How I'd use Sora 2 + Claude to hit 1M+ views on AI videos

Key Concepts

• Sora 2: An AI video generation tool.
• Perplexity: An AI research agent used for deep research and content insights.
• Claude: An AI assistant (specifically Claude 3 mentioned) used for generating and evaluating video concepts and optimizing Sora 2 prompts.
• Viral Video Concepts: Short-form video ideas designed to achieve high engagement and widespread sharing.
• Scroll-Stopping Concepts: Video ideas with strong hooks to immediately capture audience attention.
• Distribution as a Moat: The idea that owning audience attention (distribution) is a significant competitive advantage for startups and businesses.
• Hook Strength: The effectiveness of the first few seconds of a video in grabbing attention.
• Pattern Interrupts: Techniques used in video to break viewer expectations and maintain engagement.
• Algorithm Fit: How well a video concept aligns with platform algorithms (e.g., TikTok, Reels) to maximize reach, completion rates, and shares.
• MRR (Monthly Recurring Revenue): A key metric for SaaS businesses.
• SaaS (Software as a Service): A software licensing and delivery model.
• Micro-SaaS: Niche SaaS products with smaller target markets.
• Deep Research: A feature in AI tools like Perplexity to conduct thorough information gathering.
• Prompt Engineering: The art and science of crafting effective inputs for AI models to achieve desired outputs.

---

Mastering Sora 2 for Viral Content: An Unfair Advantage

This episode details a comprehensive workflow for generating viral video concepts using Sora 2, Perplexity, and Claude. The core argument is that mastering AI video generation now provides an "unfair advantage," akin to early adopters of YouTube (2006-2012), Instagram (2012-2019), or Twitter (2019-2023) who built massive media empires. By creating viral content with Sora 2, individuals can capture significant audience attention, which can then be leveraged to launch successful startup ideas, consumer mobile apps, SaaS products, or micro-SaaS. The speaker emphasizes that "distribution is the new moat," making the ability to generate engaging content crucial. The goal is to develop a "muscle" for creating videos with a higher likelihood of success, rather than guaranteeing virality for every single piece of content.

---

The Three-Tool Workflow: Research, Concept Generation, and Optimization

The workflow involves a sequential use of Perplexity, Claude, and Sora 2, with continuous iteration for refinement.

1. Perplexity as a Research Agent

The first step is to use Perplexity (or similar AI tools like ChatGPT Deep Research or Grok Deep Research) as a research agent to understand the target audience and content landscape.

• Prompt Structure: The speaker provides a specific prompt template:

  "I run a startup. In my case, this is what it does. [Describe your startup, e.g., 'Gives validated startup ideas, trends, etc. It's called ideabrowser.com.'] I need to create short form video that will go viral. To do that, I need scroll stopping concepts for my audience of [describe audience, e.g., 'entrepreneurs building mostly software. The ones who long for the idea of owning a SaaS that spits out cash.']. What kind of short form will get this engagement for this audience, show examples and get frameworks?"

• Audience Motivation: Crucially, the prompt highlights the audience's main motivation (e.g., for Idea Browser, it's entrepreneurs wanting to build software/SaaS but needing ideas or guidance).
• Deep Research Feature: The speaker recommends using Perplexity's "Deep Research" button for more thorough insights.
• Output Examples: Perplexity's output includes:
  • Core Psychology: High arousal emotions, validation seeking, FOMO (Fear Of Missing Out), pattern recognition.
  • Scroll-Stopping Video Frameworks:
    • Contradiction Hook: Leverages unresolved tension (e.g., "I'm broke, but I'll show you my 50K MRR SaaS," "Zero coding skills, but built 100K SaaS in six months").
    • Micro Tutorials: Quick, actionable guides (e.g., "Set up a SaaS analytics in 60 seconds").
    • Real Numbers Reveals: Showing actual financial data (e.g., Stripe MRR growth charts).

2. Claude as a Concept Generator and Evaluator

The research output from Perplexity is then fed into Claude (referred to as "Claude" for consistency, though the speaker notes the French pronunciation "Claude").

