Reply prompts · Earn follows, not just engagement

10 AI prompts for replies that grow your account

Replies are 27× a like in X's algorithm. The right reply on a large account's thread earns 50-500 new followers. These 10 prompts generate substantive replies — not 'great post!' filler — that demonstrate competence and convert profile-clicks to follows.

Most AI-generated replies are useless ('great insights!', 'this resonates!'). The prompts below structure the AI to produce 4 patterns proven to earn follows: insider expansion, specific counter-data, the practical missing step, and respectful pushback. Variables in {curly braces}.

The prompts

Prompt #1

Insider expansion reply

Generate a reply to this tweet: '{original tweet}'. Reply structure: 1) Establish insider credibility in 5-10 words ('After 4 years as X' or 'Worked at Y for 3 years'). 2) Add ONE specific piece of information the OP didn't mention. 3) End with a 1-sentence implication or actionable. Max 240 characters. Voice: matter-of-fact insider. Avoid 'great post' / 'this resonates' openers.

Why it works

Insider expansion replies are the highest follow-converting reply type. The 5-10 word credibility establishment is short enough to not waste space; the 'one specific piece' filter prevents the AI from listing 5 generic additions.

Best for

Claude (best at insider tone).

Variables to fill in

{original tweet}

Prompt #2

Specific counter-data reply

Generate a reply to this tweet: '{original tweet}'. The OP makes a claim you partially disagree with. Reply structure: 1) Acknowledge what's right. 2) Add a specific number, observation, or example that refines the claim. 3) End with a more nuanced framing. Max 240 chars. Voice: respectful pushback, not contrarian-for-contrarian-sake. Don't fabricate statistics.

Why it works

Counter-data replies that respect the original earn engagement from both audiences (OP's followers + observers seeing tactical depth). The 'don't fabricate' guardrail is essential.

Best for

Claude (best at the respectful pushback tone).

Variables to fill in

{original tweet}

Prompt #3

Practical missing-step reply

Generate a reply to this tweet: '{original tweet}'. The OP teaches something but skips the practical 'how to start' step. Reply structure: 1) Affirm the teaching briefly. 2) Add the practical missing step the OP didn't mention (the 'how do I actually start tomorrow' detail). 3) End with a brief example. Max 240 chars. Voice: helpful, specific, not preachy.

Why it works

Missing-step replies are highly bookmarked. Readers of educational threads often ask 'OK but how do I actually start?' — your reply that answers that earns saves + follows.

Best for

GPT-4 (good at practical specifics).

Variables to fill in

{original tweet}

Prompt #4

Respectful contrarian reply

Generate a reply to this tweet: '{original tweet}'. You disagree with the OP's main point. Reply structure: 1) Polite acknowledgment of OP's reasoning. 2) Your specific dissent with one piece of evidence. 3) The condition where OP would be right. Max 240 chars. Voice: confident but not aggressive. The condition where OP is right is critical — it shows nuance.

Why it works

Respectful contrarian replies on hot-take threads earn 100s of follows because they signal high-quality thinking. The 'condition where OP is right' framing prevents pile-on dynamics.

Best for

Claude 3.7+ (best at nuanced disagreement).

Variables to fill in

{original tweet}

Prompt #5

Diagnostic + invitation reply

Generate a reply to this tweet: '{original tweet}'. OP shares a problem they're facing. Reply structure: 1) Empathy in 1 sentence. 2) Diagnostic framework — 3 specific causes this could be, presented as a numbered list. 3) Invitation to continue the conversation. Max 240 chars. Voice: helpful, not condescending.

Why it works

Diagnostic replies on problem-sharing threads earn the OP's reply (which boosts your reach further) + observer follows (who see your tactical depth). The 'invitation' raises continuation rate.

Best for

GPT-4 (best at structured frameworks).

Variables to fill in

{original tweet}

Prompt #6

Specific example reply

Generate a reply to this tweet: '{original tweet}'. OP makes a general point. Reply structure: 1) Affirm with 'Specifically:' or 'Concrete example:' 2) Provide ONE specific real example (a company, a situation, a specific decision). 3) Tie it back to OP's general point. Max 240 chars. Voice: declarative, not opinion-led. Real examples only; don't fabricate cases.

