The technical and product decisions behind building a sponsor intelligence platform — from the initial prototype to processing 2M+ YouTube videos daily.
10,000+ brands, 300K+ sponsorship signals — searchable and filterable. Try it free.
Start Free — No Card RequiredSix months ago, SponsorTrace was a Python script running on my MacBook Air M2. Today, it processes 2M+ YouTube videos and tracks 10,000+ brands across 6 affiliate networks.
Here's how we got here.
I was a YouTube creator trying to find sponsors. The process was painful:
The data I needed existed — it was just scattered across millions of video descriptions. Nobody was aggregating it.
We settled on a stack that optimizes for data throughput and developer velocity:
The key decision was keeping everything in Postgres. No separate search engine, no separate vector database, no separate analytics DB. One database, one source of truth.
1. Start with regex, add LLM later. Our regex stage catches 80% of sponsorships for nearly zero cost. The LLM only processes the remaining 20% that need brand identification.
2. Pipeline reliability > pipeline speed. A pipeline that runs reliably every day at 6 AM UTC is worth more than one that's 10× faster but fails on weekends.
3. Free tier is a product, not a demo. Our free plan gives real value — 500 sponsors, 3 AI recommendations/week, basic channel dashboard. It's enough to prove the product works.
Q3 2026 launch. If you're a creator who wants early access, sign up free at sponsortrace.com.
This is part of our Behind the Scenes series.