Quality Control

The Defect That Costs Crores

A precision auto component manufacturer in Pune ships 50,000 parts per month to a Tier-1 supplier for Maruti Suzuki. One batch of 2,000 gear blanks passes all standard QC checks — dimensional within tolerance, surface finish within spec. But three months later, the Tier-1 supplier reports premature wear in assembled gearboxes. Root cause: a subtle metallurgical inconsistency in one heat treatment batch — the surface hardness was within spec at 58-62 HRC, but the case depth was 0.2mm shallower than usual. No individual measurement was out of spec, but the combination created parts that failed under sustained load.

This is the kind of quality problem that traditional inspection misses but AI catches — not by measuring better, but by correlating patterns across hundreds of variables simultaneously. The defect was not in any single reading. It was in the relationship between readings that a human inspector could never hold in memory across thousands of parts.

Quality in Manufacturing: What Can Go Wrong

Types of Defects

The Cost of Poor Quality in India

Indian manufacturers lose an estimated 15-25% of revenue to quality-related costs (according to CII studies). This includes:

For a manufacturer doing Rs 50 crore annually, quality costs can reach Rs 10-12 crore. Even a 20% reduction through AI-assisted quality control saves Rs 2-2.5 crore per year.

Statistical Process Control: The Foundation

Before AI, there was SPC — Statistical Process Control. Developed in the 1920s and still the backbone of quality management in Indian factories certified to ISO 9001 or IATF 16949 (automotive).

How SPC Works

SPC Limitations

This is exactly where AI adds value — multivariate pattern recognition across hundreds of process parameters simultaneously.

How AI Finds Quality Patterns

Shift Correlation

AI monitors all process parameters across shifts and finds that:

A human quality manager might notice the shift-based defect difference in monthly reports. They would not easily identify the root cause combination without AI analyzing all parameters together.

Material Batch Issues

AI tracks quality outcomes correlated with raw material batches and discovers:

Environmental Factors

AI correlates quality data with environmental conditions:

Open data/production-log.csv in the code panel. This contains 60 days of production data from a CNC turning centre — 500+ parts per day with dimensional measurements, tool wear data, spindle temperature, and material batch codes. The AI has flagged three quality patterns: a tool wear threshold beyond which surface finish degrades, a Monday morning dimensional offset, and one material batch with consistently tighter tolerances.

BIS Standards Compliance

The Bureau of Indian Standards (BIS) sets quality requirements across Indian manufacturing. AI helps with compliance in several ways:

For manufacturers supplying to government contracts (Defence, Railways, NTPC), BIS compliance is non-negotiable. AI ensures nothing falls through the cracks during high-volume production.

Six Sigma + AI: Better Together

Six Sigma's DMAIC methodology (Define, Measure, Analyse, Improve, Control) provides the framework. AI supercharges the Analyse and Control phases:

A typical Six Sigma project takes 4-6 months and analyses 5-10 variables. AI can analyse 100+ variables simultaneously and identify patterns in days — then the Six Sigma framework ensures the solution is validated and sustained.

Open data/defect-catalog.json to explore a structured defect taxonomy for machined components. Each defect type includes: visual reference description, common root causes, affected BIS/ISO standards, and recommended AI detection approach. Use this as a reference when building your quality AI prompts.

Getting Started: Your Quality AI Pilot

Key Takeaways