How I Use AI, Actually

My mentor Russ Unger has a name for something I feel constantly: working with AI mostly feels solo, and calling it a "team" flattens something real. That framing is his, not mine, from AI Doesn't Have to Be a Solo Sport. Until It Is. He also writes about the other half of this in The AI 75% Trap: the 75% keeps getting easier, and the 25% is where the work lives. My version, landed on through my own projects: AI gets you to a rough, working draft fast. Embarrassingly fast. Getting from that draft to something you'd put your name on is still the job, and no model does that part for you.

I test this against my own work, not just theory. This site went through a written plan, then every task got implemented, then reviewed twice (once for whether it matched the spec, once for whether the code itself held up) before I'd call it done. That second pass caught real bugs: a slider that snapped where it shouldn't, images that would've gone permanently invisible if a script failed to load. Nobody catches those by trusting the first draft.

That's the workflow at Agile Defense too. We build concept prototypes fast, sometimes in days, because the fastest way to find out if an idea holds up is to put it in front of someone who actually knows the problem. AI gets the prototype into someone's hands quickly. My job (and my team's) is reviewing what comes out before it goes anywhere near a client, same judgment, just applied earlier.

Land to Land Holdings runs the same principle at a smaller scale: AI drafts the listing copy and social content, and I read every word before it goes out. That's the actual workflow. Nothing ships without a human, me, reading it first.