AI Prompts Library

Write a kinetic typography promo video script for {ANNOUNCEMENT/PRODUCT/EVENT}, {15/30}s, for
{PLATFORM}.
Rules of the format — the text is the star:
1. THE SCRIPT: 25–45 words TOTAL, broken into 8–14 screen "beats" of 1–5 words each. Each beat = one screen. Write for rhythm: vary beat length (long-short-short-long), build tension, land the payoff word ALONE on screen.
2. EMPHASIS MAP: mark each beat — [BIG] for power words (full-screen, heavy weight), [fast] for connective phrases (quick flashes), [HOLD] for the beats that need a full second, [COLOR] for the 2–3 words in accent color.
3. MOTION NOTES: per beat — slam in / typewriter / slide up / scale punch. Never the same
transition twice in a row.
4. SOUND: where the beat drops, where a whoosh or click lands, where silence hits before the final word.
5. FINAL FRAME: logo + CTA + the one line that stays on screen 2 full seconds.
Give me 2 versions: hype energy and premium-calm energy.

Plan a cinematic walkthrough video of {PROPERTY / PRODUCT / SPACE / VENUE}, target length
{30–90}s, for {LISTING / SOCIAL / WEBSITE HERO}.
Deliver:
1. SHOT SEQUENCE (8–12 shots): the reveal order matters — start with the strongest
exterior/overview beauty shot, then a natural discovery path, save the "wow" feature for 60%
through, end wide. For each shot: duration, camera movement (slow push-in / lateral glide / reveal around corner / rise), and what feeling it sells.
2. AI-GENERATION VERSION: if I'm using AI video from stills — one paste-ready prompt per shot:
"Cinematic {interior/exterior} shot, [scene description], camera slowly {MOVEMENT}, {LIGHTING}, photorealistic, smooth stabilized motion, no morphing, no people, 16:9".
3. PACING & MUSIC: where cuts land (every 3–5s early, longer holds on hero shots), music mood, and the one moment to sync a cut to the beat.
4. TEXT OVERLAYS: max 4 (location/price/specs/CTA), timed.
5. PHONE FALLBACK: how to shoot each move with a phone gimbal if AI shots fail.

Script one course video lesson. Course: {COURSE TOPIC}. This lesson: {SPECIFIC LESSON}. Students: {LEVEL + WHO THEY ARE}. Target length: {5–8} minutes.
Structure (completion-rate engineered):
1. COLD OPEN (20s): the capability they'll have by the end, shown not promised ("By the end you'll be able to look at X and immediately know Y").
2. THE ONE THING: this lesson teaches exactly one core concept. State it, then explain via:
concrete example FIRST → the general principle → a second contrasting example.
3. THE MISTAKE: the error every beginner makes with this, and how to catch it.
4. DO-IT MOMENT (script a pause): a 2-minute exercise the student does NOW, with exact
instructions on screen. Lessons without a do-it moment don't stick.
5. RECAP: 3 bullets, then a 1-line bridge that makes the next lesson feel necessary.
Format as two columns: SPOKEN WORDS | ON SCREEN (slides, screen-recording moments, text overlays).
Tone: energetic teacher, not lecturer — short sentences, direct address, zero filler.

Write a 90-second whiteboard explainer video script for {PRODUCT/SERVICE/CONCEPT} aimed at
{AUDIENCE}.
Format — two columns, synced:
VOICEOVER (left): ~220 words total, conversational. Structure: meet a character with the
audience's exact problem (10s) → the problem gets worse / costs them something real (15s) → enter the solution, explained through the character USING it, not features listed (40s) → the
transformed after-state (15s) → CTA (10s).
DRAWINGS (right): for every voiceover line, what the hand draws — simple nouns and stick-figure actions ("frustrated figure buried under paper stack", "arrow from chaos-cloud to tidy
checklist"). Whiteboard rule: each drawing must be describable in under 8 words or it's too
complex.
Extras: name the character (relatability doubles); one visual metaphor that carries through the whole video; the single sentence viewers should remember.

Generate 10 AI b-roll video prompts for my content about {TOPIC}. Visual world: {AESTHETIC, e.g.
warm documentary realism / sleek tech minimal / cozy film look}. These clips will play under a
voiceover.
Rules for EVERY prompt (this is what makes AI video work instead of glitch):
- ONE subject + ONE clear motion per shot (AI video fails with complex action)
- Specify camera movement: slow push-in / drift left / static tripod / gentle orbit
- Specify lighting + time of day, consistent with the visual world
- 4–8 second clips, 16:9
- End each prompt with: "smooth motion, photorealistic, no morphing, no text"
Format each as:
SHOT {N} — {what script moment it covers}
PROMPT: {the complete paste-ready generation prompt}
Vary the shots: mix close-up details, medium human moments (hands, silhouettes — avoid clear
faces, AI mangles them), wide establishing scenes, and abstract texture shots. Keep one recurring color motif across all 10 so the edit feels intentional.

I'll paste a transcript of my YouTube video. Extract the FIVE most viral-worthy short clips.
For each clip give me:
1. START–END quote (the exact transcript lines to cut, 20–45 seconds of speech)
2. HOOK: 1-line on-screen text for the first frame (curiosity gap or bold claim)
3. WHY IT WORKS: one sentence — emotion, surprise, or utility
4. CAPTION: post text with 3 relevant hashtags for {PLATFORM}
5. TITLE: max 40 characters
Prioritize: strong opinions, surprising facts, before/after moments, and complete mini-stories.
Skip anything that needs context from elsewhere in the video.
TRANSCRIPT:
{PASTE}

Write a 60-second talking-head video script about {TOPIC} for {AUDIENCE}, to be delivered by an AI avatar.
Rules: 140–150 words total (natural 60s pace). Structure: relatable question or surprising fact (first 5s) -> the core insight explained with ONE concrete example -> three rapid-fire practical tips -> memorable closing line + CTA to {ACTION}.
Style: spoken language, short sentences, contractions, zero corporate filler.
Add [PAUSE] markers where a beat lands, and [EMPHASIS] on the 3 words that matter most — avatar tools use these for realistic delivery.
End with: a 3-option list of on-screen titles for the video.

Create a complete 30-second vertical video plan for {PRODUCT/SERVICE} targeting {AUDIENCE} on
{TikTok/Reels/Shorts}.
Deliver in this exact format:
HOOK (0–3s): one line that stops scrolling — pattern-interrupt or bold claim, related to {PAIN
POINT}.
SCRIPT: voiceover text, max 65 words total, conversational, no jargon.
SHOT LIST: 6–8 shots with duration, camera direction (close-up/pan/etc.), and what's on screen —
written so an AI video generator OR a phone can execute each shot.
TEXT OVERLAYS: caption for each shot, max 5 words each.
CTA (final 3s): one specific action.
SOUND: suggested music mood + where a sound effect lands.