The B2B Podcast Index
Index
All categories
MarketingSalesSaaSFinanceHROpsLeadershipCustomer SuccessAI & DataProductStartups & FoundersRevOpsEngineering & DevTools
MethodologySubmit
Best of:MarketingSalesSaaSFinanceHROpsLeadershipCustomer SuccessAI & DataProductStartups & FoundersRevOpsEngineering & DevTools
An independent project byFame
SearchBest episodesGuestsInsightsMethodologySubmit a podcast
Index/SaaS/The Scale Up Show
The Scale Up Show artwork

100k AI Ads Using Google VEO 3 JSON Prompting - Complete Tutorial

The Scale Up Show · 2025-08-04 · 7 min

0:00--:--

Key moments - from our scoring

Substance score

17 / 100

Five dimensions, 20 points each

Insight Density4 / 20
Originality4 / 20
Guest Caliber3 / 20
Specificity & Evidence4 / 20
Conversational Craft2 / 20

Ryan Staley demonstrates a three-step workflow for generating high-production-value videos using Google Veo 3 and JSON prompting, eliminating the need for traditional crews. The process begins with Grok 4 to identify trending JSON video prompts by scraping social platforms, capturing visual links, engagement metrics, and author information. Step two involves customizing these structured prompts - Staley shows how copying a JSON prompt into Gemini 2.5 Pro and modifying key variables (like swapping "New York City" for "Chicago") instantly generates localized, high-quality outputs. Step three deploys Veo 3 to render the final video. Staley demonstrates real examples: a selfie video from the moon, a spinning luxury suitcase ad, and a Bigfoot exploration - all created without editing in minutes. The workflow suits marketing teams, ad agencies, and product-focused companies looking to multiply creative capacity by 20-40% monthly. Key limitation: Veo 3 has daily rate limits (around five videos per day on paid accounts), requiring multiple accounts for volume production.

Key takeaways

  • →Use Grok 4 to scrape social media and identify the top 20 most visually stimulating JSON video prompts with engagement metrics and author details.
  • →JSON prompting provides superior specificity and articulation compared to natural language, enabling consistent high-quality outputs when properly structured.
  • →Customize successful JSON prompt templates by changing specific variables (location, character, theme) rather than rewriting from scratch, dramatically reducing iteration time.
  • →Google Veo 3 has rate limitations - approximately 5-6 videos per day on paid accounts - requiring multiple accounts for higher volume production.
  • →JSON prompt architecture can be repurposed across multiple use cases beyond viral videos, including ads, product demos, and various commercial applications.

In this episode

  1. 1Introduction to JSON Prompting and Veo 3 for Video Generation
  2. 2Using Grok 4 to Identify Top Trending Videos and JSON Prompts
  3. 3Customizing JSON Prompts in Gemini for Specific Use Cases
  4. 4Creating High-Quality Ads and Videos with Veo 3 in Minutes
  5. 5Limitations and Best Practices for Daily Video Generation
  6. 6Recap and Three-Step Framework for Viral Video Creation

Mentioned

Google Veo 3Grok 4Google GeminiGemini 2.5 ProOpenAIGPT-5Ryan Staley

Topics in this episode

Prompt engineeringGemini 2.5 ProGoogle Veo 3JSON promptingGrok 4Social media scrapingVideo generationUnboxing videosChicago cityscapeStormtrooper content

Questions this episode answers

What is JSON prompting and why is it better for video generation than regular prompts?

JSON prompting uses structured JavaScript-like syntax with specific curly brackets and variables to articulate exactly what you want, making it more precise and replicable than natural language prompts. The structured format works particularly well as template architectures that can be easily customized and adapted for different outputs.

How do you find the best JSON video prompts to use as templates?

Use Grok 4 to scrape social media and identify the top 20 most visually stimulating JSON prompts, requesting outputs that include visual links, the complete prompt text, author information, and engagement metrics. This surfaces proven high-performing templates you can then customize.

