MDExam.com
The $30m dollar per year GLP-1 telehealth brand running hundreds of ads.
01
AI x Human Creativity
We scale your d2c health adswith AI
“I'm the founder of AIVideo.com, and last year we spent $800k running AI-generated ads for our own business. Along the way, we pioneered a new approach combining state of the art AI with human creativity. Now we run an agency doing the same exact thing, specifically for d2c health brands.
Justin Harvey/Founder, AIVideo.com
Our clients
Who we're
scaling right now
Juvenon.com
The 26 year old supplement brand breaking new ground.
SleepKarma.com
A better mouth tape made by experienced ecom operators.
HappySoul.com
The phytonutrient gummy with a great founder story.
02
Best of the best
Our Team
We combine AI agents and human creativity in a continuous feedback loop to produce high-performing content at a massive scale.
03
How we do it
In 3 parts
01Build the
Build the
foundation
Our team of world-class human creators lock in your style and strategy.
02Explore
Explore
messaging
Rapidly generate batches of static ads to test angles and personas.
03Add production
Add production
value and scale
Double down on winners with longer, higher-fidelity super ads.
04
Step One
Build the foundation.
Our brand designers define the visual language of your brand. This helps lock in the aesthetic for automation later on.
05
Step Two
Explore messaging.
Use our proprietary pipeline to rapidly generate large batches of static ads to test angles, personas, and messaging.
06
Step Three
Add production value and scale.
When the data shows us a winning ad, we double down - iteratively adding more information, length, and human touch until we have a “super ad”. Then we restart the process.
01
Higher production value
02
Longer-form storytelling
03
Aggressive budget scaling
07
Industry Best
BTW, we go viral
08
Pricing
What you get
D2C Health Content Engine
Your AI-powered ad factory - first content in 7 days
starts at
$8,500/mo
- Daily ad creative production - video, static, and overlay variations on autopilot
- Performance-driven optimization - creative guided by real Meta Ads data
- Continuous super-ad production for winning angles
- First content delivered within 7 days
- Full onboarding and brand training within 30 days
Also Included
- Complete audit of your ad strategy to identify the highest-leverage opportunities
- Full access to the AIVideo.com product suite
- Hundreds of AI models and tools at your disposal
09
The proof
3 tests
The system we now use for d2c health brands came from a year of running our own AI ad experiments at real scale.
I spent a year running 3 AI ad experiments that got my business 5 billion views and over 20,000 paying customers.
Experiment #1
Can people tell if they're looking at AI?
We'd been working on an affiliate account with a simple goal: make content so realistic it was impossible to tell the difference.
The result was billions of views a month and a comment section full of people indicating that no, they could not tell the difference.
Then I went to the streets of Hollywood, the Entertainment Capital of the World, to give people an explicit test with a cash reward. And still they couldn't tell.
Conclusion: No. If it's done right, people can't tell.
Next I wanted to produce an entire set of high converting ads - multiple shots, voiceover, scripting, and sound effects. Over the last year I ended up spending nearly $800 thousand dollars on these AI generated Meta ads.
They ended up being the main growth driver for our business. I'm going to tell you all about that, but first you need the backstory. Which brings us to experiment #2.
Experiment #2
Can AI write all of our code?
This may seem like a weird question for a page about making AI ads, but I promise it will make sense in a second.
We launched AIVideo.com a couple of years ago as a self-serve creative tool and scaled it to nearly $4M/year. That product has been our laboratory for testing AI automations.
The platform has hundreds of features and nearly a million lines of code. We wanted to know if an AI agent could write all of it.
Well, we've gotten to the point where AI agents are writing essentially all of the code on the platform. Here's my Claude dashboard last month - we spent 17 billion tokens.
Conclusion: Yes, an AI agent can write all of our code.
And that made us wonder:
Could an AI be as good at making health ads as it is at coding?
Experiment #3
Can AI mass produce high converting ads?
At first, the answer was actually no.
Here were the biggest problems:
The smartest AI agents do not have visual tools built in. They can only output text.
There is no good interface for making edits or tweaks - you are stuck with the first draft.
There is no cloud storage, so your hard drive fills up immediately.
The AI has no taste. The rhythm, the feeling, the brand - it is helpless.
Sure, you can play the volume game and just post everything. But then your customers see a ton of bad videos with your name on it. And you waste a ton of time and money testing subpar creatives.
Why AI is great at code but bad at ads
My theory of how to get AI agents to make great video ads actually comes from evolutionary biology.
The Prefrontal Cortex
~500,000 years old
Planning, logic, math, science - the part that makes us human. It's basically just a prototype. Compared to the rest of the brain, it's inefficient and unoptimized.
AI has solved it.
Coding is solved. Math is solved. Science is almost solved.
Visual Perception & Rhythm
~1,000,000,000 years old
Visual perception, rhythm, and empathizing with your audience. Evolution has had 2,000x more time to optimize these systems.
AI still struggles here.
Taste, feeling, brand - these are ancient and deeply optimized.
The solution: human taste in the loop
The strategy we've pioneered over the last 6 months is to devise ways to keep a little bit of human taste in the loop, giving feedback.
5 ways to infuse human taste into your AI content:
1
Connect your system with analytics. Feed real retention and engagement data back to the AI so your actual customers guide improvement.
2
Invest human time defining brand style, researching reference videos, and nailing the messaging before the AI touches anything.
3
Have a talented human creator make the first example to teach the AI what the tone and flow should feel like.
4
Set up a light review process to curate the best results. At scale, $10 of human curation saves $250 in wasted ad spend.
5
Feed the latest creative strategy research into every prompt so the AI is always working with current best practices.
The bottleneck for content automation has shifted to taste.
The question is how efficiently and effectively you can deploy that taste into the system to make up for the weaknesses of the agent.
So that's what we figured out how to do.
Give the AI every tool it needs and host it on a private server dedicated to your brand.
Feed in the latest research from our creative strategists about what is performing in your niche.
Produce the first handful of videos entirely by hand with rigorous quality control.
Provide beautiful interfaces so we can make tweaks and adjustments while the system is still learning.
Apply just enough human curation at the end so your content looks better than competitors doing the same thing.
10
Our Metrics
AIVideo.com by the numbers
0.0B+
views driven by our marketing team using AI-generated content from our own platform
0B
AI tokens consumed per month to write 99% of all code in our product
0%
of customer support tickets handled by AI automation
11
End-to-End Automation
Sources & Integrations
12
Backed by the best
Our Investors
Paul Graham
Co-founder, Y Combinator
Balaji Srinivasan
Former CTO, Coinbase
Matthew Bellamy
2x Grammy Winner
13
Backed by the best
We Love Hard Projects
Justin Harvey·3rd+
AIVideo.com1yI can't believe we
Turned the Mayor's House
Into an AI Studio
(4,000 video challenge, $1M in editing costs saved!)
Did we just kill video production age?
0+videos produced
$0Min editing costs saved
0 Daysfrom idea to launch