Same draft, 4 AI-slop humanizer skills: they work at different levels and undo each other

bibryam · x · 2026-09-13

Bilgin Ibryam ran the same draft through 4 popular no-slop/humanizer skills and got four very different outputs. His takeaway: they don't compete — they operate at different altitudes. clarity reads the whole piece, no-ai-slop reads paragraphs, humanizer and stop-slop read sentences. Pick one or two at most; stacking means each undoes the last one's edits. Addy Osmani's clarity stood out: 15 passages, only -10 words, and the only skill that diagnosed the piece first — spotting that all 7 sections closed on the same cadence ("X is mature, but Y remains Z"), which none of the others caught. A full side-by-side breakdown is coming to his Substack, The Generative Programmer.

Related event: One Draft, Four Anti-Slop Skills: A Hands-On Comparison Shows They Work at Different Levels(6 posts)→

Original post →

More from coding & agent

coding & agent channel →