OTP vs Lean and Six Sigma, process quality when an agent runs the process

Lean and Six Sigma are two of the most durable ideas in the history of management. They have made cars safer, hospitals faster, and factories cheaper. They are not about the org chart. They are about the work itself, the repeatable process that produces a result, and they exist to make that process leaner and more consistent.

So this is not a contest. I run a part-human, part-AI team, and I keep coming back to one question that the framework literature could not have anticipated. When an AI agent can both run a repeatable process and watch it for waste and variation, who owns that work, and where does it live on the chart? That is the question OTP exists to answer, and it sits next to Lean and Six Sigma rather than on top of them.

What Lean and Six Sigma are

Lean traces back to the Toyota Production System and was named and popularized in the West by James Womack and Daniel Jones in "The Machine That Changed the World" and "Lean Thinking." Its obsession is waste. Toyota called it muda, the activity that consumes resources without adding value for the customer. Lean attacks it with value stream mapping, flow, just-in-time delivery, and kaizen, the practice of continuous, incremental improvement carried out by the people closest to the work. Lean asks a simple question over and over. Does this step add value, and if not, why is it here?

Six Sigma came from Motorola, where engineer Bill Smith formalized it, and it scaled famously at General Electric under Jack Welch. Its obsession is variation. A process that is right on average but wildly inconsistent still produces defects, and Six Sigma drives toward a target of no more than 3.4 defects per million opportunities. Its core method is DMAIC, five disciplined phases. Define the problem, Measure the current state, Analyze the root cause, Improve the process, and Control it so the gains hold. Practitioners are trained and ranked through a belt system, from yellow and green to black and master black belt. Lean Six Sigma combines the two, waste and variation, into a single toolkit. Born in manufacturing, both are now used everywhere from healthcare to finance to software.

These are serious, proven disciplines. Nothing about AI makes DMAIC or value stream mapping obsolete. The thinking is sound.

What OTP is

OTP, the Organization Transport Protocol, is not a process methodology. It is an operating layer for organizations that run both people and AI agents. Its core idea is the accountability chart, the same artifact an operating system like EOS® would recognize, except the seats can be filled by AI agents as well as humans.

On the OTP chart, an agent is not a feature hiding inside a dashboard. It is an Agent Employee. It has a named seat, a defined role, a scorecard, and KPIs it reports itself. That is the wedge, and it is also the difference from most of what is being sold today. The common pattern is to bolt an AI assistant onto an existing tool and call it intelligence. OTP treats the agent as accountable headcount, a seat on the chart that owns an outcome and reports its own numbers.

The real difference: a method is not a worker

Here is the line that separates the two. DMAIC is a discipline. It is something a worker practices. OTP is the place where that worker has a seat.

A green belt does not stop working when a project closes. Between formal improvement projects, the belt is monitoring control charts, watching for drift, flagging anomalies, and keeping the process inside its limits. That ongoing vigilance is exactly the kind of repeatable, attentive work an AI agent is good at. The agent can run a process and watch it at the same time, indefinitely, without the work falling off anyone's plate.

What Lean and Six Sigma do not specify is who that watcher is on the org chart, because for their entire history the watcher was a person and the chart was assumed. OTP makes the watcher explicit. It gives the agent a seat and a quality KPI, so the improvement is owned and visible rather than buried inside a tool that no one is accountable for.

Where they fit together

This is where it gets concrete, and where the two ideas reinforce each other instead of competing.

An agent owns a Control-phase metric. The hardest part of DMAIC is the C, Control, because gains erode once attention moves on. An agent assigned to that metric watches it 24/7, never gets bored, and never deprioritizes it under deadline pressure. The Control phase finally has a permanent owner.

Waste and anomaly flags become agent outputs. A Lean practitioner hunts for muda in periodic reviews. An agent in a seat can surface a waste or variation flag the moment the data shifts, and that flag is a named output from a named seat, not a line item someone might notice next quarter.

A quality KPI sits on the scoreboard. Defect rate, cycle time, variation against control limits, whatever the process demands, the metric lives on the OTP scoreboard next to every other seat's numbers. The quality work stops being invisible and becomes part of how the organization reads itself.

The seat makes the improvement accountable. When something drifts, you know which seat owns it. That is the whole point of an accountability chart, and it is what was missing when the improvement work lived inside a tool rather than on the chart.

None of this replaces the method. The agent in the seat is still practicing DMAIC and still hunting muda. OTP just gives that practitioner a home.

The agent-employee dimension

Two things follow once the agent is a seat rather than a feature.

The first is the auto-populated scoreboard. Because the agent reports its own numbers, the quality KPIs update themselves. There is no analyst compiling a monthly control report, because the seat that owns the metric is also the seat that publishes it. The scoreboard reflects reality continuously.

The second is harder for any single-company framework to match. OTP has a cross-org learning layer, the OOS, where process learnings can travel between companies that share the protocol. A control limit that one organization tuned the hard way, a waste pattern that one team learned to spot early, these become portable intelligence rather than tribal knowledge locked in one plant. Kaizen has always been local by design, improvement carried by the people in the room. A protocol layer lets some of that learning cross organizational walls. That is the part I think is genuinely new, and it is the moat. Protocol, not product.

A practitioner read

Let me be honest about where this stands. Lean and Six Sigma have been refined over decades, across thousands of organizations, with a body of evidence and a trained workforce behind them. OTP is early. The accountability chart for agents is a young idea, and the cross-org learning layer is younger still.

So I am not asking anyone to choose. If you run Lean or Six Sigma today and your processes are humming, keep doing exactly what works. The person who should care is the ops leader who has started putting agents to work, agents that run and monitor real processes, and who has noticed that this work has no clear home on the chart. If your agents are doing belt-level monitoring but nobody can point to the seat that owns the quality number, that is the gap OTP fills.

Close

Lean removes waste. Six Sigma removes variation. Both assume a person runs the process and a person watches it. The shift is that an agent can now do both, continuously. OTP does not rewrite the method. It gives the agent practicing it a seat, a scorecard, and a place on the chart, so process quality is owned and visible instead of buried in a tool.

More in this series

This post is part of a series comparing OTP to the operating frameworks companies actually run on. Start anywhere, each one stands alone.

Or read the full series index.

Looking for the head to head against named tools rather than frameworks? See OTP vs Ninety and EOS One.