# Manufacturing AI Use Cases

## metadata
- title: Manufacturing AI Use Cases | Klyff
- route: /use-cases
- format: use_case_directory
- industries: Automotive, Electronics, Metal Fabrication, Energy, Packaging, Multi-Plant Manufacturing

## Overview
Klyff applies edge AI to production problems where teams need fast decisions, dependable operation, and control of sensitive plant data. These use cases can be deployed independently and expanded as the manufacturing network grows.

## Visual Quality Inspection
Use Inspeqtr when manual inspection or conventional AOI misses defects, creates inconsistent decisions, or struggles to keep pace with line speed. Edge vision models inspect parts and packaging on existing camera infrastructure, flag surface and assembly defects, preserve visual evidence, and route uncertain cases to operators for review.

Typical applications include PCB solder joints, metal surface flaws, textile consistency, food packaging, and component verification. Klyff's published implementation content describes targets such as 95%+ defect detection accuracy and reduced manual inspection labor; actual results depend on line conditions, data quality, and validation.

## Predictive Maintenance and Asset Reliability
Use Prescptr when time-based maintenance over-services equipment and reactive repairs cause avoidable downtime. It evaluates vibration, temperature, acoustic, and process signals to identify changing asset condition, prioritize inspection or repair, and help teams plan work around production schedules.

Typical assets include pumps, motors, bearings, gearboxes, conveyors, and compressors. Models can run on site so high-frequency operational signals remain local and alerts can be acted on without a cloud round trip.

## Multi-Site Intelligence and Fleet Learning
Use Senatr when plants need consistent model improvements but cannot centralize raw production images or telemetry. Each site trains or validates locally, shares privacy-preserving model updates, and receives governed model versions suited to its equipment and operating conditions.

This supports data residency requirements, reduces the need to transfer large raw datasets, and helps successful inspection or reliability improvements move across a distributed plant network.

## Related Product Pillars
- **Analyzr:** Connect and normalize plant telemetry for shared operating context.
- **Inspeqtr:** Automate visual quality inspection.
- **Prescptr:** Prioritize maintenance using equipment condition signals.
- **Senatr:** Coordinate model improvement across sites while keeping raw data local.

## Choosing a Starting Point
- Start with visual inspection when defect escapes, inspection labor, or inconsistent quality checks are the immediate problem.
- Start with predictive maintenance when critical assets fail unexpectedly or scheduled maintenance is poorly targeted.
- Start with multi-site learning when proven models must scale across plants under privacy or data residency constraints.