3 min read
meta ai neurodivergence

AI/CI/CD/ADHD/ACDC

It started as a tagline: AI/CI/CD/ADHD. Three slashes, one thesis. Then I said it out loud a few too many times and my brain — which has never met a pun it didn’t want to make load-bearing — heard “ACDC” hiding in there and refused to let it go. I loved classic rock long before I loved Terraform. Apparently the two were always going to end up in the same sentence.

So: AI/CI/CD/ADHD/ACDC. My whole personality, slash-delimited.

The actual thesis

Take away the joke and there’s a real claim underneath it — three things that, for me, aren’t separate at all:

  • AI is a focus multiplier. Agentic coding turned scattered ideas into shipped workstreams. I haven’t handwritten a line of code in a long time, and I build faster because of it — not despite it.
  • CI/CD is externalized discipline. If it’s not automated, tested, and repeatable, it’s not done. Pipelines are the part of my brain I don’t have to hold onto manually anymore.
  • ADHD is the reason both of those matter this much. I’m neurodivergent by diagnosis and a systems-thinker by necessity — I build guardrails instead of relying on willpower, because willpower was never the tool that was going to work.

None of that is new for me. I’ve been wiring together Linux boxes and config files since I was a kid with a 486 and a stack of floppy disks. What’s new is having a collaborator that can hold the parts of a project my working memory drops, so I get to stay in the parts I’m actually good at: architecture, judgment, and knowing when something’s about to break.

Where the rock references were probably always going to show up

I grew up on the same hardware and the same radio dial — Atari and NES on one side, classic rock on the other, both playing in the background while I figured out how to hand-edit a config file to get a floppy-disk game to boot. AC/DC, Pink Floyd, Aerosmith — that stuff was the soundtrack to learning how systems work by taking them apart. Turns out that’s still exactly what I do for a living, just with more YAML.

What it looks like in practice

Right now that means building internal tooling that lets engineers actually share AI agent context — prompts, skills, workflows — instead of everyone reinventing the same automation in their own corner. It’s the same three pillars at work scale: AI doing the multiplying, CI/CD doing the discipline, and ADHD-shaped thinking doing the “why does nobody have a shared place for this yet” part.

AI/CI/CD/ADHD/ACDC. Dirty deeds, done dirt cheap, deployed via pipeline.