MyBidWise wasn’t about Nara Marketplace from the start.
With the idea that “even a one-person company can make a million dollars” in the back of my mind, I tried building all sorts of things before that. I built a BI dashboard for my younger cousin’s espresso machine business and loaded it with real sales and client data, and I tried building tools for hospital marketing. After building a few, two things became clear. Building itself had become easy enough for anyone to do. And the game isn’t decided by what you can build, but by where you can make money.
So I changed the direction of my thinking. What could I do in a setting where I could actually make money? The answer was simple. The place that holds the most money and spreads it into the market: the government. Wouldn’t the easiest thing be to go where the money is first and offer a service? That thought became MyBidWise.
Every day, hundreds of notices go up on Nara Marketplace. And people were combing through them every morning, typing the same keywords into the search box. Couldn’t I just pick out the notices for your industry and your region and send you an alert? On September 28, 2025, I pushed the first commit.
The notices come in through the Public Procurement Service’s G2B public data API. At first I thought, “There’s an API, so hook it up and you’re done,” but hooking it up was only the beginning. The collector has to run every day, and if it runs every day, it can break every day. This August, the Public Procurement Service switched its default response format from XML to JSON and the collector stopped; I tracked down the cause and brought it back by forcing the format explicitly. That’s when I learned what it means to build a service on government data: when the other side changes its format, I’m the first to find out, the hard way.
Getting paid took the longest. At first I built everything on Stripe, down to an embedded checkout. I even added dark-mode support, then ripped it all out two days later. Only after building the whole thing did I learn that Korean businesses can’t accept live payments through Stripe. So I switched to Paddle, which lets sellers in Korea register and handles each country’s taxes on your behalf. I rewrote the terms and refund policy pages from scratch and changed the refund window to 14 days to match Paddle’s consumer terms.
With payments finally set up, the system missed the very first paid order that came in. Webhook signature verification had failed, so the payment never made it into the DB. In the end, I registered that first customer by hand.
My first sale was processed not by the automated system but by hand. It still makes me laugh, and it still stings.
For an alert service, email is everything, and mine moved house three times too. SendGrid’s free credits ran out, Gmail SMTP got blocked, and I’ve now settled on Resend. Somewhere in between, the daily digest had quietly stopped; I found out late and only fixed it this July. It wasn’t that users weren’t coming. I just wasn’t sending anything.
Build it and wait, and nobody comes. So I started bolting things on. I dug into search traffic with Search Console. With Microsoft Clarity, I watched where people dropped off in the payment funnel. I opened a blog and built up bidding guides, and made card-news carousels for Instagram. Recently I even built a pipeline that collects the day’s trending keywords and, every morning, automatically generates card news and reels that tie them back to bidding and procurement. This side ended up using more tools than development did.
- Claude Code
- All development · monthly subscription
- DigitalOcean
- Droplet (2 vCPU / 4 GB) server · $24 a month
- Supabase
- DB/auth · started on the free tier
- Paddle
- Payments · 5% + $0.50 per transaction (switched from Stripe)
- Resend
- Email · started on the free tier (SendGrid→Gmail→Resend)
- Search Console / Clarity
- Search traffic · funnel analysis · free
If you’ve read this far, you’ve probably noticed: that’s far too many roles for one person. Data collection, payments, email, the blog, social media, and my day job on top. So I started giving each role its own AI. I opened channels on Discord and seated a Claude with a different persona in each one. One for bidding, one for marketing, one for code, one for legal search. At some point it took the shape of an org chart, and I call it Y Tower. As I added what MyBidWise needed one piece at a time, my one-person company had turned into a one-person-plus-AI organization before I knew it.
I kept adding what was needed, and ended up with a company.
Here’s the report card as of now. Alert subscribers now number in the hundreds. Starting from 24 last November, it was 83 in May, 129 in June, and 182 at the end of July. In the last two months alone, 99 were added. 104 members have created accounts, and paying customers: 1. In eleven months I built the collector, payments, email, even apps (I put a BidWise app on the App Store and Google Play), and revenue is still that number. Add one more feature, or go out and sell first? At that fork in the road, I still waver.
There’s a report card beyond the numbers, too. Refining each feature grew my development skills, and along the way I learned who my customers are and how a company actually runs. Dealing with real customers, I felt in my bones how hard it is to get to a single sale, and at the end of that, I got a payment system fully working. You could say I made one full lap of everything a company has to do, down to promotion, advertising, and marketing. Now I’m strengthening each of them, one at a time.
Looking back, though, for all eleven months I was doing one thing: when the collector stopped, finding the cause and bringing it back; moving payments from Stripe to Paddle; switching email three times; and checking again, in numbers, that 24 subscribers had become hundreds.
I’d like to sum up the lessons for you, but I’m still not sure what they are. One thing is certain. Before the next issue, I’m going to go meet people first, instead of building features. I’ll tell you how it went then.
The service in this story lives here: mybidwise.com · Nara Marketplace bid notices, filtered to just your criteria.
Planner and one-person organization. I spent ten years in planning and coordination at the headquarters of a large corporation, and now I’ve built my own legal, finance, and planning departments out of agents and skills. Zero employees, and the organization still runs.