Transforming Knowledge Work: The Lean Approach to AI Integration
- Dec 9, 2025
- 6 min read
Updated: Jul 30
I was rewatching a classic John Shook video about NUMMI. Thank you, Nigel Thurlow, for recording and sharing it. As I watched, I reflected on its relevance to the automation of Knowledge Work using Artificial Intelligence (AI).
One line hit me hard:
“They were trying to automate out human problems.”
Result: A disaster — followed by an extraordinary recovery once people were allowed to be ingenious, collaborative, valued, and solve problems every day.
This statement cuts straight to the heart of a significant misunderstanding of Lean in the Western world.
How Lean is often misconceived
Lean is often treated as a toolkit for process optimisation. Many organisations outsource this task to consultants who arrive equipped with PowerPoints to “redesign” the enterprise. We even coined a term for it: Business Process Reengineering.
In true Lean Thinking, there should not be a need for “reengineering.” If you need to reengineer, it means you have not managed to improve at the pace that your business needs. Improvements should be a part of standard work and come from the people doing the work, supported by leadership and stimulated by senseis — not flown-in fixers.
In true "Lean", there is no wholesale process reengineering, replatforming, or transformation, because things should never grow that out of sync thanks to continuous improvements.
Lean never mattered because it optimised processes. It mattered because it ENGAGED PEOPLE TO THINK, COLLABORATIVELY.
The Digital Economy and Lean Principles
Now, let’s turn to the digital, software and service economy. Isn’t it ironic that many organisations still behave as though the world is divided between white-collar executives who think and blue-collar engineers and delivery teams who do?
McKinsey’s concept of “digital factories” reinforced this fiction, leading many businesses down the wrong path. In traditional manufacturing, engineering and production can be separated because of the practical constraints of a physical production chain. In technology and services, however, design and delivery are deeply interwoven.
Yet organisations aspire to become digital and AI-powered businesses while relying on leadership principles designed for the mass-production era of the 1930s.
Lean represented a major advance in manufacturing. It made greater variety possible, improved quality and lowered prices through increased efficiency. More importantly, it enabled people throughout the organisation to think, become more engaged and contribute greater value. It therefore offers important lessons for the leadership of knowledge work.
In modern knowledge work, preserving the divide between business and technology delivery is one of the great failures of digital transformation. It carries the outdated distinction between white-collar thinkers and blue-collar executors into organisational structures, operating models and leadership practices.
In the knowledge economy, these silos must fall. Everyone should be encouraged to think, remain connected to execution and learn continuously. Lean showed us both how to organise such a system and how to lead it.
The Importance of Leadership in Lean
True Lean represents a major evolution from the command-and-control leadership principles of mass production—and it is more relevant than ever in the digital and technology era.
Knowledge work depends on people being able to identify problems, test ideas, learn from outcomes and improve the system continuously. That requires leaders to create the conditions for thinking, collaboration and experimentation, rather than simply directing tasks from the top. Lean leadership provides a foundation for doing exactly that: connecting strategy with frontline knowledge and treating improvement as part of everyone’s work.
AI should be understood as the next stage in the continuing digitalisation of business. Yet the record of digital transformation has been decidedly mixed. In many banks, for example, processing a trade still depends on numerous workarounds and manual interventions. Across value chains, information is frequently exchanged in unstructured formats—even when the same information already exists in structured form elsewhere in the organisation or with suppliers. Value chains are still far from being integrated.
When these problems are viewed through organisational silos, AI can appear to offer an easy solution: use it to interpret unstructured information, bridge disconnected systems and automate manual inputs. Some of these applications will undoubtedly be valuable, but they do not come without risk. Overlaying automation on fragmented processes and poor-quality data introduces accidental complexity, making systems harder to understand, manage and improve—and creating new opportunities for errors.
AI investment should therefore include an effort to simplify. Organisations must improve how information flows through their end-to-end value chains, remove unnecessary hand-offs and ensure that data is carried in forms that minimise ambiguity and error. AI should not become an additional layer of complexity when the better answer is to simplify the underlying work.
This is a core Lean principle: eliminate waste rather than optimise its handling. Standardisation can provide a strong foundation for AI, and AI can itself support greater standardisation—but neither should be used to preserve processes that should first be redesigned or removed.
This may help explain why returns on AI investment remain so elusive. Too often, organisations begin with the technology and search for places to deploy it, rather than starting with the business problems that matter.
Combining AI with Kaizen offers a more productive path. Kaizen provides a disciplined way to identify meaningful problems, understand the work and improve it iteratively. AI can then be introduced where it genuinely strengthens the solution. This encourages purposeful adoption, generates measurable business benefits and allows organisations to continually recalibrate human practices as AI takes on more of the work.
The goal is neither to preserve every existing human activity nor to automate indiscriminately. It is to create an evolving system in which people and AI complement one another—and continuously improve how value is delivered.
So, it’s a good time to watch the video.. Start there ».
Reflecting on your AI Adoption
Lean has a concept for balancing automation with human judgement: Jidoka—often described as “automation with a human touch.”
Jidoka was not simply about replacing people with machines. It meant designing automation so that problems became visible, work stopped when something went wrong and people could focus on understanding causes and improving the system. Machines took on repetitive activity, while people remained engaged—collaborating, exercising judgement and applying their critical thinking to improve both the product and the business.
That principle is highly relevant as we begin automating knowledge work with AI. A Jidoka-inspired approach does not ask only, “What work can AI take over?” It also asks, “How can people continue to contribute meaningfully, and how can human–AI activity remain focused on creating value for the business and its customers?”
The goal is neither to preserve every existing human activity nor to automate indiscriminately. It is to create an evolving system in which people and AI complement one another, continually recalibrating the work between them while improving how value is delivered.
So, ask yourself:
IS YOUR ORGANISATION MAKING THE SAME MISTAKE IN ITS ADOPTION OF AI?
Are you only trying to automate away human problems?
Draw your own conclusions.
If you want a Jidoka-led approach to AI, let’s talk.
Conclusion: Improving the Work and its Value, Not Just Automating It
AI adopted in isolation is a technology solution looking for a problem. Without a clear connection to customer or business needs, its most obvious value proposition becomes reducing cost by displacing people. That may produce a short-term saving, but it does little to address fragmented processes, poor data, unnecessary work or the organisation’s ability to create greater value.
The more powerful combination is AI with Kaizen. Kaizen begins with meaningful problems and brings the people who understand the work into the process of solving them. AI then becomes an enabler of improvement—not the objective itself. This combination suits the emergent nature of AI adoption: organisations can experiment, learn and deliver value incrementally while continually recalibrating the balance between human contribution and technology.
Jidoka provides a complementary principle. It shows how automation can take on routine activity while people remain engaged in exercising judgement, identifying problems and improving the wider system. The objective is not simply to automate more work, but to eliminate waste, simplify the flow of information and increase the value created by the whole system.
And when technology does remove manual tasks, we should not assume that the people performing them have become redundant. Freed from repetitive work, they may apply their knowledge, creativity and critical thinking to opportunities that create far more value for the business than they cost.
That is how organisations become high-performing and adaptable: not by adopting AI for its own sake, but by using AI, Kaizen and human ingenuity together to improve how value is created.
Your next step: Turn Continuous Improvement into a Leadership Habit
AI and Kaizen will not deliver their full potential through technology or tools alone. Leaders must create the conditions in which people can identify meaningful problems, experiment, learn and improve the work continuously.
If you want to develop the practical leadership skills needed to make continuous improvement part of everyday work, join us for Essential Leadership Skills to Drive Continuous Improvements.
