Make It Look Like You Mean It
Day 17. Week 4. The DOA Archive is done, documented, and submitted — and I was the first one to turn it in.
The App Is Actually Usable
I keep saying that about things I build, but I want to be clear about what I mean. Stripe works. The embeddings work. I can add entries and Booker's semantic search gets smarter with each one. The auth tiers — guest, member, admin — are enforced at the database layer, not just hidden in the UI. It deploys on Vercel, runs on a phone, and doesn't look like a weekend experiment someone cobbled together.
I wanted to get it to the point where it could function as a real asset in my world-building. A living lore database for the novel. A place readers and collaborators could actually use. That's where I got it.
The Demo Is Also a Product
Today was all about the video. And I made a decision early: I wasn't going to do a student demo. I was going to do a product launch.
I know how to make a basic tutorial video — I've probably put up 300+ videos over the past several years. At some point in 2023 I decided I was bad at (and very intimidated by) making videos and the fix was to make 120 of them in a year with only one rule: complete enough to upload. That year broke the fear. Putting up a video isn't easy, but it's not nearly as hard as it used to feel.
So when I sat down to record the demo for the DOA Archive, I wasn't starting from zero on the production side. I learned the OBS zoom feature so the code walkthrough would actually be watchable. I pulled Escape from Okinawa assets — the trailer song, the HAC Studios bumper and footer — and cut them in. I made transitions in Canva. I even left in the blooper reel at the end where I said "Strap" instead of "Stripe," because that's more human than a flawless take.
When I rewatched it — several times — it doesn't read like a class project. It reads like someone made this app and is genuinely interested in the novel behind it. That was the whole point.
I don't know how they'll score it, and honestly that wasn't the goal. The goal was to get real squeeze out of the assignment. To make something I'd be proud to show people outside of this context. I accomplished that.
Two Out of Three on My Own
The other thing worth documenting: CodeWars. We did three algorithm problems today as a group. Two of them I worked through completely on my own before the walkthrough caught up to me. Start to finish. No AI assist.
That's a different experience than where I was even a week ago.
The logic has never been the hard part for me. Break the problem into steps. Do A, then B, then C. The blocker has always been the syntax — not knowing which function handles what in JavaScript, or how Python does the same thing differently, or which one I'm even supposed to be thinking in since we're running both in parallel.
But that's just vocabulary. And vocabulary comes from reps. FreeCodeCamp makes me type a lot of code. CodeWars makes me apply it under pressure. The class problems give me a reference point. It's accumulating.
I default to the AI the same way you'd raise your hand for the instructor — when I genuinely don't know where to start. The goal is to need it less. Today I needed it less.
This week the agentic coding block ends and we move into more traditional algorithms and data structures. I'm done early, which means the rest of the week is mine — FreeCodeCamp, CodeWars, NeetCode, whatever I can get through.
More reps. More vocabulary. The finish line keeps moving, which means I keep moving.