Click the snack to open the X post and just read the post text (2 minutes, tops) — skip the longer talk today. Then open any chat tool and paste: 'In 3 bullets: what is the difference between a one-shot chat prompt and a goal_draft_critique loop with a human stop?' You're done with 3 bullets that mention loops, memory, or self-critique.
How-to · WFL · WFL.4
X snack: tool loops beat one-shot chat (Anthropic talk seed)
Stop only one-shot prompting: a small tool loop (goal → draft → critique → re-run once) beats a single clever line — use it in chat tools you already have.
Worked example
Here's a viral lesson from X (no newsletter signup required): an Anthropic engineer made a point that sounds like real platform advice — stop relying on one-shot prompting, and use a loop that re-checks your drafts instead. In plain workplace terms: in the tools you already use, run goal _ draft _ critique _ one re-run, then stop for a human check. Remembering what failed beats endless fancy re-prompting. And you still own hitting send.