Inspiration
Moving from Korea to the US just a month ago, I noticed products flying off Olive Young shelves in Seoul were only rarely unknown here. The moment that crystallized it: a number of friends' wives sent them $2,000 Korean beauty shopping lists to bring back as gifts. Korea was ahead in beauty, and the arbitrage window between "trending in Seoul" and "discovered in the US" was wide open and closing fast.
What it does
Signal from Seoul scans Korean retail rankings on Olive Young, Coupang, and Naver hourly, scores each product against six US viability signals (price range, ranking velocity, awards, bundle quantity, and more), stores signal history, and fires a Slack alert the moment a product crosses the threshold — giving US sellers a 3–4 week head start on the market.
How we built it
I used Yutori browser agents for live ranking scrapes, Fastino as the scoring engine, Senso for memory and signal history, and Slack as the alert delivery layer. The pipeline runs: Scan → Score → Remember → Alert.
Challenges we ran into
Signal noise was the hardest problem. Not every trending Korean product translates to US demand. Building a scoring model that filtered intelligently — and didn't fire false positives — required significant iteration on the six viability signals and threshold calibration.
Accomplishments that we're proud of
Noise was the hardest problem. Not every trending Korean product translates to US demand. Building a scoring model that filtered intelligently — and didn't fire false positives — required significant iteration on the six viability signals and threshold calibration.
What's next for Signal from Seoul (SfS)
Speed alone isn't the edge — signal quality is. Anyone can scrape a ranking page. The real insight is knowing which signals predict US viability before the trend crosses the ocean. We also learned that memory matters: re-scoring the same product over time and watching confidence build is what separates a noise alert from a real opportunity.
Built With
- fastino
- gliner-2
- pioneer
- senso
- slack
- uvicorn
- yutori
Log in or sign up for Devpost to join the conversation.