Automation Scales the Problem
I'm still on Elon Musk. This is the third note from the same podcast episode.
At some point I should probably move on. But every time I think I'm done, there's another part worth writing about.
In the early days of Model 3 production, Tesla built what Musk called "the machine that builds the machine." An almost fully automated factory. Robots everywhere. Minimal human involvement. It was supposed to be the future of manufacturing - the smartest, most efficient way to build cars at scale.
It nearly killed the company.
The factory kept breaking down. Production targets were missed by miles. The robots were doing things wrong, getting stuck, creating bottlenecks nobody anticipated. At one point, production was so far behind that Musk literally moved into the factory and slept there to try to fix it.
The irony: he had automated before simplifying. Before deleting. Before questioning whether every step that was being automated should exist at all.
The guy who created the algorithm violated his own algorithm.
That's funny to me.. It's also very human.
I've done the same thing.
A while back I built a content automation system on n8n. It would generate images and videos automatically, then post them directly to TikTok and other social media. The idea was to build a content factory - produce more, faster, more efficiently than doing it manually.
It ran. It posted. It produced content consistently.
The content was bad.
The prompts weren't optimized. The format wasn't figured out. The storylines didn't land with the audience. The generated videos had wrong movements, wrong product packaging, wrong everything. The automation was working perfectly - it was just automating a process I hadn't thought through properly.
I built the machine before I understood what the machine should make.
Automation feels like the smart move. It feels modern, forward-thinking, scalable. You're not just fixing a problem - you're building a system that fixes itself. What's not to like?
So companies automate early. They automate often. They automate proudly. They announce it in all-hands meetings and put it in investor decks. And then they spend the next year debugging why the automated system keeps producing the wrong output, in the wrong order, at the wrong speed.
Because they automated a process that was broken to begin with. Or one that didn't need to exist. Or one that nobody fully understood yet.
Automation doesn't fix bad processes. It scales them.
Musk eventually tore parts of the automation out. Brought humans back in. Slowed down to speed up. Only once the process was questioned, stripped down, and simplified did the automation actually work.
The sequence saved the company. The sequence he had skipped almost ended it.
Speed is only useful when you're moving in the right direction. Automation is only useful when the thing being automated is worth automating.
Most of us skip to the end before we've done the work at the beginning. We want the system before we understand the problem. We want the tool before we understand the process. We want to scale before we've figured out what's actually worth scaling.
Elon Musk - the person most associated with moving fast - learned this the hard way.
Delete first. Simplify. Then speed up. Then automate.
In that order. Always in that order.
