Intelgic · Talk to Us

Track Shop-Floor Cycle Time and Discover Unseen Manufacturing Data Using Cameras and AI

Back to Knowledge base
Intelgic · Technical Article Cycle-Time Tracking AI Manufacturing Analytics

Track Shop-Floor Cycle Time
and Discover Unseen
Manufacturing Data Using Cameras and AI

Intelgic brings visibility to manual and semi-automated manufacturing operations by deploying cameras and AI assistants at workstations to recognize process actions, verify work instructions, and measure the duration of every step.

Intelgic · Irvine, CAPublished 8/21/202622 min readAI · Cameras · Cycle Time · Throughput
01 · Introduction

Track Shop-Floor Cycle Time and Discover Unseen Manufacturing Data Using Cameras and AI

Many manufacturers know how many products leave the line at the end of a shift. However, they may not know exactly how much time is spent on every step, why one workstation runs slower than another, where operators wait for materials, or which process deviations reduce throughput.

Traditional production systems record machine signals, work orders, and finished quantities. Manual and semi-automated operations often remain less visible because their individual steps are not automatically recorded.

Intelgic brings visibility to these operations by deploying cameras and AI assistants at manufacturing workstations.

The AI observes processes through the cameras, recognizes individual actions, verifies that work instructions are followed, and measures the duration of each step. Intelgic then synchronizes these events with production data, work orders, product variants, and customer demand.

The result is detailed manufacturing analytics built from information that was previously unseen.

02 · System Definition

What Is AI-Based Cycle-Time Tracking?

AI-based cycle-time tracking uses cameras and computer vision to determine when a manufacturing process and its individual steps begin, progress, and finish.

What Is AI-Based Cycle-Time Tracking?

Instead of relying only on an operator to start a timer or scan a code, the AI recognizes actual process events.

Product enters the workstation.
Operator picks the first component.
Component is positioned.
Fastening begins.
Fastening is completed.
Inspection is performed.
Product leaves the station.

The AI records timestamps for these events and calculates total station cycle time, time spent on each process step, waiting time between steps, repeated-step duration, correction and rework time, product-transfer time, idle time, and actual productive time.

This creates a much more detailed view than measuring only the time between product entry and exit.

03 · Core Measurements

Understanding Cycle Time, Takt Time, Throughput, and Lead Time

Understanding Cycle Time, Takt Time, Throughput, and Lead Time

These terms are related but measure different aspects of production.

Cycle Time

Cycle time is the actual time required to complete a process, operation, or product at a workstation.

Complete production-line cycleWorkstation cycleProcess cycleIndividual step cycleMachine cycleManual operation cycle

For example, if an operator starts assembling a product at 10:00:00 and completes it at 10:02:30, the station cycle time is 150 seconds.

Takt Time

Takt time is the production pace required to meet customer demand.

Takt time = Available production time ÷ Customer demand

If a line has 24,000 available production seconds during a shift and customer demand is 200 units, the required takt time is 120 seconds per unit. If actual station cycle time regularly exceeds 120 seconds, that station may prevent the line from meeting demand.

Throughput

Throughput is the number of acceptable units completed during a defined period.

Units per minuteUnits per hourUnits per shiftUnits per day

Intelgic's AI can track actual process completions and connect the count with quality and production data to calculate station and line throughput.

Lead Time

Lead time is the total elapsed time from the beginning of a broader production or order process to completion. It may include processing, transport, waiting, queue, and storage time. By separating these measurements, manufacturers can understand whether a performance issue comes from slow processing, waiting, excessive work in progress, rework, or inadequate capacity.

04 · Hidden Shop-Floor Signals

The Unseen Data at a Manufacturing Station

A conventional production report may show that a station completed 350 units during a shift. It may not reveal what happened within those 350 cycles.

Time spent searching for components
Delays between process steps
Repeated tool operations
Incorrect components selected and replaced
Products waiting for workstation access
Operators waiting for material replenishment
Time lost because tools were unavailable
Assembly steps performed out of sequence
Products sent backward for correction
Work instructions consulted repeatedly
Extra handling and repositioning
Differences between product variants
Variation between shifts
Micro-stoppages that were never reported
Quality checks that consumed more time than expected
Work completed correctly but through an inefficient method

These events may last only a few seconds. When repeated hundreds or thousands of times, they can have a significant effect on throughput.

Intelgic converts these actions into structured process data that can be measured and analyzed.

05 · Camera-Based AI Assistants

How Intelgic's Camera-Based AI Assistants Work

Intelgic deploys cameras, AI software, and operator-assistance interfaces at selected manufacturing stations.

