i vibecode: ai automations & workflows

I don't have a computer science degree. I'd never even written a line of code, and I didn't know what a CLI, Git, or even VS Code was.

This post is a collection of thoughts, learnings, and demos. Most of what I know about building automations came from three places:

1. Starting at the point of friction. I jump straight in. I like to start at the point of friction. On the IDArch team, that was manual workflows. Whether it was manually updating a data asset's schema definitions or mapping a Bloomberg data file to the fields in our system, finding the common pain point is how I figure out where to start ideating (get dirty where the work sucks!).

   Finding problems and trying to fix them right away was probably what taught me the most. If you're not spotting problems, slow processes, or workflows that could be more efficient yet, that's where I'd start: go looking for them.

2. Engineers reviewing my work. I'd vibecode a solution with AI, get it working, and then hand it to people who actually knew what they were doing. Their feedback ("this works, but here's why it'll fall over in production") taught me more than any course.
3. Breaking problems down. A big, scary problem like "automate this whole reporting process" is overwhelming. "Pull this file from this folder" isn't. Most of the work is turning one big problem into a list of small ones I can solve one at a time.

WIP