Pinsetter
A friend of mine invited me to go bowling with him on Saturday, and while I suck at bowling, the invitation sparked an epiphany. I think the ideal framing for how I want to use AI is like that of a “pinsetter”— a term I just discovered after a search that looked something like “what are those things called that reset bowling pins?” Yeah, those things.
I imagine everyone would hate bowling if they had to walk all the way down the lane and pick up their pins themselves. Some work feels like this. You spend so much time setting up the pins when you’re really there to knock them down. That’s a job for a pinsetter; that’s a job for AI. To state it plainly, AI is really good at organizing things in a way such that all you have to do is push the submit button most of the time. You’re still there to review, and you have the final say on the output. You remain a human in the loop, but AI can do the boring setup stuff, and you just execute.
As with my last post, I’ll use an example related to my current job hunt. I have a ChatGPT scheduled task set up to find new job postings that fit my qualifications, then prep application materials for each, including an ATS-optimized resume and ideas for short-answer prompts. All I have to do is double-check that it didn’t do something weird, like include only projects and no work experience. I occasionally add a cover letter if I’m really gunning for the role. I put in the work up front to tell ChatGPT how to do the task and where it can source information from (mostly Notion), and it sets it up on every weekday.
The same concept can be applied to things you want to learn, which is where I’m interested in playing around with the idea more. Repetition is crucial for practice, so what better way to do it than to have AI throw as many reps at you as you want, with no time lost to prep and the challenges that grow with you. That way, you can stay in that Goldilocks zone of difficulty.
I think the important difference between this “pinsetter” concept and other AI workflows is how you design yourself to remain the focal point of the work. Other workflows are designed to remove the human bottleneck, but I find that unrealistic for something like job applications where I must still be the face of the work. When I land an interview, I don’t want to be trying to catch up on whatever my agent wrote down under “Why do you want to work at [insert company name]?”
At the end of the day, it's a system with way less friction, but I still maintain direct contact with the work. Every time a model is released, I think it may not be necessary anymore, but that feeling doesn't last long. Moving forward, this is the way I’m using AI.