For all the hype and controversy around AI, I keep coming back to a simple thought: AI is still software. A machine. Not as deterministic as I’d like it to be, but still a machine.
So I try to think about it in very unromantic terms and use it for the kind of work machines are good at: cleaning up structure, keeping formats and metadata consistent, turning loose instructions into organized documents, and doing the boring connective tissue work between apps.
Inspired by the now-obsolete Arc Max feature that renamed your downloaded files, I treat AI as a maintenance worker.
Thanks to Codex, my Obsidian frontmatter is finally in order. It can create Obsidian Bases from a simple request, clean up formatting across multiple docs, rename things consistently, and turn scattered notes into something structured enough to use later.
Format one doc and sync the same style across a dozen others? Done.
Create a Google Sheet with a specific structure and formulas? Yes, please.
I’m not asking it to create meaning. I’m not asking it to do the creative part for me. That’s the part I still want to own.
AI is more like Alfred for my digital Batcave: keeping things in order, fixing what’s messy, preparing what I need, and making sure the whole place doesn’t fall apart while I’m doing the actual work.
And the best thing about this approach is that it doesn’t depend on the biggest, most expensive models forever.
When the era of subsidized tokens is over, and frontier models are available only to the rich and powerful, open-source and, hopefully, local models should still be able to do this kind of maintenance work just as well.