01 · Cameras Observe the Process

Industrial cameras are positioned to capture relevant areas such as operator work zones, component bins, assembly fixtures, tools, product-transfer points, inspection areas, packaging locations, and material-loading positions.

Area-scan camerasHigh-resolution camerasMultiple synchronized cameras3D camerasDepth sensorsCameras integrated into machinesRobot-mounted camerasSpecialized industrial lighting

The camera configuration is designed around the events that must be recognized.

02 · AI Recognizes Manufacturing Events

Intelgic's AI analyzes the visual stream and identifies actions such as product arrival, product departure, component picking, tool selection, part positioning, drilling, fastening, assembly, inspection, packaging, labeling, material movement, rework, and process completion.

Each recognized event receives a timestamp and becomes part of the station's digital process history.

03 · Work Instructions Are Digitized

The required process is configured as a sequence of expected events. A digital work instruction may specify: scan or identify the product, pick Component A, position Component A in the fixture, install two fasteners, connect the cable, perform the inspection, and transfer the completed product.

Optional stepsParallel stepsProduct-specific stepsMaximum or minimum step durationRequired toolsRequired component quantitiesInspection requirementsAllowed process variation
04 · Actual Work Is Compared with the Expected Process
Which step is currently activeWhich steps have been completedWhether a step was skippedWhether the sequence is correctWhether an action was repeatedWhether the expected component was selectedWhether the step exceeded its expected durationWhether the product is ready to proceed

This makes cycle-time measurement and work-instruction verification part of the same AI workflow.

05 · Cycle Time Is Calculated Automatically

The system records the start and completion time of each detected step. For every production cycle, Intelgic can calculate total station cycle time, value-adding process time, waiting time, idle time, step duration, material-picking time, tool-usage time, inspection time, correction time, transfer time, and repeated-step time.

These measurements can be connected with the product, work order, station, shift, and production batch.

06 · Operators Receive Real-Time Assistance

If the AI detects a deviation, delay, or missed step, it can assist the operator through workstation displays, visual instructions, audio alerts, stack lights, Andon systems, tablets, projected guidance, connected tools, and PLC-controlled signals.

"Step 3 has not been completed.""Incorrect component selected.""Complete the inspection before transfer.""This step has exceeded the expected cycle time.""Required material is not available at the station.""Repeat fastening operation detected."

Alerts can be configured to guide the operator without unnecessarily interrupting production.

07 · Results Are Synchronized with Production Data

Visual process events become more valuable when they are connected with manufacturing context. Intelgic can integrate with MES, ERP, SAP, PLCs, SCADA, digital work instructions, Bills of Materials, quality-management systems, production databases, warehouse-management systems, tool controllers, order-management platforms, and customer-order data.

Which product was being manufactured?Which customer order did it belong to?Which product variant was involved?Which work instruction was active?Which components were required?Which station performed the work?Was the order produced at the required pace?Will current throughput meet the planned quantity?
06 · Production Intelligence

Station-Level and Step-Level Cycle-Time Analytics

Station-Level Analytics

Intelgic can measure and compare performance at every instrumented station. A station dashboard may show:

Current productActive work orderCurrent process stepCurrent step durationCurrent station cycle timeTarget cycle timeTakt timeAverage cycle timeMedian cycle timeFastest and slowest cycleWaiting timeProcess-deviation countRework countUnits completedStation throughputEstimated order completion
Live Station Status

A live status can indicate whether the station is:

Running within targetApproaching the cycle-time limitWaiting for materialWaiting for an operatorBlocked by the next stationStarved by the previous stationPerforming reworkStoppedUnder manual review
Step-Level Analytics

Station-level averages can hide the actual source of a delay. Step-level analytics show which part of the process is changing.

Average durationMinimum and maximum durationMedian durationDuration distributionFrequency of delayNumber of repetitionsSkip frequencyCorrection timeVariation by productVariation by shiftTrend over time

For example, the total station cycle may have increased by 12 seconds. Step-level analysis may reveal that component-picking time increased by 9 seconds because the line-side container was moved farther away.

This level of detail allows teams to investigate specific process conditions rather than relying on general assumptions.

Calculating Manufacturing Throughput

Throughput can be calculated from AI-recognized process completions.

Station throughput = Accepted units completed ÷ Production period

At line level, throughput can be based on the completed output of the final relevant station. The calculation can account for good units, reworked units, rejected units, incomplete cycles, product variants, planned and unplanned stops, shift schedule, and work-order changes.

Combining throughput with quality data prevents defective or incomplete units from being treated as successful production output.

07 · Delivery Performance

Matching Production Pace with Customer Orders

Customer-order data defines what must be produced, in what quantity, and by what deadline. Intelgic can synchronize actual production events with customer orders, sales orders, production orders, required quantity, product mix, planned completion time, shipping deadline, and available production time.

