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.
Printable PDF
Get the AI prompts for X (Twitter) content cheatsheet as a printable PDF
Drop your email — we'll send the PDF version (printable + shareable) plus the weekly X growth playbook each Sunday. Unsubscribe anytime.
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 tweet | Write a single X tweet (under 280 chars) about [topic]. Use a [hook pattern]. Include 1 specific number or example. No links, no hashtags. |
| Contrarian take | Write a contrarian X take on [common belief in topic]. 200-280 chars. Defend the contrarian position with specifics. |
| Personal-stake post | Write an X tweet about a personal decision in [topic area]. Format: [Action I took]. [Specific number / cost]. [What happened]. |
Thread prompts
| Topic to thread | Write 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 thread | Write 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 thread | Turn 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 variants | Here'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 sharpening | Here's a tweet hook: [paste]. Make it sharper — same idea, fewer words, more punch. Output 3 variants. |
| Hook contradiction test | Here'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 analysis | Here 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 cloning | Using the voice profile [paste profile], write a tweet about [topic]. Match the voice exactly: sentence length, vocabulary, hook style. |
| Anti-voice check | Here'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 models (Llama 8B etc.) produce more generic output. For production use, paid tier of either GPT or Claude. For free use, Llama 70B via Groq 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.