Long reads, with pictures

The first robot: a field guide for the shop that has none

October 2, 2026 · 5 min read

Shops that do their first robot well all do the same seven things, in roughly the same order, and most of them cost nothing but attention. Here they are, from the stopwatch at peak to the first unattended night, with what each one looks like and the mistake it prevents. Illustrated, step by step.

The short answer

The shops whose first robot runs for a decade did not have better robots. They picked a dull repeating job with a long cycle, measured their real peak rate, made parts arrive predictably before buying anything, ran the demo with their worst parts, named an owner, planned the first jam, and designed the first unattended night on paper. Seven steps. Most of them happen before the quote. Here is what each one looks like.

Step one: the stopwatch at peak

Stand at the station on your busiest afternoon with a stopwatch and count what actually arrives: cases a minute, parts an hour, totes a shift. Not the monthly report. The peak. Write down the heaviest part too. This single number sizes the whole project, and it is the number most quotes get wrong because nobody measured it. The piece on sizing for peak on this site exists because the undersized palletizer is the most common failure in the category.

Step two: the boring job with the long cycle

List your candidate stations and cross off the exciting ones. The first robot goes on the dullest repeating job with the longest cycle time and the most consistent part, because that is the job a robot does for ten years without anyone thinking about it. The Washington machine shop in the casebook put its first cobot on a 56 minute cycle where the robot works six minutes, as trade press reported it, and the shop's output rose 43 percent. Nobody put the first robot on the hardest part. The shops that do write the cautionary pieces.

Step three: the fixture before the robot

Make parts arrive the same way every time, before any machine is bought. A staging fixture, a tray, a vise that locates the part without judgment, a divider at the machines so parts from three sources stop arriving in one heap. Half of every robot project is part presentation and all of it can be solved in advance for a fraction of the cell's cost, and the fixture usually helps the manual process too. If the station cannot be made predictable, that is the answer, and it is a cheap one to get.

Step four: the demo with your worst three parts

Bring the lightest, the heaviest and the one that is always slightly wrong, in the containers they actually arrive in, and ask the vendor to run at your peak rate with your operator's hands on the controls. A demo is a cell in its Sunday clothes; the three ugly parts are the guests. The piece on why the demo always works lists the rest of what to bring. If the vendor only wants your average part, you have learned the most important thing the demo could teach you.

Step five: the name on the card

Decide who owns the cell before the purchase order, and put their name on a card on the controller. Usually it is the person who runs the station today and sometimes it is the one who asked the loudest question about their job. Send them to the training, let them do the first teach with the integrator watching, and pay for the skill. A cell with an owner clears its own jams in two minutes. A cell that belongs to everyone belongs to nobody at 2 AM.

Step six: the first jam, planned

Every cell stops for small reasons a dozen times a week. Before delivery, write down what the operator does when a part is upside down, when the gripper drops one, when the vision system stops seeing the part because someone changed a light bulb, and whether the program restarts from the top or from where it stopped. Tape the answers to the controller. The integrator will show the operator once; the card is for month three, when the integrator is in another state.

Step seven: the night, on paper

If the cell will run unattended, design the night before the first one: how many hours one load of blanks buys, where the chips go, how the coolant lasts, which alarms recover themselves, which wait for morning, and which call a person. Then decide who owns the phone. The shops that skip this step learn it at 2:14 AM and often conclude that robots do not work here. They work. The night was not designed.

What happens next

The second robot takes half the time and a quarter of the anxiety, because the shop has learned to fixture, to measure peak, to own a cell and to say no to the exciting job. The Colorado molder in the casebook launched more than a dozen applications after its first cart-mounted arm, as the vendor published it, and nobody launches a dozen because the first was simple. They launch a dozen because the first taught them how.

Describe your three candidate stations to Joe, the concierge on this site, and ask him which one is step two. He will also tell you, for that station, what step three looks like, which is the step that decides whether the rest goes well.

Related on this site

• Your first robot should not be your hardest job
• The second robot is easy. Here is why the first one is hard
• Why the demo always works, and how to run one that tells the truth
• What happens the day the robot arrives
• Ten questions that make a robot salesperson sweat