What I Tried This Weekend (and what actually surprised me)

Spent part of my weekend running a little experiment I've been putting off: using Claude to build a full cold outreach sequence for an AI consulting pitch — not just the emails, but the follow-ups, the subject line variants, and a scoring rubric to evaluate them before sending.

The part that surprised me most: the scoring rubric step caught vague value props and weak subject lines I totally would've glossed over. The actual first drafts were solid, but running them back through a second pass made them noticeably better.

End result: 3 emails I felt genuinely good about sending. Time spent: maybe 45 minutes including all the back-and-forth.

Has anyone else built a review/feedback loop into their AI writing workflow, or do you mostly use first drafts? Curious how others are handling the "good enough vs. great" decision.

2 comments