Can you create a customized video using Veo 3 without being a prompt engineer?

Yes - copy a high-performing JSON prompt template into Gemini 2.5 Pro and ask it to customize specific variables (like location, character, or theme). Paste the updated prompt directly into Veo 3 to generate the final video, requiring no technical expertise.

What are the rate limits for creating videos with Google Veo 3?

On a paid $20/month account, Veo 3 allows approximately five to six videos per day before hitting a ceiling on video generation capacity, though multiple accounts can work around this limitation.

What business outcomes are companies seeing with this AI video workflow?

Companies using this approach are multiplying their creative capacity by 20-40% in weeks to months, enabling faster ad production, product demos, and viral content creation without expanding their teams.

What our scoring noted

Our reviewer’s read on each dimension, with quotes from the episode.

Insight Density

4 / 20

The episode is a shallow screencast walkthrough that recycles three vague steps with almost no novel, actionable insight. The only concrete operational detail is the VEO 3 rate limit; everything else is hype and filler describing what is visible on screen.

And I'm going to show you one last one that I thought was pretty cool. And this was with Bigfoot, right? So as you can see, this is Bigfoot.
it did this. Look at this, it did another one. Uh, this one's a little bit different.

Originality

4 / 20

The core idea - use one AI tool to find popular prompts, paste into another to adapt, run in a third - is a mild process shortcut but requires no first-principles reasoning and produces no contrarian or counterintuitive argument. The framing is pure hype rather than fresh thinking.

So once I started seeing a consistent pattern of these really good videos and slight references to JSON prompting, I dug deeper and that's how I got into this.
So it's almost like a computer language that is often, it's. It's JavaScript in nature.

Guest Caliber

3 / 20

This is a solo influencer-style tutorial with no interview guest. The host's credentials are stated in the vaguest possible terms and nothing in the episode demonstrates deep practitioner expertise at scale.

I've helped thousands of executives transform their work, what's possible, and who they are with AI.
I work with teams and companies. And the amazing thing about it is I see so many different use cases on average, 50.

Specificity & Evidence

4 / 20

The headline '$100,000 videos' claim is completely unsubstantiated, and the '20-30-40% capacity multiplication' figures are asserted without a single named company or data point. The only real specifics are the $20/month account tier and an approximate 5-6 video daily cap on VEO 3.

companies are multiplying their capacity by 20, 30, 40% in months
I think I was running into issues, um, for a paid $20 a month account when I created like five videos, maybe six videos

Conversational Craft

2 / 20

There is no interview dynamic whatsoever - this is a solo screencast with no follow-up questions, no pushback, and no interlocutor. The delivery is disorganised and heavily padded with filler phrases, offering nothing in terms of conversational craft.

Okay, how amazing was that?
Right. Without any editing first. How wild is that?

Conversation analysis

Computed from the transcript - who did the talking, and the words that came up most.

Most-used words

prompt14different10json9videos8show7video7step6prompting5prompts5areas5created5create4reason4grok4link4three3

Episode notes

In this episode, I share how I'm using JSON prompting with Veo3 to create high-quality videos quickly and efficiently. I walk through my three-step process: starting with content curation using Grok 4, then refining prompts to fit my voice and goals, and finally generating the video content itself. I highlight how powerful JSON prompting can be for dialing in both specificity and engagement. I also share some sample outputs and encourage you to explore these tools if you're looking to level up your content creation workflow. Chapters 00:00 Introduction to JSON Prompting with Veo3 02:45 Step 1: Curation with Grok 4 04:49 Step 2: Customizing JSON Prompts 06:13 Step 3: Creating Videos with Veo3 Your competitors are already using AI. Don't get left behind. Weekly AI strategies used by PE Backed and Publicly Traded Companies→

Full transcript

7 min

Transcribed and scored by The B2B Podcast Index.

