The Operator's AI Brief
Automation Comes Last
Sabrina Ramonov's five-stage list says to ask, think, teach, and delegate before you automate, and it shows where most owners can gain hours first.
Listen to this brief (3 min)
The audio is AI-produced: an AI clone of Jon's voice and an AI co-host, reading a script built from the sources below.
Sabrina Ramonov put her rule in the title of a video she posted on October 11: automation comes last. I worked from the title and description of each video covered here, so this piece reports what they say. I have no test results to add.
Five stages, and where the hours are
Her description lists five stages: ask, think, teach, delegate, automate. Stage one is asking AI for an answer. Stage two is using it to find gaps in your thinking. Stage three is showing it how you like work done, with examples and rules. Stage four is giving it a real assignment. Stage five is having it repeat the work on a schedule.
My read is that most owners sit at stage one. They type a question, get an answer, and close the tab. Stage three is where hours come back. Say you answer the same estimate questions every week. Write down how you answer them, paste in two or three real replies, and state your rules. Her description says to check the output and save the process. For stage four, it says to check every figure and piece of evidence before you trust the result. If a number goes to a customer, read it first.
A question for your builder
Nate Herk posted a video the same day. He says he tried more than 100 GitHub repos built for Claude and names six he calls the best. The video targets people building things for businesses, and its description lists Supabase and Stripe repos. A builder may wire tools like those into your systems.
I can’t tell you how his tests came out. The part that matters to you is a question to ask whoever builds your system: what do you use, and what happens to my data when you use it? A good builder answers in plain English.
Stage five pays off when stages one through four have already produced work you would send.