Tubegraph

Using AI to create viral content

7 min read

From: gregisenberg

This article explores how to automate content creation with AI to generate viral videos and build a business around them, drawing insights from PJ, a creator known for his viral AI films on X [00:00:09]. The focus is on practical, step-by-step methods to create AI-generated commercials and other content that resonates with audiences [00:01:33].

The Power of AI in Content Creation

AI tools like Google V3 and large language models (LLMs) such as ChatGPT, Claude, and Grok are making it easier to produce cinematic content without needing high technical skills [00:02:25]. The key is to create entertaining content that people will want to watch and share [00:02:40]. With generative AI, content volume is expected to increase exponentially, making it crucial to understand why people engage with content, not just how to make it [00:04:58].

Content Creation Process: From Idea to Viral Video

The process involves several stages, leveraging various AI tools to streamline production.

1. Ideation and Hook Development

The first step is to have a “kernel of an idea” or a hook that makes the content unique and unexpected [00:05:40].

  • Example: The Satirical Pharmaceutical Ad The viral “Puppine” pharmaceutical commercial started as an experiment with a client request for a standard commercial [00:06:23]. PJ decided to take a “left field” turn halfway through, suggesting a placebo pill that delivers a dog [00:06:54]. This evolved with a friend’s input into a pill that secretes pheromones to attract puppies [00:07:13].

2. Scripting with Large Language Models (LLMs)

LLMs like ChatGPT are invaluable for scripting, especially for satirical content.

  • Initial Prompting: Start with generalized instructions to get a cliché script, then refine it [00:07:37]. For the pharmaceutical ad, the prompt included creating four scripts for a fake medication called “Pup” that solves depression by delivering a puppy, styled as an SNL-esque satirical drug commercial [00:07:53].
  • Iterative Refinement: If you don’t have a strong initial idea, you can ask ChatGPT for twists or satirical ideas, such as five different twists on a pharmaceutical commercial that could be funny and SNL-style [00:09:37].
  • Generating Punchlines: Feed your favorite lines back into the LLM and ask for more “zingers” or one-liners [00:14:18]. Approximately 70% of the lines in PJ’s viral ad were generated by ChatGPT through this iterative process [00:15:21].
  • Outsourcing Style: You can direct the LLM to generate jokes in the style of a specific comedian or show, like Seth Rogen or “Deep Thoughts with Jack Handy” from SNL [00:15:46].
  • Experiment with Different LLMs: Don’t just use ChatGPT; try Claude and Grok (especially Grok’s unhinged mode) for varied outputs, as some might offer more “insane” or less censored content [00:12:21].

3. Generating Video with Google V3

Once the script is refined, generate video clips using V3.

  • Prompting V3: Take each dialogue line from the script and create a prompt for V3 [00:16:34]. You can use a simple prompt structure describing what’s happening and the dialogue [00:17:47].
  • Character Consistency (Limitations & Workarounds): V3 is currently text-to-video, meaning there’s no inherent character consistency across shots [00:20:20]. To achieve similar-looking characters, describe them in detail (e.g., “an old guy with short gray hair,” “a white guy in his 30s with a mustache and a beard and like strong chin with muscular arms”) in every prompt, ideally having ChatGPT generate these details [00:20:50].
  • Cost and Iteration: V3 costs about $2-3 per shot [00:18:31]. Run clips one at a time to get feedback and make tweaks, as you’ll likely generate multiple clips (e.g., a 5:1 ratio of clips to good ones) [00:23:41].
  • Audio and Subtitles: V3 generates audio for the video, but it can sound “thin” and lack voice consistency [00:24:40]. To avoid weird gibberish subtitles, do not include quotes in your prompts [00:19:06].
  • Camera Movements and Detail: You can prompt for specific camera movements (e.g., dolly in) and granular details (e.g., “she glances up and there’s a pea stain on her kitchen ceiling”) to enhance the video [00:22:23].

4. Editing

After generating clips, stitch them together in an editing software.

