A leading escalator manufacturer assembles complex multi-component mechanical systems on a multi-station production line — each unit requiring dozens of sequential steps performed in exact order, with the right parts, by trained operators. When errors reached finished product, the cost was not just rework — it was warranty claims, field failures, and reputational risk in a safety-critical product category. Intelgic built a Process Intelligence system that watches every step, on every station, on every shift.
The Challenge
Escalator assembly is a high-complexity manufacturing process. Each unit involves hundreds of individual components — drive mechanisms, step chains, handrail systems, safety devices, electrical assemblies — assembled across multiple dedicated stations in a precise sequence. The consequences of an error at any stage cascade through the entire assembly: a missed component, an incorrect part selection, or a step completed out of sequence can compromise the mechanical integrity, safety certification, or operational performance of the finished unit.
The manufacturer's quality process relied on end-of-line inspection and periodic supervisor audits to catch assembly errors. This approach had a fundamental flaw: errors were discovered after they were made — often after they were built over. Rework at a late assembly stage is exponentially more expensive than preventing the error when it occurs. And because no continuous record of the assembly process existed, tracing the root cause of a field issue back to a specific step or operator was nearly impossible.
- Assembly errors — wrong parts, missing components, incorrect sequences — only caught at end-of-line or in the field
- No real-time verification that operators were following work instructions step-by-step
- Multiple operators across multiple shifts, each with different experience levels and working habits
- No per-unit assembly record — no traceability from field failure back to assembly step or operator
- Throughput and efficiency data collected manually — slow, inaccurate, and always retrospective
- Escalating rework costs and warranty claims with no systematic root cause data
- Real-time monitoring of every assembly step — correct parts, correct sequence, no missing components
- Operator alerts at the moment of error, not after
- Support for multiple product configurations assembled on the same line
- Automated throughput, cycle time, and efficiency measurement per station
- Per-unit Digital Quality Certificate documenting completed assembly as proof of quality
- Zero additional workflow burden on operators — the system watches, it doesn't interrupt
"The problem with end-of-line inspection is that it finds errors that have already been buried under three more hours of assembly. We needed the AI at the moment of action — not the moment of completion."
The Solution: Intelgic Process Intelligence System
Intelgic designed and deployed a Process Intelligence system built specifically for complex multi-station assembly operations. Unlike generic camera monitoring platforms, this system was trained end-to-end on the client's specific products: every part, every step, every work instruction, every valid sequence — and every deviation that constitutes an error.
Six Core Monitoring Capabilities
Correct Part Selection Verification
The AI identifies every component picked by the operator and verifies it against the work instruction for that step — flagging wrong parts before they are installed.
Assembly Sequence Enforcement
Every step is checked against the required sequence. Out-of-order actions are detected and flagged in real time — before the operator proceeds to the next step.
Missing Component Detection
The system verifies the complete component complement at each stage. Any missing part triggers an immediate station alert before the unit progresses to the next station.
Throughput Measurement
Cycle time is measured automatically for every step, every station, and every unit — with no manual data entry. Throughput trends surface in the dashboard in real time.
Units Assembled per Day
Production counts per shift, per line, and per station are tracked automatically — giving management real production data without the lag of manual reporting.
Operational Efficiency Scoring
The system calculates OEE and efficiency scores per station and per shift — identifying bottlenecks, idle time, and performance gaps with the granularity needed to act on them.
How the System Works
System Trained on Parts, Steps & Work Instructions
Before deployment, Intelgic's AI models were trained on the manufacturer's complete product catalogue — every component, every assembly step, and every work instruction for every escalator variant. The system understands what correct assembly looks like at each station for each product type.
Cameras Monitor Operators Continuously at Every Station
Overhead and angled cameras at each assembly station capture continuous video of the work area, the operator, and the assembly in progress. No operator action is required to activate monitoring — the system watches from the moment a unit arrives at the station.
AI Verifies Each Step Against Work Instructions in Real Time
As the operator works, the AI cross-references every observed action against the work instruction sequence. Part identity, placement position, fastening actions, and step order are all verified continuously — with no operator input required.
Alerts Issued at the Moment of Error
When the AI detects a wrong part, a missing component, an out-of-sequence action, or any deviation from the work instruction, a real-time alert is generated at the station. The operator is guided to the correct action before the unit moves to the next stage. Errors are prevented, not discovered.
Analytics Dashboard Tracks Line Performance Live
Every station feeds real-time data into a centralised dashboard: step completion times, cycle times, units completed, error frequency by step, and efficiency scores per station and per shift. Supervisors and managers see actual line performance at any moment without leaving their desk.
Digital Quality Certificate Generated for Every Unit
When a unit completes its final assembly station with all steps verified, the system automatically generates a Digital Quality Certificate — a complete, timestamped record of every step performed, every component verified, operator identities, cycle times per step, and final quality status. The DQC is the auditable proof that the unit was assembled correctly.
The Digital Quality Certificate
The Digital Quality Certificate is the most operationally significant output of the Intelgic Process Intelligence system — and the deliverable that most directly addressed the manufacturer's traceability and quality assurance requirements.
A Machine-Generated Proof of Quality for Every Unit
For every escalator unit that completes the assembly line, the system automatically generates a DQC — a structured, timestamped document that records the complete verified assembly history. It replaces manual inspection sign-off with an objective, AI-verified record that cannot be falsified, forgotten, or lost.
Analytics & Production Intelligence
Beyond error prevention, the Process Intelligence system transformed how the manufacturer understood their own production line. Because the AI observes and timestamps every step continuously, it generates production analytics that were previously either unavailable or collected manually with significant lag and inaccuracy.
Step-level cycle time data revealed which specific assembly operations were consistently slower than target — enabling targeted process improvement that would have taken months of manual time studies to identify. Station-level efficiency scoring made bottlenecks visible for the first time, allowing production managers to rebalance workload between stations and eliminate idle time that had previously been invisible.
Error frequency data by step provided a ranked list of the assembly operations most prone to mistakes — guiding training investments and workstation redesign decisions toward the highest-impact interventions rather than across-the-board retraining.
Error Categories Monitored
Wrong Part Selection
Out-of-Sequence Assembly
Missing Components
Incorrect Component Placement
Skipped or Incomplete Steps
Operator SOP Deviations
Results
Within weeks of deployment, the Intelgic Process Intelligence system eliminated assembly errors reaching downstream stages. The combination of real-time step verification and immediate operator alerts meant that every detected deviation was corrected at the point of occurrence — before any rework cost was incurred and before the error was concealed by subsequent assembly steps.
Operational efficiency improved from 68% at baseline to 94.2% — a gain driven partly by error elimination and partly by the visibility the analytics dashboard provided into previously unmeasured idle time and station imbalances. Throughput increased from 107 units per day at baseline to 148, without adding headcount or extending shift hours.
The Digital Quality Certificate fundamentally changed the manufacturer's quality assurance posture. Where previously a field issue could only be investigated through operator recall and paper records, every unit now has a complete machine-generated assembly history. The time to root cause analysis for a field complaint dropped from days to minutes.
Perhaps most significantly, the DQC became a competitive differentiator. Customers in the elevator and escalator industry increasingly require documented quality evidence as a condition of procurement. The manufacturer is now able to provide machine-generated, AI-verified assembly documentation for every unit shipped — a capability none of their competitors currently offer.
"Every unit now ships with a machine-generated proof of quality. The DQC is not just an internal document — it has become part of how we sell."
