Accidents in factories often occur because workers are unsure of procedures or lack the right training. New employees, such as sanitation staff or machine operators, are especially vulnerable when they want to move quickly but do not yet have the knowledge to work safely.
Quebec-based technology company Poka set out to address this challenge by creating a digital platform that replaces paper manuals and static whiteboards with instant, accessible knowledge.
The idea, first sparked by Alex Leclerc’s experience in his family’s factories and developed with co-founder Antoine Bisson, was to build a kind of “YouTube and Facebook for factories.”
Today, Poka’s connected worker platform is used by more than 400,000 people across 1,500 factories in over 70 countries, helping frontline teams at companies including Nestlé and Bosch improve compliance, safety and efficiency.
FSJA Editor Iain Hoey sat down with Alex Leclerc, CEO of Poka, to discuss connected workers and industrial safety.
Our technology is about making sure people have access to everything they need to do their job. With Poka, that starts with content, training and skill validation, but it doesn’t stop there. The next step is: are they actually applying it? That’s where our forms, checklists and audit systems come in.
Take lockout/tagout as an example. First, workers go through a generic training module in Poka that explains all the core concepts.
Then, when they’re at a specific machine, they can pull up the detailed work instruction tailored to that equipment.
After completing the task, they fill out a lockout/tagout permit directly in Poka. If they spot an issue or want to notify a supervisor, they can log it on the spot and it’s instantly directed to the right person.
Supervisors and managers can see the full picture on a dashboard: are people trained, are they accessing the right content and are they completing the necessary permits?
That closes the loop, because training alone isn’t enough – we need to know it’s being put into practice.
What systems or processes ensure that content created in Poka is accurate and reliable?
We decentralize content creation so anyone can record or document a process, but before it goes live it goes through a centralized approval workflow.
Alex Leclerc
For example, you might set a rule that anything created by an operator has to be reviewed by an operations manager, a safety manager and a quality manager. Only once all three have approved does it get published.
That way everyone can contribute their expertise, but the organization still has the assurance that the content is accurate and safe.
It avoids the old problem of everything being controlled by one person or one department, while also making sure you don’t end up with off-topic content in the library.
AI supports several parts of the process. One of the biggest is handling large amounts of existing content. Many companies have lockout/tagout procedures or work instructions in Word or Excel, often with a mix of text and images.
Our AI engine can automatically convert that into Poka’s format, making it easier to use. It can then translate the content into more than 80 languages, so the same knowledge is instantly available across different sites.
Alex Leclerc
Another application is video analysis. You can film an employee performing a task without sound and the system will break it down step by step. It recognizes the PPE – hair net, safety glasses, gloves – and then describes each action, like taking a screwdriver and loosening a screw.
From that, it generates detailed work instructions. We can add subtitles or even create synthetic voiceovers, so a video recorded in Chinese can be shared in English with an automated translation and voice track.
We also built the Poka assistant. A worker can ask, “How do I perform lockout/tagout on this machine?” and receive an answer in real time, similar to what you see with ChatGPT.
Crucially, the system always references the original source material, so the worker can double-check and ensure what they are doing is accurate.
Video makes a huge difference compared to traditional written procedures. With video you can see moving parts and during the approval process reviewers might notice something unsafe, like equipment still spinning while instructions are telling an employee to go in.
That immediately raises awareness of hazards that a written procedure would never highlight.
Safety is also about culture. You have to reinforce it constantly, almost like rewiring how people think about their work. Poka’s factory feed works a bit like Instagram.
Alex Leclerc
Workers can post updates about issues, opportunities, new employees, or good practices, but they can also post safety observations and near misses. That keeps safety top of mind every day and drives cultural change.
It also creates accountability. If you see an unsafe situation, you log it as an issue in Poka. It then moves into a Kanban board where it can be properly managed.
Customers using Poka have reduced accident rates by encouraging everyone to post near misses, by running safety behavior audits through the platform and by holding people accountable for following safe practices.
I think the rise of AI agents to handle repetitive tasks is inevitable. Factories will likely have fewer people overall, but the people who remain will take on more complex roles.
Human judgment will still be essential – AI can analyse data and make suggestions, but it won’t replace someone physically changing a bearing on a machine.
The real change will come from how machines and data connect. In the past, factories collected huge amounts of machine data but used it only for basic graphs. AI can now make that data contextual and actionable, giving operators clearer insight into what they should do and where issues might arise.
Language will be less of a barrier too. With AI translation, workers in different countries will be able to communicate seamlessly.
Alex Leclerc
For us, the direction is toward more predictive and preventive approaches. Instead of reacting to problems, we’ll use machine data to anticipate them and keep ahead of potential failures.
I also think we’ll see manufacturing execution systems and connected worker platforms come together, creating a single environment where machines, people and processes are fully integrated to support both productivity and safety.
