Swire, the first business plan I ever wrote

The timing made it sharper. After the COVID lockdowns lifted, demand came back faster than staff, and businesses were wary of hiring full time. I surveyed 241 Vietnamese students. 85.5% were looking for an internship and 88.4% said they had difficulty finding one.

Swire did not start as an internship app. My partner and I joined the Diamond Challenge late with an app that would match freelancers and clients one to one. Near the end of the competition my own market analysis showed that launching into such a crowded market was impractical.

So we narrowed to a problem I knew first hand, interns instead of freelancers, a market with no direct competitor. We rewrote the business model and the financial model, about 1,500 words, in two weeks. The survey was how we checked the new idea was worth building.

It was my first business plan, and research is what carried it. I sized the market from the bottom up, about 5.45 million secondary and 2.6 million post-secondary students, and 505,000 companies with 79% short of labor. That gave a reachable market of 4.1 million students and 398,950 companies, served only by job sites built for full-time roles.

Job boards made students search and apply one listing at a time. Swire turned that into a deck of cards. Students browse internships by field, open a role, and swipe to accept or pass. Businesses swipe on candidate profiles the same way, and both sides stay anonymous until they match. A match opens messaging and video chat inside the app for the interview.

I designed the screens in Figma and programmed the iOS interface in Swift. Combine drove the data bound UI, so cards and lists refreshed as match data changed, and Dispatch kept the asynchronous work off the main thread.

The home screen leads with search, then recommended internships by field and the most popular roles.

A swipe app is only as good as the next card. I built a real time matching service in Node.js using collaborative filtering. Each swipe is a signal about what a student or a business wants, and the service uses the patterns across users to rank what each person sees next.

Firebase carried the back end. Firebase Auth handled token based sign in, and Firestore stored the app's data as NoSQL documents.

Our first plan could not show a profit, which meant many sleepless nights. The fix was adding advertising on top of the subscription. Businesses pay per click or per thousand impressions, priced below local competitors at $3.40 and $4.03, and students get 100 free swipes a day with a $2.59 a month premium tier.

To keep the build affordable the plan was to outsource part of it to a Vietnamese studio at about $15 an hour, with me building the rest. The estimate came to about $8,500 for 550 hours. The deck then projected revenue over five years and a growth path from Vietnam to Southeast Asia, Asia and then global.

For the Diamond Challenge I wrote the business concept, from the problem and the matching model to the market, the financials and the growth plan, and produced a five minute pitch video around an animated walkthrough of the app.

As I wrote afterwards, "This was my first time producing a business plan, so research was an important aspect that has helped me succeed." Research is also what told me to drop the first idea.

In 2026 I redesigned Swire from scratch, built in Claude Code, to show how I would design it today. The idea is unchanged. All content is placeholder.

Setup asks for interests once, when you can start and whether remote works, so the first deck is already relevant.

The 2022 plan described a model that learns from every swipe. The redesign makes that visible. Each card carries a fit score that updates as you swipe, and you can drag or tap, with an undo.

When a business likes you back it becomes a match, kept apart from the roles you only applied to.

The business side, which the 2022 plan served through candidate profiles, can now post a paid role in under a minute.