of their toolbox talks and training materials has been maintained or even improved, even though they spend far less time producing them. It’s a perfect example of AI delivering on the promise of doing the heavy lifting, so the experts can focus on refining and delivering the message.

From an ROI perspective, these time savings and productivity improvements are extremely tangible. The safety team can now handle a larger scope of work without needing additional headcount, effectively “lightening the load,” as Wilson put it. “Yeah, the return on investment’s there,” he affirmed simply. By automating routine tasks like document drafting and information gathering, the AI has paid for itself in the time reclaimed – time which is now spent on higher-value safety activities that further protect Blackmores’ employees and assets.

Key Takeaways for EHS Managers

For EHS professionals evaluating AI for their own organizations, Blackmores Group’s real-world experience offers several valuable lessons:

Embrace new tools to manage complexity: Traditional safety processes can become unsustainably time-consuming as regulatory and business complexity grows. AI solutions offer a way to handle large volumes of information and repetitive analysis more efficiently, allowing you to work smarter, not harder in managing risk and compliance. It’s increasingly important to explore these tools to keep up with constant changes without burning out your team.
Start with a clear need and quick wins: Identify specific pain points or use cases where AI can immediately add value (e.g. analyzing ergonomic risks, automating document checks). Blackmores began with areas like ergonomics where results were tangible and visible. Successfully addressing a known problem with AI will build momentum and support for broader adoption. As Lalich noted, they “picked their battles” – winning one domain at a time – which helped gain wider buy-in.
Address skepticism through involvement and education: It’s natural for team members (and leaders) to be skeptical of AI at first. Involve them in demos and trials to show what the technology can and cannot do. Andrew Wilson openly discussed his initial hesitation and how hands-on experience overcame it. Provide training and reassurance that AI is a tool to assist them, not replace them. Also educate senior management about how AI can help meet due diligence – for Blackmores, a dedicated session highlighting legal responsibilities and how the AI mitigates risk was pivotal in securing leadership support.
Ensure data quality and verify outputs: AI is powerful, but it’s not magic – the old saying “garbage in, garbage out” applies. To get reliable results, feed the system good data (clear images, accurate and up-to-date documents) and ask clear questions. The Blackmores team learned to refine their queries and double-check the AI’s suggestions against regulatory sources. This not only builds trust in the AI’s outputs but also catches any anomalies. Maintaining a human-in-the-loop for verification is especially important in the early stages of use or for critical decisions.
Leverage AI for efficiency, but keep humans in control: Automate what is tedious or data-heavy – like searching regulations, generating reports, or analyzing video – so your safety experts can focus on interpretation and action. Blackmores saved significant time by having AI draft toolbox talks and analyse hazard videos, then humans reviewed and implemented the findings. This human-AI collaboration yields better outcomes than either would alone. Always have professionals review AI-generated content for accuracy and context before final use.
Use data and visuals to engage the workforce: One of the powerful aspects of AI is the ability to turn raw data into digestible insights (charts, risk scores, annotated images). Use this to your advantage when communicating with employees and executives. In Blackmores’ case, showing a worker a color-coded stick figure of his posture instantly triggered corrective action, and showing executives a dashboard of compliance gaps grabbed their attention. Visual evidence can make safety issues more concrete and motivate action more effectively than abstract descriptions.
Plan for integration and scalability: Think about how an AI tool will fit into your broader safety management system. It’s okay to start with a standalone pilot, but have a roadmap for integration into existing processes (incident reporting, audits, etc.) as you prove its value. Also, widen the circle of users over time – train supervisors and other departments to use the tool, not just the safety team. Blackmores is actively moving toward a more integrated system and expanding access to the AI across their operations. This ensures AI isn’t a siloed experiment but a core part of the safety culture.
Keep learning and stay adaptable: The technology will evolve quickly, with new features and improvements rolled out frequently. Be prepared to update your procedures and training as the AI platform grows in capabilities. Stay engaged with the solution provider for updates (as Lalich was advised, “stay tuned” for new developments). Internally, encourage your team to experiment with the tool’s functions – you might discover additional uses. Continuously measure the impact, gather feedback from users, and refine how you use AI to ensure it delivers on your safety objectives.

Blackmores Group’s foray into AI-driven safety management demonstrates that, when implemented thoughtfully, AI can dramatically enhance workplace safety programs. It can take over the drudgery of data processing, provide sharper insights into risks, and even help communicate those risks more effectively – all of which empowers safety professionals to focus on strategic prevention and worker engagement. As Andrew Wilson and Judd Lalich have shown, the keys to success are being clear about why you need AI, taking your people along on the journey, and continuing to adapt as you learn. For EHS managers everywhere, their experience serves as a real-world case study of AI’s transformative potential: it’s not about fancy algorithms, but about making the workplace safer and healthier in practical, tangible ways. By combining the strengths of humans and AI, we truly can transform “work harder” into work smarter – and in the process, protect our workers like never before.