• Step 1: Ingest Perplexity Research: Copy and paste the entire deep research output from Perplexity into Claude. Claude can handle long inputs.
• Step 2: Generate 10 Video Concepts: Prompt Claude to act as a "skilled social media content strategist and short-form video expert" and brainstorm 10 creative short-form video concepts based on the ingested research, including goals, requirements, and psychological triggers.
  • Example Concepts Generated:
    • "Refresh Stripe 47 times today. Got this." (Screen recording of Stripe dashboard refreshing, cutting to tactics, tools, and final MRR number).
    • "My ADK SaaS mistake in 23 seconds."
    • "POV, you just got your first dollar on Stripe."
    • "Cloned $100 million SaaS in four days using these three tools, it's legal."
    • "Your SaaS onboarding has a 14-second problem."
• Step 3: Evaluate Concepts: This is a critical step often missed. Prompt Claude to evaluate each of the 10 concepts against "key engagement criteria," rating them 1-10 and providing brief comments:
  • Hook Strength: How attention-grabbing are the first three seconds?
  • Pattern Interrupts and Pacing: Does it break expectations and maintain engagement?
  • Emotional Curiosity Trigger: Does it invoke an emotional response?
  • Algorithm Fit: How likely is it to lead to high completion rates or shares (i.e., trend-friendly for TikTok and Reels)?
• Step 4: Get Top Recommendations: Ask Claude for the top three overall concepts based on its evaluation. Claude provides average scores (e.g., 9.75) and detailed explanations for why a concept wins (e.g., "Perfect storm of controversy, social proof and trending format. Leverages an existing popular content. Remix strategy. Drives massive comment debate. Algorithm gold."). The speaker advises reading these rationales and adjusting concepts if they feel "cringe" or uncomfortable, while potentially retaining the format.
  • Example Top Concept: "Everyone says build an MVP. I did the opposite and hit 25k MRR."
• Step 5: Generate Sora 2 Optimized Prompts: Finally, prompt Claude to create Sora 2 optimized prompts for the top recommended concepts. This involves instructing Claude to research Sora 2's capabilities and optimize the prompts accordingly.
  • Specific Instructions for Sora 2 Optimization:
    • Duration: Specify 15 seconds (Sora currently maxes out at 15s on web, 10s on mobile, but this will increase).
    • Spoken Dialogue: Use quotes for exact lines to be spoken, as Sora 2 can generate synchronized speech. Add a "spoken dialogue" label for clarity.
    • Visual Elements: Describe key visual elements (e.g., "person at a desk," "timer").
    • On-Screen Text: Mention "on text explicitly for the ending" as Sora can render some text, though quality may vary.

3. Sora 2 for Content Production

With the optimized prompts from Claude, Sora 2 is used to generate the video content.

• Platform Usage: Sora 2 can be used on the web (often requiring a pro membership), which is generally easier for copying and pasting long prompts.
• Duration Adjustment: A "hack" mentioned is the ability to sometimes change the video duration from 10 to 15 seconds directly on the web interface, though this might be unavailable due to traffic.
• Example Ad Creation: The speaker demonstrates an ad created using a Sora 2 prompt. The ad shows Notion, then text like "I clone this. Do you want to clone this?" followed by tools and a call to action to visit a link in bio.
• Critique and Effectiveness: While the generated video might not be "perfect" (e.g., "bubble IO" text, sketchy writing), it effectively generates curiosity, which can increase click-through rates and potentially lower cost per click on advertising platforms. The speaker notes that such a concept and production would typically cost hundreds or thousands of dollars if done manually.

---

Iteration and Continuous Improvement

The workflow is not a one-off process but an iterative loop. Users are encouraged to continuously go back and forth between Perplexity, Claude, and Sora 2. For instance, if a Sora 2 video is good but needs a different visual style (e.g., more cinematic), Claude can be prompted for advice on how to adjust the Sora 2 prompt. This continuous iteration and posting are key to refining the content creation process and increasing the likelihood of success.

---

Conclusion: The Power of Hollywood in Your Hands

The speaker concludes by emphasizing the immense opportunity available to those who master Sora 2 and prompt engineering. The ability to "write incredible Sora prompts" grants users "the power of Hollywood in their hands," a highly valuable skill in the current media landscape. The episode encourages immediate action and creative exploration with these tools.