The System Can Compare
Planned units versus actual unitsRequired takt time versus actual cycle timePlanned throughput versus actual throughputCompleted order quantity versus remaining quantityCurrent production pace versus required delivery time
Example Dashboard
Customer order: CO-84521Required quantity: 1,200 unitsCompleted: 735 unitsRemaining: 465 unitsCurrent throughput: 92 units/hourRequired throughput: 100 units/hourProjected completion: 47 minutes behind planPrimary constraint: Assembly Station 6Longest step: Cable routing

This connects real-time shop-floor behavior with delivery performance.

08 · Bottlenecks and Micro-Stoppages

Finding Bottlenecks with AI

A bottleneck is the process or resource that limits overall production flow. Intelgic can help locate bottlenecks by analyzing station cycle time, queue formation, product waiting time, blocking and starvation, work-in-progress accumulation, step duration, rework frequency, station utilization, throughput, and variation over time.

The slowest average station is not always the true bottleneck. A station with high variation or frequent micro-stoppages may constrain the line even when its average appears acceptable. Camera-based data provides additional context by showing which physical events accompany the delay.

Detecting Micro-Stoppages

Short interruptions are often missed by conventional downtime reporting because they are too brief to trigger manual documentation.

Searching for a toolRepositioning a partClearing a fixtureWaiting for a component binAdjusting a cableRepeating a fastening actionMoving a finished part manuallyCorrecting an orientationConsulting the work instruction

Intelgic can record these events and calculate their cumulative effect. A five-second delay repeated 500 times creates more than 41 minutes of lost production time.

09 · Stability and Compliance

Understanding Cycle-Time Variation and Work-Instruction Compliance

Cycle-Time Variation

Average cycle time alone does not show whether a process is stable. Two stations may both average 90 seconds. One may consistently operate between 88 and 92 seconds, while the other varies between 60 and 120 seconds.

Standard deviationPercentilesMinimum and maximumMedianOutliersStep-level variationProduct-specific variationShift-specific variationTrend over time

High variation may indicate inconsistent material availability, unclear work instructions, difficult product variants, workstation-layout problems, tool issues, or repeated corrections.

Work-Instruction Compliance

Cycle time should not be improved by skipping required work. Intelgic measures process duration while also verifying that expected steps are completed.

Skipped stepsIncorrect sequenceRepeated actionsWrong component selectionIncorrect tool usageMissing inspectionIncomplete processPremature product transfer

A short cycle is not treated as a positive result if mandatory operations were omitted.

Product-Variant, Shift, and Station Comparison

Different product variants may require different components, tools, or process steps. Intelgic can compare cycle time by product model, variant, option package, work order, component combination, and customer configuration.

The analysis may show that one variant consistently requires additional time at a particular step. Teams can then update line balancing, work instructions, tooling, staffing, or scheduling.

Manufacturers can compare process behavior across workstations, production lines, shifts, facilities, products, and time periods. The objective is to identify process patterns and improvement opportunities, not to draw conclusions from isolated operator-level measurements.

Step duration before and after a layout changeCycle time across two identical stationsWaiting time by material-delivery scheduleProcess variation before and after trainingThroughput across product mixesError frequency before and after an SOP update
10 · Dashboards and Investigation

Real-Time Production Dashboard and Historical Analytics

Live Production Dashboard
Line statusActive customer ordersUnits completedUnits remainingCurrent throughputRequired throughputCurrent cycle timeTakt-time comparisonStation bottleneckDelayed process stepsWaiting and idle timeSOP deviationsRework eventsProjected order completionAlerts requiring attention
Dashboard Locations
WorkstationsSupervisor areasControl roomsQuality officesProduction-planning teamsManagement dashboards
Historical Analytics

Historical data helps teams understand patterns beyond the current shift. Intelgic can analyze hourly, daily, weekly, and monthly trends; cycle-time changes; step-duration trends; throughput trends; order-performance history; recurring bottlenecks; process-deviation trends; material-related delays; rework patterns; and results following process changes.

Filters can be applied by station, product, variant, line, work order, batch, or time period.

Root-Cause Investigation

When a cycle exceeds its target, authorized users may review which step caused the delay, what event occurred before the delay, whether material was available, whether an action was repeated, whether the wrong component was initially selected, whether the product required correction, whether the station was blocked or starved, and whether a work-instruction deviation occurred.

Configured image or video references can support investigation according to the organization's data-governance policies.

