DOI: 10.11648/j.ajmie.20261105.15 ISSN: 2575-6060

DMAIC, Augmented: Generative AI Strengthens Lean Six Sigma Execution

Oleg Smirnov
Show PDF Cite
This article examines how generative artificial intelligence (AI) assistants can strengthen Lean Six Sigma (LSS) execution across every phase of the Define-Measure-Analyze-Improve-Control (DMAIC) methodology. The objective is to give LSS practitioners field-tested patterns, quantified cycle-time data, and a governance framework for integrating AI assistants into quality-improvement projects without compromising methodological rigor. The investigation draws on recent industry deployments using Anthropic’s Claude assistant and a construct called skills: versioned playbooks that bundle written instructions, scripts, reference data, and validation rules into reusable artifacts. Methods include direct measurement of cycle-time deltas across Define, Measure, and Control tasks; four worked Analyze-phase examples spanning hypothesis testing, Measurement System Analysis, Statistical Process Control, and regression interpretation; and analysis of a multi-skill portfolio deployed at a small aviation services client. Results demonstrate substantial efficiency gains: project-charter drafting time fell from 90 to 120 minutes to roughly 6 minutes; monthly payroll consolidation fell from 90 minutes to under 30 seconds (a 99.4 percent reduction) with the first-pass error rate dropping to zero; cross-system benefits reconciliation fell from 4 hours to under 60 seconds, with monthly completion improving from 5 of 12 cycles to 12 of 12. The aggregate skills portfolio at one client achieved approximately 84 hours of administrative cycle-time savings per year. The article identifies six required governance controls (human review, version control, controlled prompts, sensitive-data boundaries, hallucination mitigation, and auditability) and aligns them with ISO 9001: 2015 and the NIST AI Risk Management Framework. The conclusion is that generative AI does not alter DMAIC methodology; it lowers the cost of executing each phase well, expanding the share of LSS that is practiced in fact rather than only in theory. The findings support Master Black Belts and quality champions in deliberately piloting AI-enabled tooling, prioritizing Control-phase tasks first, and treating skill development as a formal deliverable with version control, ownership, and quality gates.

More from our Archive