Episode Details

Dear Andy,

This letter has two names on it. The second one spent a career building the machines.

Over the last two days I made an argument to you in two parts: that Amazon holds a fifth pillar it has never named, and that you have already built the instrument to fund it and pointed it at the door.

Both letters left one question standing, and it is the question anyone inside your building would ask first.

Why form people for work the machines are about to take?

It is not a cheap objection. It is the strongest one available, and if the answer is wrong then everything in the first two letters is sentiment. So I am not going to answer it myself.

Dr. Roland Menassa is a co-founder of the Capability Capital Institute. Before that he was Principal and Senior Manager for Advanced Technologies at Amazon, where he led the development and deployment of advanced robotics across the fulfillment network. Before that, the GE Advanced Manufacturing Technology Center. Before that, twenty-four years at General Motors, ending as a GM Fellow in flexible manufacturing systems. He was elected to the National Academy of Engineering in 2022 — the citation reads

for contributions to robotics and automation in aerospace, automotive, and e-commerce. He holds over a hundred patent filings.

He has spent thirty years removing human labour from buildings like yours. I asked him whether the people who remain matter less. He said the opposite, and then he told me about a night shift.

What the robots actually did

Workers once walked miles a day through endless aisles of inventory. Driven by a desire to improve both fulfillment speed and employee safety, Amazon introduced automated robotic storage fields: instead of workers traveling to products, products were brought to workers at ergonomic stations. Projection systems illuminated the correct bins. Vision-based scanning identified barcodes in motion. Fulfillment times shrank from days to hours, and what remained for the employee was final product selection, verification, packaging, and shipment.

Read that last sentence slowly, because it is the whole letter in miniature. The walking went away. The verifying did not. One of those is a task and the other is a judgment, and only one of them was ever going to be absorbed.

The night I walked ten miles

My experience at GM placed me inside one of the highest-volume automotive production environments in the world — a three-shift body shop, roughly 425 robots across 750,000 square feet, welding sheet metal into a complete vehicle body before paint. My first assignment was to lead a third-shift skilled-trades team of forty, maintaining quality and throughput while managing downtime across hundreds of automated stations.

On my first night, I spent the entire shift walking from one issue to the next — robot faults, sensor failures, code execution errors, quality-sensor anomalies. By the end of the shift, I had walked more than ten miles responding to one automation interruption after another. After several months of this routine, two conclusions became obvious. First, you become very fit. Second, and more importantly: the more automation, sensors, programming, and complexity you add, the higher the downtime and the greater the maintenance burden required to keep the line running.

Factories that attempted lights-out operations learned this faster. Fully automated, unattended manufacturing does not work in practice: skilled trades cannot diagnose, repair, or maintain in the dark. The very people required to keep the system alive are removed the moment the lights go off.

Automation is not a substitute for people. It is a system that depends on them. This is why the future of manufacturing is not lights-out. It is lights-on, people-in, automation-with.

Andy, hold those