Free · Hook + sections + b-roll · Shorts / Standard / Long
Describe your topic and audience. Pick length, tone, and hook style. We return a primary hook + 2-3 A/B alternates, sections with b-roll cues, pacing notes, and SEO tags.
See how it works — click any example
Five tones — educational / entertaining / conversational / informative / energetic. Six hook styles — curiosity / stat / question / bold-claim / story / controversial. Every script ships with a primary hook plus 2-3 alt hooks in different styles, ready for YouTube Studio's built-in A/B test.
Hook variants (A/B), intro, 3-7 timestamped sections with talking points and b-roll suggestions, engagement cues, outro, 3-6 pacing notes tuned for 2026 retention curves, title alternates, thumbnail caption, 5-10 SEO tags, and demonetization flags. Pair with the description generator for the full publishing flow, or summarize an existing video for research.
Niches we don't recognize — add topic context so the model has something to ground to. Scripts that invent facts — if a section needs a stat you didn't provide, we leave a bracketed [fill in] placeholder. YMYL topics get flagged (health, finance, politics) for advertiser restrictions before you record.
Free with an account. Sign up for 30 runs/month (~10 scripts). Upgrade to Pro for 1,500/month (~500 scripts).
Brief: why most people fail IF in the first 2 weeks — 4 hidden mistakes + 3-rule starter protocol. Audience: 25-45, generally healthy, tried IF once, not medical patients. Primary hook style: bold-claim. Here's what the result page surfaces:
"The strangest thing about fasting: the people who eat normal at dinner lose more fat than the people who go strict. For two weeks I'll show you why — and the three rules that actually work."
Why this works: bold-claim opens a knowledge gap by directly contradicting viewer expectation. Concrete ("two weeks", "three rules") sets the payoff contract. No "hey guys, welcome back" opener — the first 15 seconds earn the retention.
Mistake #2: Day-1 window too aggressive
Purpose why 60% of IF attempts die in the first 72 hours
Talking points: most beginners jump to 16:8 day one → ghrelin spike at 10am → by day 3 the decision fatigue breaks willpower → 12:12 for week 1, 14:10 for week 2, 16:8 only after body adapts → "easy > optimal" on day 1 → typical drop-off curve graphic. B-roll: ghrelin line chart, split-screen week 1 vs week 2, kitchen clock fast-cut at 10am.
"Re-hook at 4:30 with a callback to the opening contradiction"
Why: 9-minute video hits its retention cliff at the 50% mark (~4:30). Re-tie to the opening hook ("remember the people who ate normal at dinner?") before introducing mistake #3 — viewers who made it this far are high-intent; re-hook holds them through the protocol payoff.
3 variants in curiosity, question, and story style
Why this matters: YouTube Studio A/B tests thumbnails AND hook lines automatically — queueing 3 stylistically-distinct hooks lets the algorithm find the highest-retention opener for YOUR specific audience. One curiosity variant ("the mistake nobody tells you…"), one question ("why does eating less make you gain more?"), one story ("I coached 200 people through IF — here's what broke them"). Three swings, one winner.
Run also returned the full intro (~80 words), sections 1, 3, 4 with their own talking points and b-roll, outro with next-video teaser ("next week — the 3 supplements that actually matter"), 5 pacing notes, 2-4 alternate titles, thumbnail caption ("The 4 hidden IF mistakes"), 7 SEO keywords (intermittent fasting mistakes, if beginners, fasting protocol 2026, when to break fast, 16:8 vs 12:12, if fat loss, fasting willpower). Plus 1 compliance flag: "Health claim territory — flagged for Made-for-Advertisers category, ok for channel if disclaimed."