Why it works

Specific examples on general-advice threads earn engagement because they make abstract advice concrete. Real-examples-only constraint prevents the AI from generating plausible-sounding fictional cases.

Best for

Claude (best at the 'don't fabricate' constraint).

Variables to fill in

{original tweet}

Prompt #7

Forward-look insider reply

Generate a reply to this tweet: '{original tweet}'. OP celebrates a milestone. Reply structure: 1) Genuine congratulation in 1 sentence. 2) Insider knowledge of what's coming next (the 'thing nobody tells you' about that stage). 3) Brief actionable advice. Max 240 chars. Voice: warm but not sycophantic. Knowledge must be specific to that stage.

Why it works

Forward-look replies on milestone threads earn warmth + respect (you're sharing instead of competing). The 'thing nobody tells you about this stage' framing makes the reply actually useful, not just congratulatory.

Best for

Claude (best at the warm-but-substantive blend).

Variables to fill in

{original tweet}

Prompt #8

Pattern-recognition reply

Generate a reply to this tweet: '{original tweet}'. OP shares an observation. Reply structure: 1) Affirm with 'I see this pattern too' or similar. 2) Provide the pattern as you've seen it — the common element across multiple cases. 3) End with the implication. Max 240 chars. Voice: experienced observer. Reference real (or generally-credible) cases.

Why it works

Pattern-recognition replies signal experience without bragging. The 'common element across cases' framing is more useful than just adding one more anecdote.

Best for

Claude 3.7+ (best at synthesis).

Variables to fill in

{original tweet}

Prompt #9

Reframe-the-question reply

Generate a reply to this tweet: '{original tweet}'. OP asks a question or frames a problem. Reply structure: 1) Briefly acknowledge their framing. 2) Suggest the question they should be asking instead. 3) Briefly explain why this reframe matters. Max 240 chars. Voice: thoughtful, not condescending. The reframe should be one they'd find genuinely useful, not just contrarian.

Why it works

Reframing the question is a higher-order signal than answering it. Earns follows from observers who learn to think differently from your reply.

Best for

Claude (best at reframes).

Variables to fill in

{original tweet}

Prompt #10

Cost-and-decision reply

Generate a reply to this tweet: '{original tweet}'. OP asks about a major decision (career, business, relationship). Reply structure: 1) Briefly share your relevant experience (with specifics). 2) Provide a concrete framework or condition for the decision. 3) End with a warning about the most common mistake. Max 240 chars. Voice: honest, lived-experience, not preachy.

Why it works

Decision-framework replies on big-life-question threads convert at high rates because the reader is in 'I need to decide' mode. The framework gives them tools; the warning gives them wisdom.

Best for

Claude (best at decision framing).

Variables to fill in

{original tweet}

Common questions

How many replies should I post per day to grow on X?+

10-20 substantive replies per day. 'Substantive' means 2+ sentences adding real value, not 'great post!' filler. Reply discipline is the most underused growth tactic — replies are 27× a like in X's algorithm, and the right replies on large accounts earn 50-500 followers per reply moment.

Should I use AI to generate replies?+

Yes, but with heavy editing. The prompts above are structured so AI produces the framework; you add the personal specifics (your real experience, your real numbers, your actual industry). Pure AI replies without editing read as generic and don't earn the algorithm-boost or the follow-conversion. Plan to spend 30-60 seconds editing each AI-generated reply.

What accounts should I reply to for growth?+

Accounts 5-50× your size in your niche. Replying to mega-accounts buries your reply under 1,000 others; replying to peer-sized accounts doesn't borrow reach. The sweet spot is the next tier up — large enough to give you reach, small enough that your reply surfaces. Build a private X list of these accounts and reply daily from it.

Will substantive replies actually drive follows?+

Yes, when they demonstrate competence. The pattern: large-account thread → substantive reply showing depth → readers click your profile → strong bio + recent quality posts → follow. The reply alone doesn't drive the follow; the reply + profile-arrival experience does. Optimize your profile first; replies amplify what's already there.

Skip the prompt engineering

AutoTweet's AI generates tweets in your voice using these structural patterns automatically. No prompt-tweaking required. 14 voice-matched tweets queued the moment you connect X.

Cancel anytime