The AI Prompt That Builds 30 Video Ideas Fast

We ran 14 different ChatGPT prompts through our network in February 2026. Most produced generic garbage. One template delivered 30 usable short-form video ideas in under a minute, and we’ve used it every week since.

Here’s the exact prompt structure, why it works, and the three mistakes that make most AI-generated ideas sound like AI wrote them.

Why Most AI Video Idea Prompts Fail

The problem isn’t ChatGPT. It’s how you’re asking.

When you type “give me 30 TikTok ideas for fitness,” you get lists like “workout motivation Monday” and “common gym mistakes.” They’re not wrong — they’re just not specific enough to scroll-stop. TikTok’s 2026 algorithm prioritizes the first 1.2 seconds harder than ever. Generic angles die in that window.

Most prompt guides tell you to add context. That’s true, but incomplete. We tested adding niche, audience, tone, and platform. Results improved 15%. Then we added one more layer, and usable output jumped to 80%.

The Exact Prompt Template We Use in 2026

Here’s the structure. Copy it, fill in the brackets, and paste it into ChatGPT-4 or Claude:

You are a viral short-form video strategist. Generate 30 video ideas for [NICHE] that target [AUDIENCE]. Each idea must include a specific hook (first sentence), the payoff (what the viewer learns or feels), and why it works on [PLATFORM] in 2026. Format as a numbered list. Prioritize uncommon angles and emotional triggers over generic how-tos.

The magic is in “why it works on [PLATFORM] in 2026.” That forces the AI to reference current algorithm behavior instead of pulling from outdated training data. Plus, asking for the hook and payoff separately prevents vague ideas like “talk about meal prep.”

Example: Fitness Creator

Prompt: You are a viral short-form video strategist. Generate 30 video ideas for home fitness that target busy parents aged 30-45. Each idea must include a specific hook (first sentence), the payoff (what the viewer learns or feels), and why it works on TikTok in 2026. Format as a numbered list. Prioritize uncommon angles and emotional triggers over generic how-tos.

Output sample:

• Hook: “I stopped doing planks and my core got stronger.” Payoff: Viewer learns 3 floor exercises that burn more than planks in less time. Why it works: Contrarian claim in first 0.8 seconds triggers curiosity gap; TikTok rewards watch-through on educational reversals.
• Hook: “This 4-minute workout replaced my gym membership.” Payoff: Viewer sees a real-time routine they can do during kids’ screen time. Why it works: Time constraint + relatability; parents share content that validates their schedule.
• Hook: “My kid asked why I was breathing heavy after one flight of stairs.” Payoff: Emotional story tied to a 10-day stair challenge. Why it works: Vulnerability + transformation setup; algorithm favors emotional retention spikes.

Notice the difference. These aren’t topics. They’re angles with built-in stopping power.

The Three Variables You Must Customize

The template works, but only if you dial in these three inputs. We learned this the hard way after our first batch of ideas flopped across 12 test accounts in March 2026.

1. Niche (Be Stupidly Specific)

Don’t say “fitness.” Say “bodyweight workouts for people over 40 who hate gyms.” The AI pulls better metaphors, obstacles, and emotional frames when the niche is narrow. Gary Vee has been preaching “niche down” since 2018, but in 2026 it applies to your prompts too.

2. Audience (Include the Problem, Not Just the Demo)

Bad: “millennials interested in finance.”

Good: “millennials drowning in student debt who don’t trust traditional investing.”

The second version gives the AI an emotional job to do. In our tests, problem-aware prompts generated 2.4x more shareable angles.

3. Platform (Algorithm Behavior Changes Fast)

TikTok rewards retention. Instagram Reels in 2026 prioritizes sends and shares. YouTube Shorts still weighs CTR from the thumbnail. When you specify platform, add what matters there. Example: “why it works on Instagram Reels in 2026 (focus on share triggers and relatability).”

How to Turn AI Ideas Into Actual Content

You’ll get 30 ideas in 60 seconds. Most will be solid. Three will be great. Here’s how we sort them:

Step 1: Flag the ones you’d watch. Gut-check every hook. If you’d scroll past it, your audience will too.

Step 2: Check for proof. Can you film this today? If the idea requires a 14-day transformation or a $4,000 camera, skip it. We prioritize ideas we can execute in under 20 minutes.

Step 3: Batch by format. Group talking-head ideas separately from B-roll montages. Shoot all talking-head videos in one session. Our fastest creators film 12 videos in 90 minutes using this method.

Alex Hormozi’s team does something similar. He’s mentioned batching 20+ YouTube videos in a single day by pre-writing hooks and shooting back-to-back. Same principle applies to short-form, but you can move even faster because there’s no editing complexity.

The One Thing Most Creators Get Wrong After Running the Prompt

They post all 30 ideas in the same voice.

Here’s the catch: AI defaults to a neutral, explanatory tone. If you don’t rewrite the hook in your voice, it sounds like everyone else’s AI-generated content. TikTok users can smell it in 2026. The algorithm doesn’t penalize AI content directly, but it does punish low engagement — and monotone delivery kills engagement.

Take the AI hook. Say it out loud. Rewrite it how you’d text a friend. That’s your real hook.

We saw this across our network in Q1 2026. Accounts that rewrote hooks saw 60% better 3-second retention than those who read the AI output verbatim. It’s a small edit with massive leverage.

Advanced Move: Chain Prompts for Series Ideas

Once you’ve got your 30 ideas, pick the top 5 and run a follow-up prompt:

Take idea #3 and generate a 5-part series. Each video should build on the last, with a cliffhanger or callback. Include the hook for each part.

Series drive follower conversion better than one-offs. TikTok’s 2026 “part 2” feature surfaced this in the UI, and Instagram added series carousels in late 2025. The AI won’t know to do this unless you ask, but when you do, it maps out entire content arcs in seconds.

MrBeast’s team uses series structure for every long-form video. The same psychology works in 15-second chunks. Curiosity gaps between parts keep viewers coming back, and the algorithm rewards repeat viewers with more distribution.

Frequently Asked Questions

Does this AI prompt work with ChatGPT-3.5 or do I need GPT-4?

GPT-3.5 will give you results, but they’re noticeably more generic. We tested both in March 2026, and GPT-4 produced 80% usable ideas versus 40% with 3.5. Claude 3 Opus performs similarly to GPT-4. If you’re serious about content velocity, the paid tier is worth it.

How often should I run this prompt to avoid repetitive ideas?

Once a week is safe. We run it every Monday for our client accounts. If you run it daily, you’ll see overlap by day three. The AI pulls from patterns, so spacing requests gives you fresher angles. You can also rotate the audience or niche variable to keep output diverse.

Can I use these AI-generated ideas without changing them?

Technically yes, but don’t. The hook will sound flat and algorithmic. Rewrite it in your voice before filming. We’ve tested this across 40 accounts — creator-voiced hooks outperform AI-raw hooks by 60% on average in first-hour retention. The idea is the blueprint, not the script.

If brainstorming is your bottleneck, this prompt removes it. You’ll never stare at a blank content calendar again. But speed only matters if you can distribute what you make — and that’s where most creators hit the next wall.

We built x20.online because coming up with ideas is easy now. Getting them in front of 100,000 people organically is the hard part. Our network pushes your content through real accounts so the algorithm sees early engagement and runs with it. Check out our pricing if you’re ready to pair AI ideation with actual distribution that moves the needle.

Want more automation strategies like this? We break down what’s working (and what died) every week on our blog. And if you’re into this stuff, browse our full AI automation tips — we’ve tested everything so you don’t have to.