Skip to products
Park & Koh Works

Independent product studio · Seoul

We build software people actually use.

Park & Koh Works is an independent studio run by two people. We find a problem worth solving, build the product, put it in front of real users, and let what happens next decide the roadmap. 3 products are live today.

Products

Everything we’ve shipped, still running.

3 products, all live, all serving users right now. Each one started as a problem one of us kept running into.

01Consumer · Self-understanding

Read Me나를 읽다

A 16-type tendency test and a saju birth chart, analysed together — including where the two disagree.

Most personality tools hand you one frame and stop. 나를 읽다 runs a 16-type tendency test alongside a traditional Korean saju birth chart, then reports the points where the two results agree and the points where they contradict each other. The contradictions are the part people come back for.

Live
  • 16-type tendency test, about five minutes
  • Birth chart generated from birth date and time
  • Agreement and conflict mapped between the two readings
  • Detailed AI report delivered 1–3 minutes after payment
  • No account, no email, no ads, no subscription
Languages
한국어 · English
Pricing
Basic reading free · full AI report ₩2,900 one-time, VAT included
02B2B · Funding intelligence

startup-tips

Tells a company which government funding programmes it can actually apply for — and what it would take to unlock the rest.

Country portals answer “what exists here.” startup-tips answers “can we apply, and if not, why not.” It judges a company against each programme's own published text, marks every requirement eligible, conditional or blocked, and cites the line it read. Requirements the announcement never states are marked as needing confirmation rather than guessed at.

Live
  • Eligibility judged per programme: eligible, conditional, or blocked
  • Every verdict cited back to the announcement text
  • Unlock conditions — local entity, eligible partner, minimum consortium size
  • Collected from six countries' and unions' official portals and public APIs
  • Search by country, sector, deadline or administering body
Programmes collected
789
Currently open
655
Administering bodies
439
Countries & unions
6
Languages
한국어 · English
03Consumer · Careers

JobAgent

Reads your resume and the job posting as one problem, and shows you what a recruiter will see.

A Korean 경력기술서 and an English job description are treated as a single problem, not two documents to translate. Every requirement in the posting is judged against the line in the resume that proves it, and rewrites are cited to what you actually wrote. When a number is missing, JobAgent asks you for it instead of inventing one.

LiveHuman review tier not open yet
  • Requirement-by-requirement match against the posting
  • ATS readability check on the document as submitted
  • Line-level rewrites, each tied to the line it came from
  • Korean resume ↔ English posting, matched by concept not translation
  • Numbers are never fabricated — gaps come back as questions
  • No sign-up, no card — resume goes to the model for analysis, never to our servers
Languages
English · 한국어
Pricing
AI analysis free · human reviewer tier announced, not yet open

About

Two people, 3 products, no clients.

Park & Koh Works is a product studio, not an agency. We don’t take briefs and we don’t build other people’s roadmaps. Everything on this site is something we decided to make, shipped under our own name, and still operate.

The pattern repeats: someone hits a problem that should already be solved — a personality test that only tells you one thing, a funding portal that lists programmes without saying whether you qualify, a resume you cannot see the way a recruiter will. We build the version that answers the actual question, put it online, and find out whether we were right.

Most of what we build leans on AI, because for these problems it is the fastest route to a useful answer. None of it is AI for its own sake.

01Notice
A problem one of us actually has, or watched someone else grind through. Never a market we read about.
02Build
Smallest thing that answers the question honestly. Weeks, not quarters.
03Ship
Public domain, real users, no waitlist. The product has to survive contact.
04Watch
Where people stop, what they retry, what they never touch. That is the roadmap.

Principles

How we decide what to build, and what not to.

  • 01

    Ship it, then find out

    A product in front of a real user teaches us more in a week than a quarter of planning does. We put the first working version in public early and let the response redirect us.

  • 02

    The model is a component, not the product

    AI is the fastest way we know to solve certain problems, and the wrong tool for plenty of others. Every product here has rules, data and judgement around the model — the model alone was never the answer.

  • 03

    Say what the answer is based on

    Our products cite the announcement they read, the resume line they judged, the two readings that disagreed. An answer you cannot trace is an answer you cannot act on.

  • 04

    Ask for as little as possible

    No forced accounts, no dark-pattern subscriptions, no collecting data because it might be useful later. Most of what we build works before you have told us who you are.

  • 05

    Build for whoever has the problem

    Our problems started in Korea; they were never only Korean. We build in Korean and English from the first commit and let the product go where the problem is.

  • 06

    Small on purpose

    Two people, no outside money, no client work. It keeps the distance between noticing a problem and shipping something for it measured in weeks.

Building next

More things are
being built.

There is always a next one in progress. We announce products when they are live and usable, not before — so this page stays a list of things that exist rather than things we intend.

Got a problem that keeps costing you time, or want to hear when the next one ships? Write to us — a real person reads it.

park.koh.works@gmail.com