Prose Style

Transitions and Connective Tissue

• Cut throat-clearing. Delete sentences that exist only to announce what you're about to say: "This should not surprise anyone," "It's worth noting that," "The implication is clear."
• Avoid hedging formulas. "It depends on where you sit," "regardless of which X you prefer," "the broader point stands" — these signal AI equivocation. Take a position or cut the sentence.
• Earn your transitions. "But," "However," "That said" are fine when they mark a genuine turn. If the next paragraph continues the same argument, drop the contrastive opener.
• Don't summarize before concluding. The conclusion should advance the argument, not restate it. If your final paragraph could serve as an abstract, rewrite it.

Word Choice

• Prefer concrete over abstract. "Subscription seats replacing headcount requisitions" is better than "a fundamental shift in resource allocation paradigms."
• Avoid AI-favored adverbs. "Fundamentally," "essentially," "ultimately," "importantly," "significantly" — cut these unless they carry genuine meaning. They rarely do.
• Watch for hollow intensifiers. "Incredibly," "extremely," "absolutely," "truly" — if the noun or verb needs propping up, choose a stronger noun or verb.
• Limit "landscape/ecosystem/paradigm" language. These words have become AI markers. Use them only when the technical meaning is precise.
• Cut fluff phrases. "It's worth noting that," "in terms of," "the fact that" — these pad word count without adding meaning. Find the verb. Say it.

Voice and Ownership

• Take ownership. "My title" not "the title." "I'll qualify" not "the title needs qualifying." First person signals the author stands behind the argument.
• Write as the author, not a commentator. AI defaults to observational distance ("one might argue," "it could be said"). State positions directly.
• Avoid distancing constructions. "What this means depends on where you sit" hedges. "Engineering leadership already knows this" takes a position.

Structural Tells

• Don't start with a question you immediately answer. "What does this mean for developers? It means..." is a pattern LLMs default to. State the claim directly.
• Avoid the "to be sure" sandwich. Stating a counterpoint only to immediately dismiss it reads as performative balance. Either engage the counterpoint seriously or omit it.
• Lists vs. narrative. Default to narrative prose. Use bulleted lists only for genuinely enumerable items. A blog post with more than one bulleted list probably needs restructuring.

Self-Check Before Finalizing

1. Could I identify this as AI-written from the first paragraph? If yes, rewrite the opening.
2. Do more than two paragraphs start with the same structural pattern? If yes, vary them.
3. Is there a sentence I included only because it "sounds professional"? If yes, cut it.
4. Does the conclusion say something the introduction didn't? If not, sharpen it.