11 · Platform, Integration, and Traceability

AI Assistance at Every Station and the Intelgic Certainty Platform

Real-Time Process Assistance
Display work instructionsVerify process sequenceConfirm step completionDetect incorrect actionsAlert operatorsSupport correction
Process Intelligence
Measure cycle timeMeasure individual step durationTrack waiting and idle timeCalculate throughputIdentify bottlenecksAnalyze variationConnect production with customer ordersGenerate detailed analytics

Combining these roles creates a feedback loop between execution and improvement.

Intelgic Certainty Platform

Intelgic's Certainty platform can coordinate camera connectivity, AI analysis, process logic, production integration, and analytics. For cycle-time tracking, Certainty can:

Connect with industrial camerasReceive images and video eventsRun action-recognition modelsConfigure process workflowsRecord step timestampsCalculate cycle timesVerify work instructionsGenerate operator alertsConnect with PLCsSynchronize with MES and ERPLink production with ordersStore process recordsProvide dashboards and analyticsManage recipes and AI-model versions

Each station can operate with its own workflow while contributing data to a line-level manufacturing view.

Factory-System Integration

PLC
MES
ERP
SAP
SCADA
Quality-management software
Warehouse-management systems
Order-management systems
Digital work instructions
Tool controllers
Barcode and RFID systems
Production databases
Cloud or on-premises applications

The integration creates context around each observed process event.

Process Traceability

Every manufacturing cycle can create a structured record containing:

Product IDPart numberProduct variantWork orderCustomer orderStationProcess stepsStep timestampsTotal cycle timeWaiting timeProcess deviationsCorrective actionsFinal process statusAI model versionProcess recipe versionShift and production time

This record connects the physical manufacturing process with its digital production history.

12 · Governance and Deployment

Responsible Use and Implementation of AI Cycle-Time Tracking

Responsible Use of Camera-Based Analytics

Camera-based manufacturing analytics should be deployed with transparent governance. Organizations should define:

The operational purpose of the systemWhich stations and process areas are monitoredWhich data is calculated and storedWhether images or video are retainedWho can access detailed recordsHow long data is retainedHow operators are informedHow uncertain results are reviewedHow cybersecurity and access control are managed

Analytics should be interpreted in process context. Station design, product mix, material availability, equipment conditions, and upstream constraints can all affect cycle time.

Predicting Production Completion

By combining current throughput, remaining order quantity, cycle-time trends, and station conditions, Intelgic can estimate when an order is likely to be completed. The projection can be updated as production conditions change.

Units remainingCurrent throughputRecent cycle-time trendPlanned breaksProduct changeoversCurrent bottleneckRework rateAvailable production time

The estimate is an operational projection rather than a guarantee, but it provides planning teams with a more current view than a fixed schedule.

Implementing AI Cycle-Time Tracking

A typical implementation follows these stages:

Select stations and processes.Define process steps and work instructions.Identify required cycle-time events.Study camera visibility and lighting.Connect product and work-order data.Collect representative process images or video.Train and validate action-recognition models.Configure timing and compliance rules.Design operator alerts and dashboards.Integrate PLC, MES, ERP, and order systems.Run a controlled production pilot.Validate measurements against approved references.Deploy the system.Monitor and improve performance.
13 · Complete Line Visibility

Data That Was Previously Unseen

A production line contains thousands of small actions that are rarely converted into structured information. Intelgic's cameras and AI assistants make it possible to measure:

How long each step actually takes
Where time is lost
Which actions are repeated
Which stations are waiting
Which products create variation
Which work instructions cause difficulty
Where throughput is constrained
Whether production is keeping pace with customer orders
What physical events occur before delays and errors

This turns the shop floor into a source of continuous, actionable manufacturing intelligence.

See Every Step. Measure Every Cycle. Improve the Complete Line.

Intelgic deploys cameras and AI assistants across manufacturing workstations to observe process execution, verify work instructions, measure cycle time, and calculate throughput.

The system connects these physical events with production plans, product variants, work orders, and customer demand. Certainty then converts the combined data into detailed dashboards, alerts, traceability records, and operational insights.

Manufacturers can finally see the information hidden between the start and completion of each production cycle.

Intelgic · AI Cycle-Time Tracking

Ready to see every step
and measure every cycle?

Contact Intelgic to discuss camera-based cycle-time tracking, AI work-instruction monitoring, and shop-floor analytics for your manufacturing line.

Head Quarter

17352 Murphy Ave Suite 101
Irvine, CA 92614
Phone: (949) 317-2420

Branch Office

5ES7-H, 5Th Floor, East Tower, Mani Casadona, Plot no -IIF/04, Action Area -IIF, Newtown, Kolkata-700160

Certificate

intelgic ISO 27001 certification

Social links

©2026 Intelgic Inc. All Rights Reserved.