[Watch the video of this post here.]

If you don't know Rube Goldberg: he was a cartoonist in the early 1900s who drew absurdly complicated machines that accomplished simple tasks. A bowling ball rolls down a chute, which tips a bucket, which pulls a string, which lights a match, which... eventually butters your toast.

The machines weren't meant to be efficient. That was the point.

Goldberg was satirizing his era's obsession with mechanization. Everyone was so drunk on "we CAN automate this" that nobody stopped to ask "should this take 47 steps?"

Sound familiar?

Right now I can:

• Use one AI to research a topic

• Pipe that into another AI to write a draft

• Send it to a third AI to critique the draft

• Have a fourth AI revise based on the critique

• Run it through a fifth AI to check for tone

• Then a sixth to format it for LinkedIn

I’ll admit this is a little contrived, but we are spending more and more time creating “agents” to do things like this.

Often at the end of these magnificent contraptions... we have a social media post.

Which we could have just written in 30 seconds.

Don't get me wrong—I love this stuff. Building AI apps is my favorite puzzle to solve right now. I've got truly befuddling automations that would make Mr. Goldberg proud.

And watching a piece of software come to life is the very act of creation itself. It’s life! It is art!

But I've noticed something:

Many AI workflows I’ve seen out there often accomplish what a clear-thinking person could do in a fraction of the time. We're so excited that we CAN chain these tools together that we forget to ask if we should.

Goldberg's machines were commentary disguised as invention.

I wonder if some of ours are too.

The next era won't be about building longer chains. It'll be about knowing when the chain is the point—and when you should just butter the damn toast yourself.