Here’s what’s actually happening on X.

Representative of Alex's growth threads — generating 30 hook variants with Claude, shipping the top 5 as thumbnails or first lines, and keeping only the winners.

“I don't pick my hooks anymore. I pick which ones to test.”

Generate 30 hooks for a <platform> post about <topic>. The promise is: <one sentence>. Use 5 archetypes, 6 hooks each: curiosity gap, contrarian, numbered list, personal story, shocking stat. Keep each under <char limit>. Don't explain — just the hooks.

Representative of Eli's templates for encoding brand voice into Claude projects — tone markers, banned phrases, example library, rejection criteria.

“If your prompt doesn't say what NOT to say, your brand voice is mush.”

Representative of Matt's workflow threads — using NotebookLM to ingest the week's AI news, Claude to draft scripts, and ElevenLabs for B-roll voice takes.

“The channel doesn't exist without three AIs doing three different jobs.”

Here are this week's sources: <paste brief>. Draft a 10-minute YouTube script in my voice. Structure: hook (15s), 3 stories (2m each), one takeaway, CTA. For each story include one on-screen graphic cue and one B-roll line. Do not invent facts; cite which source each claim came from.

Representative of Katie's playbook for AI-assisted drip campaigns — using LLMs to generate 30-day nurture sequences seeded with real customer interviews.

“The AI doesn't replace customer research. It scales the output of good research across 30 emails.”

Representative of Ryan's Product Hunt trend posts — AI tools dominating top-of-day launches, and the positioning patterns that break through.

“Every launch day is now 60% AI tools. Positioning against a specific painful workflow is the only signal.”

Representative of Dharmesh's threads on using AI to turn one long essay into dozens of platform-shaped posts — each one rewritten for the medium, not just truncated.

“Repurposing isn't copy-paste. It's rewriting for a new room.”

Rewrite the attached essay as a <platform> post. Conventions of that platform: <hook style, length, emoji use, line breaks>. Keep the core argument; change the shape. Produce 3 variants I can choose between. Do not summarize — rewrite.

Representative of Aleyda's breakdowns of how Perplexity, ChatGPT Search, and Google AI Overviews pick sources — and the schema + freshness signals that correlate with being cited.

“Brand mentions are the new backlinks. Consistency across the open web beats keyword density.”

Representative of Dickie's threads on using Claude to repurpose long-form into 10+ social posts without losing the voice — by feeding style examples as 'voice anchors'.

“Repurposing isn't summarizing. It's remixing with voice preserved. That's a prompt, not a feature.”

Representative of Glenn's case studies tracking traffic drops after AI Overviews rollouts — and the content patterns that survive (original data, expert quotes, specific how-tos).

“Generic content is dead. Original research and firsthand experience is what survives AI Overviews.”

Representative of Shaan's prep workflow — Perplexity Spaces preloaded with every recent guest interview, plus Claude for question generation tied to their actual history.

“Good interviews used to take 4 hours of prep. Now it's 30 minutes and the questions are sharper.”

Representative of Marc's marketing commentary — solo marketers running what used to be 10-person content teams via ChatGPT + Jasper + Ahrefs.

“The 10-person content team is a nostalgia artifact. One person with good AI hygiene outperforms.”

Representative of Brian's threads on content engines — how to use Claude for first drafts that don't trigger AI-content penalties, combining with human edit passes.

“AI-assisted is fine. AI-generated is dead. The edit pass is the job now.”

Representative of Marc's posts on ranking in both Google and ChatGPT/Perplexity answer boxes — the mechanics of getting cited as a source by the LLMs themselves.

“Citations in ChatGPT drive more traffic than page 1 of Google now. Optimize for both.”