Peter Drucker
“There is nothing so useless as doing efficiently that which should not be done at all.”
Peter Drucker

01 · Woongie

1.6 million Korean blog posts, read so you don't have to.

  1. Finds the places

    Every restaurant is scored from what Koreans actually wrote about it, not from a star rating a tourist left.

  2. Filters to the ones locals still use

    A place only surfaces once 110 or more people have blogged about it in the last six months. 5,837 clear that bar out of 7,761.

  3. Reads in English, shares in one tap

    Korean reviews summarized in English, and a place you can send to a friend without explaining where it is.

Woongie map view, places pinned across a Seoul neighborhood
A single Woongie place page
Woongie ranked list of places
A Woongie place page: Korean menu items with English names alongside

02 · ArriveKorea

A guide site is only as good as the day it was last checked.

Everything a person needs in their first year in Korea, written for search and for the assistants people now ask instead.

119
pages live

Shaped by what people actually search

Guides, travel, schools and community answers, shaped by the questions people were already asking.

0
approvals to ship

No budget line,
no vendor, no ticket

119 pages went live with no procurement cycle and no queue to sit in. What limits most content work is not budget, it is how many people have to agree first.

3
products, one pipeline

It finds them, writes them up, then checks back

Discovery, scoring, English summaries and closure checks run as scheduled jobs. Weak output gets audited and rewritten, not corrected by hand.

03 · MarketSignal Korea

Eight answers in. A ranked plan out.

A company tells the diagnostic how it currently sells. It returns a readiness score for Korea, the six things driving that score, and which one to fix first.

MarketSignal Korea
Diagnostic
Benchmarks
Recommendations
Platform mix
Localization plan
Report
Blueprint
Export

Diagnostic Session 8412 Readiness score

Korea readiness

Eight answers, scored across six weighted categories.

49/ 100 overallModerate risk
100030 / 100CRMPlatformCRMLocaliz.BudgetContentCreative

Hover a point. CRM Infrastructure is the weakest at 30, but not the first thing to fix.

CategoryWeightScoreSignal returned
Localization Readiness25%38/100Your current landing experience may not match Korean consumer information expectations.
Platform Alignment20%45/100You have some Korea platform presence, but critical gaps remain.
Budget Signal Strength20%72/100Budget strength supports full ecosystem signal generation.
CRM Infrastructure15%30/100Your CRM infrastructure may be reaching a fraction of Korean consumers.
Content Ecosystem10%55/100Some content ecosystem presence. Expanding reach could accelerate acquisition.
Creative Refresh Cadence10%60/100Your refresh cycle is slower than Korean brand benchmarks.

The product ships a score and a written blueprint. This screen does not exist. I drew it to show what those numbers turn into once somebody has decided what matters. Deciding was the work. Drawing it was an afternoon.

What I can claim

I don't hold the certificates.
The platforms underneath do.

Certified platforms, chosen on purpose

Cloudflare has held SOC 2 Type II and ISO 27001 since 2019. That is Cloudflare's audit, not mine, and saying so plainly is the point. I pick infrastructure for what it is certified to do, then stay inside it.

The processors get counted first

One assistant is usually three companies: the interface, the model, the host. Each one has to be named in a DPA before it sees anything. That count happens before a tool is connected, not after.

Components

I learned how Naver ranks so you don’t have to.

Naver is Korean search, and it blocks the fetchers every AI tool uses. So the model guesses. These two skills hand it the ranking rules, and a reader that gets in.

naver-reader

Reads Naver blog posts that normal fetchers cannot open. Body, photo captions, image counts, the embedded Place. Many URLs at once.

naver-blog

How Naver actually ranks. C-Rank, D.I.A.+, SmartBlock, AI 브리핑, keyword method, title and body limits, and what gets a post penalized.

Free, MIT licensed

Set up in under a minute.

Two skills. No API key. Works with Claude Code and Cursor.

1
Pick your tool

We'll give you the exact command to run.

Anything that reads a file can use them.

2
Run this

3
Then just ask

Give it Naver URLs and a question. It reads them before it answers.

Read these three Naver posts and tell me why the top one outranks mine.

I built it to be annoying.

I kept swiping reminders away and forgetting them ten seconds later. So I built one that will not let me: it pings at the time, then every five minutes until you deal with it.

NagBot09:41
send helen the budget friday 4pm
TaskSend Helen the budget
WhenFriday, 16:00
Heads-up15:55, five minutes before
RepeatEvery 5 min until closed
Scheduled. I’ll keep asking until you close it.
Send Helen the budget

Due now. If you want this closed, type 4821. Otherwise I ask again in five minutes.

Type it the way you would say it out loud.

Start your day with a Jarvis for your inbox.

Nothing important gets missed, and nothing unimportant gets your morning.

Mail Search mail

Sorted

Folders

Inbox74 Sent Archive

2 of today’s 6 meetings have no agenda and name no decision. Those are the two worth moving.

Your industry probably isn’t on this list.
It wasn’t on mine either.

  • PGA Tour
  • EF Education First
  • New York Life
  • Tampa Bay Rays
  • LX Pantos
  • U.S. House of Representatives

About & contact

Let's talk.

I'm Steve Nam, an AI Operator. Growth and operations across very different industries, the last three years building the workflows that make real teams faster. If you're working out how AI fits yours, I'm easy to reach.