100 Days Building in Public
─ The Real Day 1-8 Log

Last night a family member asked, "Does this coding thing pay anything?" I said no. "Then what are you doing?" I said, "Building credibility." Five seconds of silence, then: "OK. Get some sleep." This one's for everyone who, like me, is still rewriting their hero copy at 2am.

※ To protect the people involved, every conversation with family and friends in this piece has been blurred — identifying details removed, only the core situation kept. All numbers reflect my actual results and aren't meant to be representative.

Day 1I built a website nobody would see

5/19, 8pm. I bought the domain mentry.dev for NT$540/year.

I picked .dev because it "looks like an engineer." Honestly .com is better (Google prefers it), but someone had already grabbed it. At the time I thought — whatever, go with .dev, figure it out later.

I finished exactly one thing that day — the first draft of my hero copy: "I'll walk with you from beginner to engineer."

I stared at it for 30 minutes and decided it was right.

Seven days later I'd kill this version. But that's a Day 7 problem. On Day 1 I knew nothing — I just felt this version was "honest."

Day 2-3Building a site without knowing who it's for

Day 2, I used an AI assistant to build the site. In one afternoon I wrote the hero, about, services, and contact sections. Vanilla HTML/CSS/JS, no framework.

Day 3, I did something that turned out to be really dumb — I added 5 services: courses, 1-on-1, technical partnerships, the newsletter, and "TBD."

"TBD" meant I didn't even know what I wanted to sell.

5 services = 0 clients reaching out.
Because they have no idea "what you actually specialize in."

I later changed it to 5 Stage cards (Getting Started / Halfway / Launch / Scale / Interview) — but that's a Day 6 thing. On Day 3 I was still lost in the "more services is better" fog.

Day 4First wall I hit — nobody's watching

I posted my first IG post. A 6-slide carousel, way too long to design, with 12 hashtags.

The numbers 24 hours later — single-digit likes, 0 comments, 0 new followers, reach in the double digits.

I felt a little deflated that day.

But here's the thing — a brand-new account's first post barely reaching anyone is completely normal. I didn't know that back then. I thought what I'd made just "wasn't good enough."

Only later did I get it: a new account's reach is heavily shaped by account trust, not just content quality. The first few weeks are usually ice-cold — you have to keep stacking to unlock a bigger traffic pool.

Day 5Rewriting the hero, take 1

Day 5, I reread the hero — "I'll walk with you from beginner to engineer" — and it suddenly hit me: "Wait, no. A PM won't click this. A designer won't click this. A founder won't click this."

I talked it through with a friend for 30 minutes:

Damn.

That one line made me realize — I was talking to the outside world in engineer insider-speak, and the outsiders couldn't follow.

Over the next few days the hero went through 5 rewrites:

• V1 "I'll walk with you from beginner to engineer" ← too engineer-centric
• V2 "Your AI-written code? I'll bail you out" ← too passive
• V3 "I'm building myself as a product, in public" ← too abstract
• V4 "AI gives you code, but it can't give you software" ← right concept
• V5 "You have the idea, AI gives you the code, software—I help you build it" ← this is it

The key to V5 — putting "you" first.

In that first second, a visitor isn't wondering "who are you" — they're wondering "is this for me?" The hero has to answer that one first.

Day 6Realizing I was avoiding the real work

Day 6, I spent 4 hours on responsive design. I went from 4 breakpoints to 7. Checked every screen size, one by one.

11pm, a family member walked past, glanced at the screen: "You skipped Threads again today, huh?"

…Yeah.

I'd spent the whole day tweaking the site — didn't write a single post, didn't reply to a single comment, didn't do anything that would actually "get seen."

Tweaking the site = visible progress.
Publishing content = no visible payoff.
People pick the visible thing and dodge the invisible one.

That night I wrote up a "do-not list" and taped it to my desk:

• No touching the hero (unless someone says they don't get it)
• No optimizing responsive design (unless it's genuinely broken)
• No adding new services (unless a client asks)
• No researching new tools (unless I literally can't do the work without one)

The next day — Day 7 — I rewrote the hero again.

That's just how humans work.

Day 7I deleted all 14 AI icons

This was the day I did something a designer might want to punch me for — I deleted all 14 AI-style icons on the site and swapped in Lucide line icons.

Why — that day I was scrolling Threads and saw someone say: "These days you open any site and know within 3 seconds whether it was made with AI. The easiest tell is the icons."

I opened my own mentry.dev.

…Damn. They were right.

All 14 icons were AI-generated — too consistent in style, too much 3D depth, instantly recognizable as the "skeuomorphic look" AI hands you.

I spent an hour hand-picking 14 Lucide line icons. The site instantly felt like "there's actually a real person behind this."

That's the most important thing I learned on Day 7 — in 2026, the scarcest thing isn't AI, it's a human touch. Even the icons have to show it.

Day 8Fixed a bug nobody would ever notice

Day 8 — which is today — I ran PageSpeed Insights and found my mobile performance score was only 51.

The diagnostics said: script.js:30 forced reflow 1,264ms.

1,264ms.

I dug into it for 30 minutes and found the culprit was a poorly written scroll listener (even though I was using rAF batching). After switching to IntersectionObserver, the forced reflow went from 1,264ms to 1ms.

This is what engineering discipline is — using the right tool for the right job.
It's not enough to "know how to write it." It's "knowing which way to write it won't blow up."

Most visitors will never notice this bug. But I noticed.

That's the difference between "building software" and "writing code."

The real numbers on Day 8

It looks rough.

But I chose to make it public — because I believe a real zero is more useful than a polished story.

I don't know if I'll ever take off.

But I know I'll keep writing. Check back in 92 days.

To everyone doing the same thing

If you're also building a personal brand, also come from an engineering background, and are also rewriting your hero at 2am — here are 3 things I want to tell you:

1. Don't wait until it's done to show it. Wait until it's done and you'll discover — it's never done.

2. Tweaking your site is avoidance, not work. The real work is writing content, finding clients, and handling rejection.

3. Your most important supporters might not understand what you do. They don't need to — they just need to believe you're serious about it.

That's the truth of my Day 1-8.

The next post, Day 9-16 — roughly 2 weeks out. Subscribe to the newsletter and follow along.