Case study 02 · My own product · 2026
Woongie: where locals actually eat.
A food guide for Korea that ranks more than 10,000 places using what locals write, not what tourists rate.
The problem
Google reviews barely work in Korea. Locals write about food on Naver blogs, in Korean. So visitors and expats end up at the same hyped places and miss where people here actually go.
What I built
- A data pipeline that reads Naver blog posts about more than 12,000 places across Seoul, Busan, Jeju and other cities.
- A Local Signal Score from 0 to 10. It weighs how many different locals write about a place, how often, how positive they are, whether it's trending and how recent the posts are. Sponsored posts count against a place.
- Everything in English: every name, address and menu translated.
- An iOS and Android app built on Naver Maps. Tap any spot on the map and it finds the place.
- A website with neighborhood guides and 85+ articles.
What AI did, what I did
AI did the heavy lifting: reading thousands of blog posts, translating, summarizing and writing most of the code.
I made the judgment calls. I designed the scoring weights, decided sponsored posts should count against a place, capped chain restaurants so they don't crowd out local spots, and fact-checked more than 185 details on the site before they went live.
It needed it. One check caught 1,379 places where the AI had saved its own error message as the description. A person reading the output is the only reason that never reached users.
Results so far
Live on the App Store and Google Play.
What it shows
This is how I'd work for you. AI turns a mountain of raw data into something useful, and a person decides what's true, what's fair and what ships.
Want something like this for your team?
A few lines about your team and what's slowing you down. I reply within one business day, and if it's a fit, I'll send a link to book a free 20-minute call.
hello@eighttwotwo.comI only use your email to reply to you. No mailing lists, and I never share it.