Four things. First, structured output built for how creators actually work — hook + intro + timestamped sections + outro + pacing notes + thumbnail caption + title alternates + SEO tags in one pass, not a wall of prose you have to carve up. Second, hook A/B built in — every script ships with a primary hook AND 2-3 alternate hooks in DIFFERENT styles (curiosity / stat / question / bold-claim / story / controversial) so you can actually A/B test on YouTube itself, not guess which opener works. Third, retention-aware pacing — each script gets 3-6 specific pacing notes that target the 30-second and 3-minute retention cliffs ("Re-hook at 4:30 with a callback to the opening question — this is where long-form videos lose 20% retention"), not generic "speak slowly" advice. Fourth, compliance flags built in — if your hook uses a stat you didn't provide, or the topic crosses YMYL / demonetization territory, we flag it before you record instead of after Content-ID claims your revenue.
All three lengths. Short (<60 seconds, YouTube Shorts / TikTok / Reels) — no intro block, no bloat; hook → single payoff beat → one CTA. The script is 100-180 words total, 2-3 sections max, b-roll suggestions stay simple, and pacing notes specifically target the first 3 seconds (where Shorts live or die). Standard (5-10 min, classic creator length) — full intro with value promise, 3-5 sections, 1-2 engagement cues, pacing notes for the 30-second cliff. Long (15-30 min deep-dive) — 5-7 sections with deeper b-roll + graphics suggestions, multiple engagement cues, multi-CTA outro, pacing notes for the 3-minute, 8-minute, and 15-minute retention cliffs. Pick the length that matches your channel's AVD target — we adapt every section accordingly.
The 2026 YouTube algorithm makes the first 15 seconds decide whether a video is promoted at all — if viewers abandon in the first 15 seconds, CTR stops mattering and the video dies. We write hooks against this reality. Three disciplines: (1) "hey guys, welcome back to my channel" is a hard-banned opener — viewers click off the moment they hear it; we get to the payoff in the first sentence. (2) The primary hook is 15-40 words, readable in 10-15 seconds aloud, and is SPECIFIC to your topic — not "you won't believe what I found", but "the people who eat normal at dinner lose MORE fat than strict fasters." (3) The alternate hooks use DIFFERENT styles from the primary — never 3 curiosity hooks. So you record 2-3 versions of the opening, run them through YouTube's A/B thumbnail+hook test, and let the algorithm tell you which one holds retention. This is the single highest-leverage thing you can do for a growing channel and most scriptwriters skip it.
Curiosity — "You'll never believe…" — opens a knowledge gap, works on informational + story videos, safest default. Stat — "97% of people get this wrong." — specific number, only use if the stat is REAL (in your brief or widely-known public data); if fabricated, we flag it rather than publish. Question — "What if I told you…" / "Why does X always Y?" — opens an unresolved question the rest of the video answers, excellent for educational + explainer. Bold-claim — "This one trick beats every diet." — contrarian statement, highest engagement, RISKY for demonetization if in YMYL (health / finance / politics) so we flag it there. Story — "Last Tuesday at 3AM…" — narrative opener, HIGHEST retention for long videos but slower to payoff; use only with standard or long, never Shorts. Controversial — "Hot take: X is killing Y." — polarizing, highest engagement AND highest risk of getting demonetized or brigaded; we flag for YMYL. The ALT-HOOK DIVERSITY rule means your 2-3 A/B variants always mix styles, so you get a data-driven read on which style fits your channel.
Every script ships with complianceFlags (0-5) — specific warnings about advertiser restrictions and Content-ID risks BEFORE you record. The 9 categories we watch for: health/supplement/weight-loss claims (Made for Advertisers category strictness), mental-health topics (monetization restricted without care + disclaimers), politics/elections (partisan = demonetized, community guidelines risk), financial outcome claims ("I made $10k", "beat inflation" — advertiser-restricted), alcohol/cannabis/tobacco (regional restrictions), graphic/accident/true-crime content, crypto/NFT promotions, copyrighted clip or music reference (Content-ID will redirect ALL revenue), reaction content using large portions of another creator (fair-use dispute risk). We don't moralize — we just tell you which revenue risk you're taking so you can either (a) adjust the hook or (b) record it knowing ads will be limited. The script also hard-bans the phrases that trigger YouTube's clickbait detection: no fake urgency ("channel will be deleted"), no fake scarcity ("limited time only"), no subscribe-cue spam. Never fabricates stats or named sources either — if a section needs a stat you didn't provide, the script leaves a bracketed [fill in: specific number] placeholder.
