That Weird AI Workflow Might Just Work
AI workflow experiments deserve curiosity before dunking, because today’s weird loop might be tomorrow’s normal development practice.
Rob Zolkos
Building Software. Lots of AI lately. Curious and creative. Sharing what I learn.
You can find me on Twitter, GitHub, or write me an email
AI workflow experiments deserve curiosity before dunking, because today’s weird loop might be tomorrow’s normal development practice.
A 60-second PR screencast proves the feature works and often reveals the rough edges tests did not catch.
A private GitHub Action can turn ONCE Campfire’s source ZIP into daily commits and readable diffs.
One `setup-node` setting can save a legacy GitHub Actions build from using the wrong Node version.
A five-minute morning huddle can do more for a team than another strategy deck ever will.
Every “we might need this later” line of code is a tiny tax on today’s feature.
The next feature should not be the easiest one; it should be the one that moves the business goal.
A stubborn bug became a lesson in pairing, documentation, onboarding, and why pride is expensive.
The inbox wins when it controls your attention, so this is a simple system for taking that control back.