Where AI Actually Saves Time in a Lean Six Sigma DMAIC Project (And Where It Doesn't)

I've been tracking where AI saves time in Lean process improvement projects. Not the marketing claims - the actual hours, on real engagements, with government clients.

The short answer: on the projects I’ve measured, I typically see around 20–40 facilitator hours that can be returned to the mission per DMAIC project when AI is used thoughtfully across the work.  Your mileage will vary based on tool maturity, project scope, and how much of the process is text-heavy vs. data-heavy.

The biggest single savings is the final report.

Assembling a 20-page executive report from multiple project artifacts - charter, data analysis, root causes, experiment results, recommendations - used to take 6 to 10 hours. With AI, the structure and first draft come together in 2 to 3 hours. You still add the narrative judgment, the political sensitivity, and the story arc that makes the sponsor care. But the assembly time is cut by more than half. That's 4 to 7 hours back on a single deliverable.

The second biggest is interview synthesis.

On a typical project I conduct 8 to 12 stakeholder interviews and keep typed notes for each one. Synthesizing them manually - re-reading every notes page, building a theme matrix, identifying contradictions, spotting exceptions - takes 3 to 5 hours. AI does the first-draft synthesis in 20 minutes or less. It analyzes all 12 interview documents simultaneously, which a single human brain can't do. I still validate every theme and catch what the AI misses. But the 3-hour scramble becomes a 40-minute review. Two to four hours saved.

A reasonable objection: synthesizing interviews is the analysis. If AI does it, you don’t actually understand what stakeholders said. Fair point.

I read every transcript first and take my own notes. Then I use AI synthesis to surface what I missed. AI is a second pair of eyes, not a substitute for my own analysis.

The third is drafting your plans to test out your ideas.

A typical project produces 4 to 6 improvement ideas to test out. We plan each of these experiments by filling out a customized Lean Agility template. Typically, each one takes 45 to 60 minutes to write from scratch. With AI generating the first draft from a good prompt, it's 20 to 30 minutes of refinement per plan. Across 4 to 6 experiments, that's 2.5 to 4 hours saved – around half a day!

Where AI doesn't save time: the human work.

Building trust with a nervous sponsor. Coaching a director who reviews everything because of one problem three years ago. Creating psychological safety for a team member who's afraid to say "I don't understand." Reading the room when someone is about to shut down and check out of the workshop. None of that got faster. None of it should.

AI handles the arithmetic so you can handle the meaning. That's a pretty good deal.

Key takeaways:

• 20-40 hours saved per project when AI is used across all DMAIC phases.
• Biggest wins: assembling report (4-7 hrs), synthesizing interviews (2-4 hrs), planning experiments to test out new ideas (2.5-4 hrs).
• AI helps you between workshop sessions, so you can be more effective during sessions.
• Every AI output is only a first draft—it always requires your review and judgment.
• The human work doesn't get faster from AI. But AI gives you more time to focus on this critical piece of the project.

Craig Szelestowski is a recovered executive, the founder of Lean Agility Inc. an instructor at the Telfer Centre for Executive Leadership, University of Ottawa, and a Subject Matter Expert at the New Jersey Institute of Technology.