AI changes how improvement opportunities are found and implemented, but not why kaizen works. How AI tools can support continuous improvement, and why people, standards and the gemba remain at its heart.
A kaizen team investigating changeover delays had the usual tools: stopwatches, observation sheets and a spaghetti diagram. This time, they also had machine data showing every changeover for the past year, analysed by an AI tool that grouped delays by product, shift and cause. The analysis pointed them in minutes to the two product transitions that caused most of the lost time. The team still had to go to the gemba, watch the changeover, talk to setters and try improvements. But they started in the right place.
AI is changing how improvement teams work. It is not changing why improvement works.
Where AI helps kaizen
- Finding opportunities: analysing machine, quality and process data to highlight where losses are greatest.
- Understanding problems: identifying patterns and correlations in large data sets that people would miss.
- Faster documentation: drafting standard work, summaries, A3 reports and training materials from notes and photos.
- Knowledge sharing: knowledge assistants that help teams find previous improvements, lessons learned and solutions, as described in building an engineering knowledge assistant.
- Idea management: categorising, routing and tracking ideas from employees.
- Sustaining gains: monitoring performance automatically and alerting teams when processes drift.
What AI does not replace
- Going to the gemba. Data shows where to look; observation shows what is really happening.
- People's knowledge and ideas. Operators understand their work in ways data does not capture.
- Trying things. Improvement comes from experimenting on the floor, learning and adjusting.
- Standard work and discipline. New methods must be standardised and followed.
- Leadership. Leaders who ask questions, remove obstacles and recognise effort remain essential.
- Respect for people. Improvement that is done to people rather than with them does not last.
Risks to manage
- Analysis paralysis: spending more time on data than on action.
- Over-trusting AI conclusions without checking them at the gemba.
- Excluding people who are less comfortable with technology.
- Automating processes before improving them, the mistake described in Lean before automation.
A balanced approach
- Use AI to point teams in the right direction and to speed up analysis and documentation.
- Keep kaizen events grounded in observation, experimentation and team participation.
- Train facilitators to use AI tools well, building on internal capability like the facilitators in our facilitator case study.
- Measure results the same way: performance improvement, sustainment and engagement.
The future of kaizen
The best improvement cultures will combine human insight with AI-powered analysis: faster problem identification, better data and more time for people to do what only people can do, which is to understand, experiment and improve their own work.