Speaker A: Today, I am going to walk you through how to use JSON prompting, specifically with Vayo3 to create $100,000 videos that would have previously taken a whole crew and team to generate. You're going to be able to do it only in three steps. I'm going to walk you through with exact steps in terms of curation all the way to execution with this. For those who don't know me, my name is Ryan Staley. I've helped thousands of executives transform their work, what's possible, and who they are with AI. And I'm going to show you a glimpse of kind of how I help people. What I'm going to do is start sharing my screen and we're going to rip right into this. And reason being is I was originally going to do a GPT5 video today, but for some reason OpenAI decided, hey, we're going to kind of pretend like we're going to release it in the end of July and then we're going to change our mind. Right? Maybe they did. Maybe that was just the hype machine getting after it. So let me share my screen and what you're going to see on this is. It's pretty eye opening. Okay, so step number one, I went over to Grok 4 and the reason why I went to Grok 4 is because it's really good at curation of social scraping. Okay, so I said identify the top 20 most visually stimulating, useful JSON video. Three prompts for Google Gemini. Please include a link to the visuals, the entire prompt, author engagement of wherever it is located and why it made your top 20 list. Okay, so some critical things within that prompt is I asked for the visual, I asked for the entire prompt, the author and the engagement. And then why. Okay, so what you're going to see here is like, I was blown away with like, how good Grok spit this out, right? And so what you're going to see here is that it went through. It's got a link to the visual, it's got, uh, basically a prompt with character named Torben, but as you can see at the end of it, it's got how many, how much engagement, the author, and actually the link. This had close to 200,000 views and it went through. And this was what was amazing about it. It did this. Look at this, it did another one. Uh, this one's a little bit different. I'm going to show you examples what some of these look like. But by the way, it goes all the way down and it did all 20 of them. And I Didn't have to keep asking for it. What you're seeing here with these little curly brackets and these specifics, this is what JSON prompting is. So it's almost like a computer language that is often, it's. It's JavaScript in nature. But the reason why it's good at prompting, because it's really good for specifics and articulating, uh, exactly what you want. Okay. So once I started seeing a consistent pattern of these really good videos and slight references to JSON prompting, I dug deeper and that's how I got into this. Okay, so now I'm going to show you some outputs on what this looks like. But as you can see, if we go all the way to the bottom, I mean, it, it included everything from like, um, basically, you know, Indian areas where there was like, basically different, um, parts of nature. In India, this one was more stained glass. Right. And so there's a lot of different areas where we went into it. So I'm going to show you what this looks like. All right, step number two. So we're going to go over this. How do you customize the top prompts in JSON for exactly what you need? And you don't even need to be a prompt engineer. This is so simple. So there was one with an unboxing video where basically New York City came out of it. And I'm like, hey, I live in Chicago. I want to create one for Chicago. The only thing I did is I hopped over to Gemini 2.5 Pro and I said, can you just update this prompt to be for sure? So all I did, and then I copied and pasted the entire JSON prompt. All right, because this has, I mean, specific timeline sequences, all these different areas, right? So watch this. This is what's wild. Uh, so once I did that, this is what it created. I just pasted the prompt and I started a new, new chat. And this was the prompt, right? So customize everything in Chicago city and with the whole assembly. And this is what the output looked like. Right. Without any editing first. How wild is that? I basically created or, um, transformed what was a top 1% prompt in JSON with VO3 in seconds with that. Now there's other areas that it created as well. I want to show you some different examples. This one was really popular as well. This was, um, basically a stormtrooper creating a selfie video on the outpost. I'm going to play this for you. Look at that. A selfie on the moon on top of it too. Another one. This was a hundred thousand Dollar ad that would have been created for a suitcase, as you can see, spinning around. Transform. Freaking amazing. And I'm going to show you one last one that I thought was pretty cool. And this was with Bigfoot, right? So as you can see, this is Bigfoot. Okay, how amazing was that? So what you can see is like literally this is five minutes, I think the video we're at so far. So I'm going to wrap this up and just kind of recap this and if you want I can include those 20 prompts, uh, basically at the end, drop some, some love in the comments section, right? And spread this and share this. And if we get some traction, I'll put it together, I'll put it in sweet format and I'll give you those 20 prompts of the top performing videos on the Internet right now with fail so you can replicate that JSON prompting. So once again to recap it, use Grok4 basically to scrape social and identify what are the top trending videos. That'll give you insights. And the other thing that I didn't hit is the architecture of those JSON prompts is why they're so good. So what that'll do is you could basically use those as different prompt templates, um, and use it for anything that you want. It doesn't need to be for just a viral video. It could be for ads, product demos, it could be for a lot of different areas, right? So that's step one. Step two. Then you customize your prompt, okay? So you take the prompt template, you say okay, use this style and structure, but make it about this instead. So like for example, instead of Bigfoot, I don't know, maybe we do something with like a yeti instead, which I kind of want to do that now. I'm m going to test it out, right? So you could change it up and say, okay, use this prompt and make it about a yeti. Make it about a different area. That's step two. And then step three is just popping in a VO3. Now the only limitation with this that I will give you or tell you about is effectively with Vail 3, if you put this in there, um, once you start creating videos, it has a lid in terms of, or I should say, um, ceiling in terms of how many videos you can create in a day. So I think I was running into issues, um, for a paid $20 a month account when I created like five videos, maybe six videos. So just food for thought. That's something that you got to take care of. I have multiple accounts. I was able to go back and forth, but it's amazing to use. And one of the things I had to share with you because the outputs were so simple. And we're doing all this in less than seven minutes. Okay. So if you want to hear more things like this, feel free to click on the link and actually join my newsletter. I go through different walkthroughs like this on YouTube. I have tons of content I create on LinkedIn every single day. And I do this for you because I work with teams and companies. And the amazing thing about it is I see so many different use cases on average, 50. But what I'm seeing is companies are multiplying their capacity by 20, 30, 40% in months. Okay. Sometimes even. Even weeks. And so I'm going to start sharing this with you more and more. This is off the cuff. I wanted to do it and hopefully we'll be back. We'll show some stuff with, uh, GPT5 or something else. Cutting edge. But thanks for joining me and we will see you all on the next video.

