Cobots have one heck of a pitch. No fence, easy to program, works right alongside your people. For the right job, all of that's true. The trouble is when folks buy one for a job that really needed a regular industrial robot...and then spend months wrestling with it.
A cobot's sweet spot is light, moderate-speed work with people around: tending a CNC, packing, screwdriving, palletizing light cases at a modest clip. They really are easier to reprogram, which matters a bunch if you change products every week and want your own crew handling it instead of calling an integrator. They also squeeze into spots where a fenced cell never would.
Where they struggle is speed. When a cobot works near people, it moves slow on purpose - that's how it stays safe without a fence. If your cycle time is tight, a cobot may just not keep up, and no amount of tuning beats physics. Heavy parts, long reach, precision at high speed, hot or washdown environments, three shifts of the same job...that's industrial robot territory.
The thing I'd push back on hardest is the idea that a cobot is safe because it's a cobot. "Collaborative" describes the whole job, not just the arm. Stick a knife, a hot part or a sharp sheet-metal blank in its gripper and it's suddenly not something you want to be cuddling up next to. Every cell needs a risk assessment against the robot safety standards, and plenty of cobot cells come out the other side with scanners or guarding added. The current ones agree with me, by the way: ISO 10218 was revised in 2025 (in the US it's ANSI/A3 R15.06-2025), and it now talks about collaborative applications rather than collaborative robots. Totally fine. But if you end up fencing it in and cranking the speed up anyway, you've paid extra for collaborative features you're not even using.
There's also a middle option folks tend to forget. An industrial robot can share space too, with safety scanners (or zoned light curtains) that slow it down when someone walks up and stop it if they get close. Full speed when the area's clear, protection when it isn't. For a lot of palletizing and packaging cells, that beats either extreme.
Still on the fence? Answer a few questions honestly. What cycle time do you need, and what's the heaviest thing it'll pick up, gripper included? Will people be in its space while it runs, or only to load and unload? How often will the job change, and who's doing the reprogramming? Is the part sharp, hot, heavy or fragile? Usually the answer's obvious by question three. If it isn't, a cobot pilot is a reasonable way to find out - as long as the safety side is planned from day one and not duct-taped on afterward.