  • Software Choices: Final Cut Pro is used by PJ, but CapCut is a free and equally capable option for this type of content [00:23:19]. DaVinci Resolve is another free, top-tier option for more advanced editing [00:23:30].
  • Example Cost: The “Puppine” commercial, despite looking high-budget, cost about $500 in V3 credits and took 6-12 hours to produce [00:21:22]. This contrasts sharply with traditional pharmaceutical commercials that can cost hundreds of thousands to millions of dollars [00:21:31].

5. Crafting a Viral Hook for Distribution

The marketing aspect is as important as the content itself.

  • Platform-Specific Hooks: On platforms like X, the headline is crucial; it has 0.2 seconds to hook viewers. The content itself then has 2-3 seconds to make them stay [00:25:50].
  • Mr. Beast’s Funnel: Apply the Mr. Beast philosophy: title > thumbnail > content [00:26:20].
  • Example: “Puppine” Ad Hook: “I used to shoot 500 in V3 credits in less than a day. What’s the argument for spending 500K now? Steal my prompt below.” [00:26:47]
    • This hook works because it combines a “low-key flex” (legitimacy) with a surprising cost comparison (500) [00:27:13]. It uses a question to generate replies, feeds the algorithm, and offers value (“steal my prompt”) to encourage clicks [00:27:40].
  • Provocative Statements: Viral posts often start with a provocative statement (e.g., “AI video quality just 10xed overnight. I’m speechless.“) [00:29:05]. Avoid overused phrases like “Hollywood is cooked” and instead find your own unique wording [00:29:10].

Building a Business Around AI-Generated Content

Here are several content strategies and using AI for content creation for monetization:

  1. Build a Personal Brand: Use V3 to create content that you’re proud of, attracting an audience and leading to downstream work [00:31:36]. A large audience is a valuable asset [00:44:55].
  2. Attract Direct Clients for Commercials: Create AI-generated commercials for specific industries (e.g., food products, beer companies, CPG brands, underwear brands) [00:32:00]. These brands often want “unhinged special effects” that agencies can’t provide within budget [00:33:19]. Charge at least $5,000, valuing your skill as a filmmaker [00:33:30].
  3. Build a Portfolio for Agencies: Create a specialized portfolio, focusing on a niche that interests you (e.g., knife commercials). You can even “steal” (copy shot-for-shot) award-winning commercials to learn pacing and flow, then adapt them to new products [00:34:41].
  4. Create Narrative Series with Brand Integrations: Build a unique universe or series, like “Neural Viz’s” “moniverse,” and integrate products into the storyline [00:37:04]. This involves taking popular social formats (e.g., “Man on the street” videos) and adding a twist [00:38:57].
  5. Sell Courses/Digital Products: Package your knowledge and skills into courses, PDFs, or other digital products [00:39:50].
  6. Consult with Larger Companies: Offer consulting services to Fortune 500 companies on how to integrate generative AI while adhering to legal and compliance constraints [00:40:08]. Charge more based on the size of the company [00:41:01].
  7. Build a Sketch Comedy Channel: Create a Key and Peele-style sketch comedy show [00:41:14]. While harder to monetize directly, large audiences can attract product integrations (e.g., Bang energy drinks on TikTok) [00:41:31].
  8. Do Narrative Films (with caution): While long-form narrative content is possible (e.g., a 25-minute TV show), the AI space iterates too quickly, leading to inconsistent quality within a single long film [00:41:49]. Prioritize episodic weekly content, which aligns with modern consumption habits and is easier to produce without large budgets [00:42:17].
  9. Pitch Brands on Narrative Content: Offer to create episodic narrative series (e.g., 10 two-minute episodes) for brands, where their products are integrated as relics or artifacts [00:43:02].
  10. Target Small Sub-niches: Find niches with low content supply but high demand. Examples include faith-based content (like stories from the Bible, which has 2 billion daily active users) [00:43:30], or “Romantasy” (romantic fantasy) BookTok, where authors would pay for book promos or scenes from their books [00:45:40]. This allows for creating high-impact niche content using AI.

The overall advice is to do what fascinates you, copy successful content frameworks, apply them to a unique niche, and follow the technical instructions for utilizing AI tools for efficient content strategy [00:44:19]. This approach enables a wide range of business ideas using AI-generated content.

Graph View