Ford didn’t invent the assembly line because he wanted to build cars.
What McKinsey’s AI Assembly Line really means for European industry.
Henry Ford didn’t invent the assembly line because he wanted to build cars. He invented it because he had to change the economics of production. In 1908, his factory produced 12,000 Model T cars per year at $825 each. By 1925, output had reached two million vehicles at $260 each.
Everyone knows these numbers. What is usually overlooked is the sequence of steps that made them possible.
Ford didn’t start with the line. He started with the standardization of parts. Interchangeable components, consistent tolerances, defined interfaces — that was the real industrial revolution. The assembly line was merely its logical consequence. Without standardized parts, no line in the world would have worked.
This sequence is essential to understanding what McKinsey proposes in its recent paper “The AI Assembly Line: Strategic Imperatives for CEOs.”
What McKinsey gets right
The paper offers one of the clearest strategic framings of Agentic AI I have read in months. Three points deserve unequivocal agreement.
First: AI is not a technology project. It is a business transformation. McKinsey backs this with a number that belongs in every boardroom: CEO-led transformations are 1.5 times more likely to succeed than those driven by technology teams. Anyone still locating AI inside the IT department has not understood the problem. This is about organization, roles, spans of control, performance metrics. It is about the operating system of the enterprise.
Second: The architecture is clearly described. McKinsey distinguishes between functional agents — operating within a domain — and enterprise agents that orchestrate the end-to-end flow. This is not semantics. It is the next stage beyond the chatbot. It is the moment AI moves from the information layer to the action layer.
Third: The urgency is well founded. Chinese automotive manufacturers develop new models in 24 months. Their Western competitors need 40. They have captured more than 50 percent of their home market. Any European OEM still clinging to “artisanal engineering” processes will not survive the next decade. This is not a prediction. It is arithmetic.
Up to this point: complete agreement.
Three places where it stalls
This is exactly where my work as a translator between McKinsey’s vision and industrial reality in Europe begins. The paper describes a destination most companies cannot reach from where they stand today. It describes the running line — but not the road to it.
Stall 1 — The data foundation
McKinsey writes: “Leaders should ensure that all AI solutions share the same data and operational framework.” A half-sentence, embedded in a paragraph on architecture.
In industrial practice, this half-sentence is the 18-month foundation work on which most projects stumble. In 30 years, I have watched ambitious transformations break on exactly one question: Do we actually know what we know?
A typical example from logistics: an industrial company implements a Transport Management System. Algorithmically sound, professionally designed. The system cannot calculate optimized routes — because no one knows how much the parts weigh. Or how large they are. Master data lives in Excel spreadsheets that haven’t been updated in years.
This is not an isolated case. It is the rule. And it is why the discussion about Agentic AI floats in mid-air without an honest inventory of data reality. A functional agent assessing supplier risk needs a consistent view of suppliers. An enterprise agent orchestrating source-to-contract needs a continuous process and data architecture across twelve SAP systems.
McKinsey treats this prerequisite as given. In European industry, it is the bottleneck.
Stall 2 — The standardization gap
Here we return to Ford. Ford needed interchangeable parts before the line could run. In the European industrial mid-market, the “parts” are processes — and they are typically the product of history: site-specific, department-specific, sometimes person-specific.
A typical automotive Tier 1 supplier with 800 suppliers and twelve SAP systems rarely has one procurement process. It has twelve variants of the same process, each with its own approval workflows, its own material group logic, its own exceptions.
McKinsey’s Assembly Line presupposes that these processes have been standardized. It does not say: “Standardize first, then automate.” It says: “Here is the architecture in which you automate.”
Reverse the sequence — deploy agents on non-standardized processes — and you produce pilots that work in one plant and fail in the other eleven. That is not the technology’s fault. It is the sequence Ford understood in 1908: first the parts, then the line.
Stall 3 — The translation gap
The paper addresses the CEO. The CEO decides. Fair enough. But between the CEO’s decision and operational reality lies a layer that does not appear in McKinsey’s argument: the people who have spent 20 years thinking in SAP transactions and are now expected to work with agents.
This layer is not the problem. It is the precondition for success.
A planner who has known for 15 years that a supplier in Poland will deliver despite the red light in the system — because he personally knows the plant manager — carries knowledge that exists in no database. If an agent is to take over that decision, the knowledge must be made explicit. That is not a technical task. It is a translation task.
McKinsey describes elegant architectures. It does not describe who does the translating. And it is precisely this gap that decides whether a board presentation becomes a working transformation.
The right sequence
What does this mean in practice?
It does not mean McKinsey is wrong. It means the road to the AI Assembly Line in European industry has three stages that McKinsey compresses in the paper — and that must be unfolded again in translation.
Stage one: The inventory. Before anything is orchestrated, you have to know what exists. Where does data originate? Where does it get lost? Which processes are documented, and which live only in the heads of experienced employees? Where are the cognitive bottlenecks McKinsey refers to — the “white-collar-intensive workflows” in which thousands of hours disappear into emails and meetings? This is not glamorous work. It is the foundation for everything that follows.
Stage two: Standardization. Not of all processes — but of the three to five that carry the highest volume. Only on a standardized foundation does the Assembly Line begin to make sense. Invest 18 months here, and you gain three years at the end.
Stage three: Orchestration. Now — and only now — do functional agents begin to deliver their value. Now the enterprise agent becomes the strategic backbone. Now McKinsey’s vision becomes a reality that shows up in the P&L.
Closing
Ford didn’t start with the assembly line. He started with the standardization of parts. The line was the consequence, not the starting point.
European industry will not build McKinsey’s AI Assembly Line by buying agents. It will build it by opening the factory gate, taking an honest inventory, and standardizing the parts. Then — and only then — will the line begin to run.
This is not a delaying tactic. This is not German thoroughness as an excuse. This is the sequence Ford himself respected. Skip it, and you build showroom models. Respect it, and you build industry.
The question for boards is not whether the AI Assembly Line is coming. It is coming. The question is whether, when the line starts moving, the parts in your own organization are ready.
E-Mail: sven.vollmer@business-quotient.com
Sven Vollmer is “The Industrial Translator.” He bridges the gap between industrial operational reality (SAP, supply chain) and the possibilities of generative AI. His focus is on value-creating applications—beyond the hype.
Transparency Note: This article was created with editorial support from AI (Gemini/Claude). The ideas, technical validation, use case selection, and adult supervision were 100% authored by Sven Vollmer.
LinkedIn: www.linkedin.com/in/sven-vollmer-bq
