Two websites, one session each
This week's theme was website. I ended up building two completely different ones. One was a real client project with a deadline. The other was a personal experiment that went sideways in the best way.
The client site: NLG Challenge 2026 (https://www.nextlevelgrowth.fi/challenge/)
Patrik needed a website for Next Level Growth Challenge 2026 — a startup pitch competition here in Vaasa. Deadline: now. Hosting: a basic Zoner FTP server. No Vercel, no GitHub, no build pipeline. Just drop a file on a server and it works.
What we built in one session:
Sticky nav, hero section with a live countdown to the finals
A prominent banner for the Startup World Cup integration — NLG is the official SWC regional for Ostrobothnia, meaning the Scale track winner goes to the Grand Finale in San Francisco competing for a $1M investment prize
Two clearly separated competition tracks: START (idea stage, no company needed) and SCALE (registered company, bigger prize, SWC ticket)
Prizes, partner grid, event details, an Ostrobothnia region context section with map, footer
The most interesting part wasn't the code. It was how the content came together. Instead of working from a brief, we pulled from real sources mid-session — Instagram posts, a LinkedIn article about the Ostrobothnia startup gap, and email chains Anton forwarded from the NLG planning threads. The $1M San Francisco prize detail came directly from an internal email, not anything public-facing yet. That one detail changed how the whole Scale track was positioned on the site.
End result: One 320KB HTML file. One session. Ready to deploy.
The personal experiment: interactive pixel portfolio
The second build started as a simple question — what if my personal website was a world you could walk around in?
That question turned into a Phaser 3 pixel-art island. You control a small cloaked character. You walk around a hand-painted island. When you get close to a building, a prompt appears. You press E and a panel slides open with actual content — about me, projects, writing, contact.
The island itself is a 30×22 tile grid I painted by hand in a JavaScript array — every 0, 1, and 2 placed deliberately to make the coastline look organic. Water surrounds it. A single-tile sand border rings the grass. Four buildings mark the four corners: a Nordic timber cabin, an industrial lab, a stone archive tower, an observatory dome.
What it ended up being: a 1,200-line single HTML file. Drop it anywhere. Open it in a browser. Walk around.
What I actually learned
The NLG site taught me something I already knew but keep forgetting: constraints clarify decisions faster than freedom does. "FTP server, one file" eliminated three hours of infrastructure decisions before I even started.
The pixel site taught me something different: iteration on the wrong abstraction is expensive. I spent multiple sessions tweaking building sizes, colors, and transparency before realising the actual problem was which Phaser API I was using to generate textures. The fix took four lines of code. The wrong path took hours.
The thing both projects had in common: the interesting problems weren't the ones I expected going in.
Week 3 is VES. Something for the entrepreneurship community.
