The 5-Step Guide For Creating Hollywood-Caliber Ads Using AI

Respond “🧠” and we’ll send you a joke clearly created by AI.

PJ Accetturo spent 10 years making commercials the old-school way: 

In May, he said f*** that.

Now, he runs Genre.AI, an AI-native ad agency that’s racked up 300M+ views online (working with brands like Disney, Ramp, and Kalshi). PJ shared his method for producing next-level creative from his office chair.

the problem: Take the 2017 Pepsi x Kendall Jenner ad fiasco. Pepsi paid $2–$5 million to create the commercial, from start to finish. Part of that huge price was Jenner's reported celeb fee ($400K–$1M). Lesson: Even Pepsi is not immune to failed creative. 

PJ’s engagement volume proves that you can create a successful piece of branded video content in under 2 weeks, for less than $1,000. The snowball effect here is that it can test more over time and ultimately win faster than going down the traditional ad route. 

the strategy: Instead of risking everything on a single expensive ad, use this workflow. These are the steps:

scroll-stopping idea: Your only goal is to stop someone from scrolling. The first 0.5 seconds matter just as much—if not more—than the next 30. 

Use ChatGPT to write a hook. Then, use the tool to finalize a script that ties into the product.

storyboard: Paste your approved script into Reve. Ask for a shot list and generate starter frames for each scene.

pictures to videos: Now, animate the frames using Google Flow + Veo 3.1. Ask for different angles, camera movements, and character actions.

voices & sounds: AI loves handing you stock audio if you don’t intervene. Write the dialogue, choose the music, and control the ambient noise—or the model will invent vibes you didn’t ask for.

edit & ship it: Iterate until your video is perfect. Edit it yourself using CapCut or DaVinci Resolve (both free), or hire a freelancer to help. Don’t skimp out on the edit. 

➿ Want more pro tips from PJ? Check out the full playbook here. 

why it matters: It’s going to be one of the largest stock debuts ever + obviously a huge benchmark for AI valuations.

state of play: OpenAI restructured into a public-benefit corporation and reduced Microsoft’s sway—clearing the path to raise capital and use stock for deals.

what remains: The partnership is still strong (Azure remains a major provider; long-term access/rights persist; there is a large multi-year Azure spend).

by the numbers: Private marks hover ~$500B today, while reports peg annualized revenue in the low-double-digit billions and rising.

between the lines: More independence from Microsoft = broader fundraising and M&A flexibility ahead of a 2026–27 window.

Perplexity Patents adds a natural-language layer on top of USPTO filings. You ask normal questions, it reads the legal maze for you, pulls the biggest claims, and shows related inventions with citations you can actually click.

ELI5: You ask, “Did anyone already invent this?” It replies, “Yep. Here’s who tried, what they said they built, and who they borrowed from.”

It lowers the “I need to understand the terrain before I pay $900/hour for a lawyer” barrier.

Also, yes. It looks like 2025, not Windows 98. A massive win for anyone who’d rather not drown in 300-page-long patent PDFs before coffee.

Congratulations GitHub. You now have as many developers as Bangladesh has people.  

One chart shows that the monthly number of distinct contributors to generative AI projects increased from around 90,000 to 200,000 between January and July 2025.

View the full report here.

background: Anthropic introduced MCP (Model Context Protocol) to the world about a year ago. And people are still confused about what it is and how it “will change their lives.”

in plain english: MCP is a handshake between your AI model (like Claude) and the software you want it to work with. It allows your LLM to actually do things, beyond simply fixing the grammar in your email. It's the connection layer that lets AI communicate with external tools and services. 

There are three tiers of AI:

MCP isn’t a new “Tier 4” or some AGI breakthrough. It's the plumbing that lets each of those three tiers plug into real systems and automate work.

apply it: We built an MCP repository so you don’t have to start from scratch. It breaks down real-world use cases and shows how to bridge Claude with the tools you already use to run your business (Vercel, Zapier, Canva, Figma, Stripe, HubSpot, Airtable, and more).