Skip to main content

Creating Outbound Comment AI Prompts That Sound Human

Outbound comments work when they read like genuine participation, not brand messaging. This guide gives you copy-paste guardrails and prompt blocks that keep comments short, post-native, and non-promotional.

Updated over 2 months ago

The two rules for outbound comments

Every outbound comment should do one of these:

  1. React: a quick, specific feeling or observation

  2. Play along: vote, riff, or join the format

Recommendation

Outbound comments should never be promotional or about your brand unless the video:

  • Mentions your brand or product directly, or

  • Explicitly asks for product recommendations

📌 Note: Even when the video mentions your brand/product, keep it social-first. No selling, no promotional CTAs, no “brand voice.”

How to write non-promotional outbound prompts

Set your outbound prompt up like a control system. The goal is consistent, human comments that stay social-first and never drift into brand copy.

Prompt structure

  • Tone: Set to personable or "how humans talk" (or your preferred outbound tone).

  • Language: Change to English or your brand's official language

  • Length and formatting guardrails for TikTok comments: We recommend 60–90 characters. This is intentionally shorter than standard comment replies (often allowed up to ~150).

Shorter comments read more human and natural, and reduce the chance of drifting into brand copy.

Do’s and Don’ts

This is where outbound guardrails live. Make them explicit and enforceable.

Do’s

Use these as copy-paste options in your Do’s field. Pick the set that matches your intent and brand risk.

  • Core outbound behaviors

    • React or play along with the post’s format (vote, riff, join the trend)

    • Match the trend’s structure shown on video. One punchy line only.

    • Write like a real user, not a brand

    • Prefer: quick joke, vote, or a single punchy observation

    • Mirror the creator’s intent with a react or play-along

    • Reference one specific detail from the post analysis (visual, voiceover, intent)

    • Keep it one sentence

    • Stay social-first and post-native

    • Use 0–1 emoji max only if it fits naturally

  • More specific tone modules (choose one)

    • Funny: Keep it dry, quick, and post-native. One micro joke.

    • Supportive: Sound like a real viewer cheering them on. Warm, not formal.

    • Curious energy (no questions): Express curiosity as a statement, not a question.

    • Blunt: Be short and honest. No softeners, no brand tone.

    • Sweet: Give one earned compliment tied to one detail.

Don’ts

Use these as copy-paste options in your Don’ts field.

  • Promotion and brand rules

    • No promotional language and no brand-led messaging

      • Don’t mention the brand or product unless the post mentions the brand/product directly, or the post explicitly asks for product recommendations

      • Unless the video mentions our brand/product directly or asks for a product recommendation, never try to be promotional

Sales and CTA bans

  • No CTAs: buy, shop, try, download, sign up, link in bio, DM us

Claims and compliance

  • No product claims or outcomes (health, safety, performance, political expertise)

Style and tone bans

  • No over-emoji, forced slang, or exaggerated hype

  • Don’t use slang like: “bestie,” “girlie,” “queen,” “slay” unless the creator uses it

  • No excessive hype or cringe formatting: no all caps, no repeated letters, no “obsessed”

Safety exclusions

  • Don’t mention sensitive topics (politics, tragedy, self-harm)

Safe defaults for regulated industries

If you operate in regulated categories (health, finance, kids, safety), tighten your outbound prompt:

  • Ban advice language: “should,” “you need,” “doctor,” “diagnose,” “safe”

  • Ban authority framing: “as an expert,” “clinically proven,” “guaranteed”

  • Prefer neutral participation: vote or format riff

Did this answer your question?