Things I find outside the Lab. Some smart. Some ridiculous. Some expensive. A few follow me home.
↓
SEEN IN THE WILD / 01
THE AUTOMATION GUY.
THE SITUATION
n8n. Make. Python. Custom scripts.
HOW DO I GET CLIENTS?
SO HERE’S THIS GUY.
The guy is ready to automate the crap out of something.
One little problem: no customers.
Hold on. You picked the tools. Learned the technology. Built the capability. And now we’re going looking for somebody who needs it?
Maybe this guy doesn’t have a marketing problem. MAYBE WE GOT THE ORDER BACKWARDS.
SEEN IN THE WILD / 02THE BUILDER
THIS GUY CAN BUILD.
Real technical experience. Real systems. Real capability. Selling it? Different story.
SEEN IN THE WILD / THE JOB
Data & AI Automation Engineer
GenAI / agentic AI
data pipelines
APIs + integrations
cloud + databases
workflow orchestration
ALSO REQUIREDTalk directly to clients. Lead technical sessions. Explain complex systems to people with different levels of expertise.
SEEN IN THE WILD / 04
AIN'T THAT SOME BS.
So lemme get this straight.
You pay somebody to do the work. Over time, she learns the customers. The exceptions. The workarounds. What breaks. Who to call. What nobody bothered putting in the manual. Why everybody stopped doing it the “official” way.
Then she leaves.
And everybody spends weeks trying to reconstruct what she knew.
OH. WE DIDN’T ACTUALLY HAVE THE PROCESS. SHE HAD THE PROCESS.
Think about that.
THE COMPANY PAID FOR THE LEARNING. SHE WALKED OUT WITH THE TEXTBOOK.
BACK AT THE LAB Apparently this isn’t just small-business chaos. NASA has studied the underlying knowledge-continuity problem at institutional scale.
HOW MUCH DO COMPANIES SPEND TEACHING PEOPLE THINGS THE COMPANY ITSELF NEVER LEARNS?
SEEN IN THE WILD / 05
THE REQUEST SAIDAI + AUTOMATION
WHAT HE ACTUALLY WANTEDUNDERSTAND → MAP → FIND BOTTLENECKS → PRIORITIZE → IMPLEMENT
THEN THERE’S THE OWNER WHO THINKS HE WANTS AUTOMATION.
The request says AI. Automation. Fine.
Then you read what he actually wants. Look at the business. Understand the workflow. Map what’s happening. Find the bottlenecks. Figure out where the time and money are going. Prioritize what should change. Then implement.
Hold on.
That’s not the same problem.
ILLUSTRATIVE / 06
$250,000
ONE EXECUTIVE / ONE YEAR
Here’s an expensive one.
Let’s say a company decides it needs an executive. $250K a year. Illustrative number.
Fine.
Now add the recruiter. The search. Interviews. Negotiation. Salary. Benefits. Onboarding. Months learning the company.
Maybe that’s exactly what the company needs. Great. Hire the guy.
BUT HUMOR ME FOR FIVE MINUTES.
What are we actually paying him to do?
LEAD PEOPLE?Great.
MAKE CONSEQUENTIAL DECISIONS?Human.
BUILD RELATIONSHIPS?Human.
CHASE INFORMATION BETWEEN SIX SYSTEMS?Hold on.
RECONCILE REPORTS?Hold on.
REMEMBER WHAT EVERYBODY ELSE FORGOT?Now we’re talking about something else.
Maybe hire the executive. I’d just like to know what capability the money is buying.
The number is illustrative. The decomposition is the point.HYPOTHETICAL / 07
OKAY. YOUR TURN.
So your people are copying information from one system into another. Then they check it. Then somebody approves it. Then they copy something else. This happens all the time. Where would you automate?
ORDER ARRIVES→PERSON COPIES DATA→SECOND SYSTEM→MANUAL CHECK→APPROVAL
Look at the person in the middle. They’re carrying information from one system to another, then checking it by hand. Of course they want relief. Fine. What do you do?
CHOOSE A MOVEWHAT WOULD YOU DO?
DON’T AUTOMATE THE WORKAROUND BEFORE UNDERSTANDING WHY THE WORKAROUND EXISTS.
I’ve seen this enough times now.
OKAY. THIS ONE KEEPS SHOWING UP.
THIS ONE FOLLOWED ME HOME.STREETS → LAB
BACK AT THE LAB.
Most of this stuff stays on the street. But every once in a while, something keeps showing up and the obvious question won’t go away. That’s when it comes back to the Lab.