Yes — every section comes with a TITLE (usable as a YouTube chapter marker), a TIMESTAMP in M:SS or MM:SS format, a one-line PURPOSE, 3-8 TALKING POINTS (these are the BEATS, not verbatim lines — you bring the voice, we bring the structure), 1-4 B-ROLL / visual suggestions, and 0-2 ENGAGEMENT CUES placed at natural payoff moments inside the section (not in a separate "subscribe" block). Shorts get 2-3 sections covering the whole <60s arc. Standard (5-10 min) gets 3-5 sections with timestamps distributed across the total runtime. Long (15-30 min) gets 5-7 sections with deeper talking points and more b-roll. You paste the timestamps straight into your YouTube description to get automatic chapters, which boosts AVD (average view duration) by 2-4% on explainer content because viewers can skim-watch to the part they want instead of bouncing.
Every run ships with a primary title (10-100 chars, tuned for the CTR-vs-clickbait line), 2-4 alternate titles with DIFFERENT angles (one benefit-led, one curiosity, one contrarian, etc. — not 4 variations of the same) so you can A/B test titles on YouTube Studio, a 3-7 word thumbnail caption idea (the punchy text that goes on the thumbnail image — not the title), and 5-10 SEO keywords (short phrases, lowercase, ranked for the video's tag suggestions). Title alternates matter: YouTube's Studio A/B tests different titles automatically, so when you upload with 4 strong alternates queued, you let the algorithm find the highest-CTR option across your actual audience instead of guessing. Thumbnail caption pairs with the title — it says WHAT, not WHO or WHY. SEO keywords become your video tags and channel-keyword optimization signals.
Works for both. The script is structured as a teleprompter / outline: TALKING POINTS are beats for YOU to fill with your voice, not verbatim sentences to read flat. For faceless channels (AI-voiceover, screen-recording tutorials, stock-footage compilations, news roundups) — you use the talking points verbatim if you want, paste them into your TTS pipeline, and the b-roll suggestions tell you what stock footage or screen captures to pair. For on-camera creators — you internalize the beats, bring your own voice, and the b-roll cues tell the editor where to cut. The pacing notes work for both (the 30-second cliff is the same whether the face is yours or an AI avatar's). SHORTS DISCIPLINE is especially important for faceless Shorts channels — we target aggressive hook + 1 cue max + no intro block so the TTS output stays under 60 seconds every time.
Standard 5-10 min scripts generate in 25-40 seconds on our Haiku-4.5 backend (that's the speed/quality sweet spot for structured longform). Long-form 15-30 min scripts take 45-80 seconds because the section count and word count roughly double. Shorts take 15-25 seconds. If the script misses on tone or length, re-run with the same topic and a different preset — results cache for 7 days so you can also share a URL with co-creators or editors before recording. If the model returns invalid JSON or violates a schema rule (rare — <2% of runs), you get a clear MODEL_OUTPUT_INVALID error and a free re-run. If your topic is genuinely un-scriptable (pure spam, random characters, hard-banned content), you get a POLICY_REFUSED or UNUSABLE_CONTENT flag with a reason — no silent failures, no hidden charges.
First 3 runs per day are completely free — no signup, no credit card, full access to the tool including all lengths (Shorts / standard / long), all 5 tones, all 6 hook styles, hook A/B, pacing notes, compliance flags, title alternates, SEO keywords. A free account bumps you to 30 credits/month; each YouTube-script run costs 3 credits, so 10 full script generations per month on the free tier. Creators who push 2-3 videos/week should look at Pro (1,500 credits/month = ~500 scripts). Every result URL is saved for 7 days on the free tier (indefinitely with a free account) so you can share a script with your editor, voiceover artist, or sponsor for review before recording. Each result page also cross-links to our YouTube Description Generator and YouTube Summarizer so your full post-publish workflow — script → description → summary for email/socials — runs in one place.