Step 1: Spot the early warning signs

We identified key indicators of potential errors - those moments when problems are just waiting to happen. Here’s the list we used, based on real-world experience:

• Over-complicated SOP's: more than 8 pages? You’re asking for trouble.
• Manual, human-reliant processes: everyone has a bad day eventually.
• High multitasking demands: when you multitask you make mistakes.
• Frequent procedural changes and retraining: prevents habit formation
• Batch Manufacturing Records (BMR): “Right first time” below 90%
• Staff turnover exceeding 20%: high ‘churn = high error rates
• Contract workers exceeding 5%: not always, but often
• Manufacturing RFT (Right First Time) less than 80%: enough said
• On-time delivery below 80%
• Supervisors unavailable >40% of the time: encourages bad habits to remain
• Equipment utilisation <70%: causes disruption and mistakes
• Highly customised processes (lack of standardisation): leads to confusion.
• Repeat deviations >60%
• ‘Check by’ culture: Too many signatures that dilute accountability so no one really cares.

Step 2: Prioritise the Risks

Not all indicators carry the same weight. Each characteristic was assigned a priority score (out of 10) based on its impact on error likelihood. This made it clear where to focus our energy.

Step 3: Ask AI (your co-worker) for help

Here’s where the magic happens. We used a Large Manufacturing Model (LMM), a predictive AI system, to answer this specific question:

“Which manufacturing area is most at risk of human error? Use the weighted data to identify where preventive action will have the biggest impact.”

The AI analysed our data, designed an algorithm and flagged areas where errors were likely to emerge. These insights gave us a precise starting point for intervention and prevention.

Step 4: Validate and act

No AI system is perfect without human expertise. We validated the AI’s findings, refined the algorithm and pinpointed an area ripe for improvement. The results were staggering.

A four-person task force trained in our six step simplification process spent just five days addressing the top three contributing factors:

• Overly complex documents
• Excessive multitasking
• High levels of procedural change

Step 5: Celebrate success and scale up

The outcome? A 4,000-hour reduction in resource use and a leap in on-time delivery from 72% to 98%. By prioritising prevention over reactive investigation we achieved more with less.

What can you do?

Here’s how you can take the first step toward transforming your human error prevention approach:

Brainstorm contributing factors: What signs indicate errors waiting to happen in your processes? Engage as many people as possible. Would you add anything to my list?

Leverage AI wisely: AI is a powerful co-worker, but the quality of your results depends on how you frame the problem. Let me know if you need help with 'prompting'.

Validate AI outputs: Never take predictions at face value. Refine and test the results. LLMs can lie. Critically think about the outputs. Never blindly accept.

Focus on simplification: Even small changes in process and SOP design can yield massive results.

Celebrate wins: Prevention saves time, resources, and stress. Every win is worth a party