I Almost Quit Claude. Then Opus 4.8 Dropped.

For the past few weeks, Claude was driving me up the wall.

Here’s the exact thing it kept doing. I’d hand it a big job — a real one, with a bunch of smaller pieces buried inside. And somewhere in the middle it would hit the hard part, decide it “wasn’t important right now,” punt it, and keep building. On a foundation full of holes. My writing, my content, my software — same story every time. I was this close to walking away from it.

Then Opus 4.8 dropped yesterday. And so far, that problem is just gone.

Subscribe now

I broke the whole thing down in the video above — real day-one testing, not a benchmark recap.

The short version:

It stopped being lazy.

It handles the hard part in the moment instead of punting it. It builds on something real. And it holds onto what matters without me telling it five times to remember.

It follows my system.

That’s the whole game when you run an operation on Claude. To prove it, I rebuilt my entire publishing workflow into a working skill — start to finish, one sitting, about fifteen minutes. It held every instruction without losing the thread once.

It takes initiative.

While I was working on something else, it suggested a feature for my own platform — then built it and shipped it into my live app on its own. Zero failures. I didn’t touch it.

Full honesty: it’s been a day.

A day feels incredible right up until week three when the shine wears off. So this is what I’m seeing right now, and I’ll tell you if it holds.

Watch the full breakdown above. Then go put it on your real work and see what changes.

—Alex

Follow me: Instagram | X