Related episodes across the Index

Other episodes covering the same guests and topics, from across The B2B Podcast Index.

  • Why B2B Brands Are Using AI to Write Sales ProposalsThe Growth Operator with Fexingo · on Prompt engineering85 / 100
  • Episode 018: Season 2, the $75 Consult and the Frankenstein StackAI Tools for Practicing Lawyers · on Prompt engineering84 / 100
  • Episode 111: Building Your Defences Against AI MisinformationValue Driven Data Science · on Prompt engineering83 / 100
  • How B2B Marketers Use AI to Personalize at Scale for EnterpriseB2B Marketing with Fexingo · on Prompt engineering82 / 100
  • 477. The Nitty Gritty of AI From an Attorney and AI Expert with Mike BrownThe Game Changing Attorney Podcast with Michael Mogill · on Prompt engineering81 / 100
  • Greg Joseph: AI’s Future in Personalizing Streaming Ads10 Minute Martech · on Prompt engineering75 / 100

More from The Scale Up Show

All episodes →
  • Beyond the Demo: Why Most AI Implementations Fail and How to Build Ones That Actually Work69 / 100
  • Stop Using GPT-5 Like GPT-4 - Do This Instead41 / 100
  • How a $100M+ Company Deploys AI Agents for Real Business Impact48 / 100
  • The Four-Stage AI Implementation Framework: Taskmaster to Orchestration for Revenue Teams35 / 100
  • The Super Agent: One AI Tool to Rule Them All?39 / 100
Explore the best B2B SaaS podcasts →
All The Scale Up Show episodes →