“We’d already driven huge gains manually - but eventually hit a wall. Almetra showed us what we couldn’t see and helped us increase output again.”

Filip Marjanović
PROJECT MANAGER

THE CHALLENGE

After various improvement workshops - eight kaizens, tighter 5S, and optimized flow - the line had reached one of the factory’s highest efficiency levels. It was running so well that new improvement opportunities were hard to find, yet cost pressure kept rising as wages grew, absenteeism increased, and skilled operators became harder to hire.

To break through the plateau, the team explored whether AI could help them find what traditional CI had missed.

WHY THE OLD WAY WASN’T ENOUGH

On paper, the team was doing everything right - following proven CI methods. But were they really seeing the whole picture?

MES in place

Tracked output, not process - the small inefficiencies stayed invisible.

Manual Stop Logging

Depended on human input - brief, frequent stops often went unrecorded.

Quality and scrap reports

Showed when defects appeared, but not where they came from. Missed steps stayed invisible.

Clean and consistent-looking data, but hidden inefficiencies that quietly drained performance.

The team wasn’t out of improvements - they were out of visibility.

ALMETRA’S APPROACH

Within weeks, Almetra connected all 9 stations on the manual pump assembly line with cameras and AI.

The system continuously tracked how operators, tools, and materials moved, revealing what was really happening inside each cycle.

How the Team Used AI to See More

Detected True Causes of Long Cycles

After one month, Almetra flagged the longest cycle times and ran Non-Value-Added (NVA) analysis to explain why. Hidden losses were transformed into visible, actionable root causes.

Identified Downtime & Idle Pockets

Activity-time and presence views detected idle pockets when stations lacked either an operator or a product. This exposed recurring issues such as carts occasionally locking, extra motion, and materials not always within reach.

Standardized Best Operator Practices

Almetra identified the fastest repeatable cycles and surfaced click-to-video best practices. It also revealed where assembly steps were being skipped - the source of several quality issues.

Key Actions

• Rebalanced work and optimized the layout

• Simplified material handling and changeovers

• Standardized packaging, labeling, and tools

• Digitized tracking and error-proofing with AI checks and live feedback

THE RESULTS

11% throughput increase

Achieved within 3 months through focused waste elimination and bottleneck stabilization.

82% All-time high OPR

For the first time since launch, the line’s OPR exceeded 80%.

Manual data collection eliminated

Teams no longer spend hours doing time studies - AI now captures everything automatically, saving time and improving accuracy.

14% labor cost savings

Grundfos was able to maintain the same output while using 14% fewer labour hours, allowing the team to reallocate operators to other lines where additional capacity was needed.

Ready to see what AI could reveal on your line?