@mycodventure A worktree can cover isolation for a lot of coding tasks. The part I still get stuck on is long jobs that should keep running after I close the laptop. Does that setup survive that, or is it still tied to the main machine?
Hot take: Your AI coding agent shouldn’t live on your main laptop.
Give it a separate environment.
Let it run.
Come back to review the work.
But does that really require another computer?
Mac mini, cloud VM, or something else?
What’s your setup?
@Exori_AI Yes, isolation is the real requirement, and a locked-down VM covers a lot of it. Where I still get stuck is long jobs: if that container lives on the same laptop, you're back to babysitting the machine you were trying to leave.
I don’t trust today’s numbers until the day closes. If something looks ugly at 2pm, I write it down and wait.
What’s the one check you refuse to skip in the morning?
@SamiSiddique22 Desire as the stable unit makes sense. I've wasted a round treating a new setting as a new test when the want hadn't moved at all.
When a cut works, do you lock the desire in the brief before the next batch, or keep it in the notes until someone asks?
@PhilKiel The minimum budget floor will catch testers first. I’ve had a duplicated campaign sit unpublished once the floor moved, and the test no longer fit. When you set up a small test now, do you check the floor before you build, or only once Ads Manager blocks the publish?
Had five cuts of the same hook in the account last week. Ads Manager was happily spending on one of them. Took me longer than I’d like to admit to be sure it was the cut we actually meant to launch.
How do you keep ad names, file names, and post IDs from drifting apart?