@angus.sewell
Instagram · Greenscreen Talking Head
@angus.sewell Time to start hoarding your data ☮️✌️ #ai #startup #saas
@angus.sewell
Instagram · Greenscreen Talking Head
@angus.sewell Time to start hoarding your data ☮️✌️ #ai #startup #saas
Performance
Video #28539 · Greenscreen Talking Head
[01]
The creator makes the argument that many multi-billion dollar AI SaaS companies are just simple wrappers on top of existing AI models. He breaks down the business model into three layers (data, AI orchestration, output) and then systematically analyzes several unicorn companies (Glean, Harvey, Clay, etc.), showing how their core functionality can be replicated for a low monthly cost using basic tools and system prompts.
[02]
[03]
[04]
[05]
What stops the scroll in the first 3 seconds
Title hook · 55%
Eight unicorns, nobody inside
Verbal hook · 40%
$27,000,000,000 of AI companies are all the same chatbot.
Visual hook · 5%
A direct, close-up shot of the creator's face as he speaks expressively to the camera, creating an immediate personal connection.
[06]
Steal this for your own video
worth of companies are all doing the same
[07]
The promise the opener makes, and how the video pays it back
Hook
A bold, numerically-backed proclamation that challenges the perceived value and complexity of an entire industry or category, immediately establishing a controversial premise.
Payoff
The viewer is left feeling intellectually superior, armed with 'insider knowledge,' and questioning the established value of the critiqued category. The final beat builds rapport by admitting uncertainty about the future, inviting further engagement.
[08]
The structural pattern this video runs on
Framework
The Deconstruction Gauntlet
To expose a widely accepted, high-value category (e.g., an industry, a trend, a product type) as being based on a simple, repeatable, and easily replicable formula, thereby devaluing the established players and empowering the viewer with 'insider' knowledge.
The Universal Blueprint
Presents a simplified, abstract model that allegedly underpins the entire category being critiqued. This primes the audience by framing the complex subject as a simple, repeatable trick.
The Gauntlet of Examples
Systematically presents multiple, high-profile examples from the category. Each one is rapidly processed through a repeating micro-pattern to prove the initial thesis over and over, building overwhelming evidence.
The New Reality
Delivers a final judgment based on the evidence presented, declaring the old model obsolete and positioning the creator as a fellow explorer in a new, emerging landscape, fostering a sense of community with the viewer.
Time to start hoarding your data ☮️✌️ #ai #startup #saas
title · beginning
Eight unicorns, nobody inside
title · beginning
ANATOMY OF A $38B INDUSTRY
informational · beginning
LAYER 01 - DATA Existing data nobody wants to rebuild
informational · beginning
LAYER 02 - AI ORCHESTRATION Model reads, indexes, and reasons
informational · beginning
LAYER 03 - OUTPUT Vertical-specific structured result
informational · beginning
$500-$2,000/mo per seat. That's the product.
informational · beginning
GLEAN $7,200,000,000 Enterprise AI Search
informational · middle
GLEAN Build It Yourself $20/mo
informational · middle
HARVEY $11,000,000,000 AI for Law Firms
informational · middle
HARVEY Build It Yourself $20/mo + API
informational · middle
CLAY $3,100,000,000 AI Sales Platform
informational · middle
CLAY Build It Yourself from $49/mo
informational · middle
JUICEBOX $850,000,000 AI Recruiting Platform
informational · middle
JUICEBOX Build It Yourself from $49/mo
informational · end
SPELLBOOK $350,000,000 AI Contract Review
informational · end
SPELLBOOK Build It Yourself $20/mo
informational · end
TOME $1,300,000,000 AI Sales Intelligence
informational · end
TOME Build It Yourself $20/mo
$27,000,000,000 of AI companies are all the same chatbot. Data layer. Some AI bullshit. Vertical output. This is $27,000,000,000. Take a glean for example. Just a measly seven bill valuation. Their AI search probably can't be replaced by, say, Claude. It totally rank, guys. Like, you already have this for $20 a month. No what about something vertical specific, you say? They have all this niche training data. Right? I straight up think this would be, like, just as good. You don't need all that training data to just retrieve documents. What about Clay? Clay's pretty cool. Right? They're totally not a wrapper for a bunch of other data sources. Clay might actually still be valuable. You have to set up all the data sources yourself. Oh, boy. We already know where this one's going. Does JuiceBox actually have unique data that can't just be easily found on a pod? I don't know. I've only gotten a demo of the platform, but it didn't seem like. This one's the easiest yes. Like, it's editing documents, Scott. How can you argue this won't be replaced by Claude? I guess they have their own, you know, models or whatever, but you don't need, like, a super verticalized model to do this. I don't even know what to say anymore. Here you go. Yeah. There's a lot of shit to build out here, but I don't think Sass is the move anymore. Bro, I am trying to figure it out too. Trust.