Identify breakpoints and insert relevant ads
Companies can create Projects
Tag projects and upload video ads
Upload Content to monetize

Mad Ads — Smart Ad Matching for Content Creators

Inspiration

For years, ads have been annoying.

Creators hate interrupting their content.
Viewers hate irrelevant, random ads.
Brands hate wasting money on impressions that don’t convert.

There was a time we were all just scrolling TikTok/YouTube and thinking — “wait… this video would be PERFECT for a Nike ad” but instead the platform was sending a totally irrelevant, generic filler ad.

Meanwhile — billions of dollars are burned every year on spray-and-pray advertising.

Global digital ad spend = $740B+ by 2025
65%+ of advertisers say they struggle to reach the right audience
54% of creators say brand deals are their #1 revenue source but only 12% actually get inbound brand sponsorship requests

This is inefficient.

That inefficiency became the spark.

We wanted to build a world where ads don’t interrupt — they blend.

Where ads become native, contextual content, not just an interruption.

That is where Mad Ads started.

What it does

Mad Ads is an AI ad marketplace that auto-matches brands to creators and then programmatically inserts the right ads into the right moment in a video.

Flow:

Brands upload their product + ad videos.
Creators upload their raw content.
Our model:
- understands the context of the content
- finds breakpoints — natural cut points in the video
- chooses relevant ads based on semantic match
- edits the video to insert those ads precisely where they make sense

In the future — Mad Ads will not only place ads.
It will generate ads.

Creators don’t need to pitch brands.
Brands don’t need to cold-email creators.
The entire matchmaking + insertion pipeline is automated.

How we built it

Computer vision + scene segmentation to detect breakpoints
Vector embeddings for contextual similarity between content themes and product categories
Rule-based and ML-based ad selection scoring
Automated video editing pipeline to stitch ad clips into the creator content timeline

We made creators and marketers speak the same language — through embeddings.

Challenges we ran into

Finding breakpoints that feel natural (audience perception matters more than just scene cuts)
Dealing with variable video formats, codecs, fps differences
Semantic similarity scoring that avoids false positives
Making sure ads don’t ruin pacing or comedic timing

Accomplishments that we're proud of

Fully automated ad insertion without manual editing
A pipeline that helps small creators access brand money — not just the top 1%
Interpretation of content at semantic level (not just keyword matching)