Cheatsheet · AI prompts

AI prompts for X content (12 tested)

Quick-reference card for AI prompts that produce X-native content. Tweet generation, thread expansion, hook iteration, voice matching.

Weekly playbook

Get more guides like AI prompts for X (Twitter) content cheatsheet by email

Drop your email for the weekly X growth playbook — tactics like this one, every Sunday. This page is bookmarkable/printable directly from your browser for offline reference right now.

We'll never share your email. Unsubscribe with one click.

Generic AI prompts produce generic content. The 12 prompts below are X-tuned — specific format constraints, hook patterns, length bounds. Copy + adapt for your specific topic.

Single-tweet prompts

Topic to single tweetWrite a single X tweet (under 280 chars) about [topic]. Use a [hook pattern]. Include 1 specific number or example. No links, no hashtags.
Contrarian takeWrite a contrarian X take on [common belief in topic]. 200-280 chars. Defend the contrarian position with specifics.
Personal-stake postWrite an X tweet about a personal decision in [topic area]. Format: [Action I took]. [Specific number / cost]. [What happened].

Thread prompts

Topic to threadWrite a 7-tweet X thread on [topic]. Hook tweet with contrarian opener. 5 setup/payoff tweets in middle. Closer tweet ties back to hook.
Framework threadWrite a 5-tweet X thread explaining the [framework name]. Tweet 1: hook + framework name. Tweets 2-4: one component each. Tweet 5: when to use it.
Story threadTurn this experience into a 9-tweet X thread: [paste experience]. Tweet 1: hook. Tweets 2-7: chronological story. Tweets 8-9: lesson + general principle.

Hook iteration prompts

5 hook variantsHere's a tweet draft: [paste]. Write 5 alternative hooks (first lines) using 5 different patterns: contrarian, specific-number, curiosity-gap, personal-stake, framework-intro.
Hook sharpeningHere's a tweet hook: [paste]. Make it sharper — same idea, fewer words, more punch. Output 3 variants.
Hook contradiction testHere's a hook: [paste]. Argue against it. The strongest counter-argument shows me where the hook is weak; rewrite the hook to defuse it.

Voice matching prompts

Voice analysisHere are 5 of my tweets: [paste]. Describe my writing voice in 5 attributes (e.g., 'short sentences', 'concrete examples'). Use these for future tweet generation.
Voice cloningUsing the voice profile [paste profile], write a tweet about [topic]. Match the voice exactly: sentence length, vocabulary, hook style.
Anti-voice checkHere's a generated tweet: [paste]. List 3 things in it that DON'T match my voice (in the previous voice analysis).

Common questions

Which AI model is best for X content?+

GPT-4 class and Claude 3.5 class models produce X-quality content with the right prompts. Smaller open-weight models produce more generic output. For production use, paid tier of either GPT or Claude. For free use, Groq's free-tier inference is the strongest free option.

Will AI-generated tweets sound robotic?+

Only if the prompts are generic. The 12 prompts above include hook patterns + format constraints + voice anchors — which produce content that doesn't read as AI. Bad prompts produce bad AI tweets; good prompts produce competitive-with-human content.

How do I keep AI-generated tweets in my voice?+

Use the voice analysis + voice cloning prompts. Feed the AI 5-20 of your best tweets, ask it to extract a voice profile, then condition all future generations on that profile. AutoTweet does this automatically via the Voice Training feature.

AutoTweet bakes these patterns in

Optimal times, hook patterns, content shapes — AutoTweet applies the cheatsheets to your content automatically. Try free at /try — no signup.

Cancel anytime