Day 26: Wrap Up
Twenty-six days ago, I started with an empty repository and a simple question: what happens when you rebuild a business website from scratch, document every decision, and ship it in exactly 26 days?
Now I have an answer.
What This Project Was Really About
26to26 was never just about building a website. It was an experiment in disciplined iteration—proving that meaningful work happens when you show up every day, make one thing better than it was yesterday, and resist the temptation to chase perfection at the expense of progress.
The daily constraint forced honesty. There was no hiding behind “I’ll polish this later” or “Let me think about it more.” Each day demanded a deliverable, and each deliverable demanded decisions.
The Top 3 Days
Looking back across all 26 days, three stand out as defining moments:
1. Day 20: Accessibility Audit
This was the most humbling day of the project. After 19 days of building what I thought was a thoughtful, well-crafted site, a proper accessibility audit revealed gaps I hadn’t anticipated: keyboard navigation that didn’t work for radio groups, color contrast that failed WCAG AA in ways I’d missed, heading hierarchies that made no sense to screen readers, and an auto-advance behavior that was actively hostile to keyboard users.
Fixing these issues wasn’t glamorous work. It meant rewriting interaction patterns, creating new color token variants, and rethinking assumptions baked into the design from Day 1. But by the end, we hit 100% on Lighthouse accessibility—not as a vanity metric, but as proof that the site actually works for everyone.
The lesson: accessibility isn’t a feature you add at the end. It’s a lens that should inform every decision from the start. I learned this the hard way, but I learned it.
2. Day 1: Foundation and First Pixels
Day 1 set the philosophical foundation that carried through the entire project. The decision to embrace Japanese aesthetic principles—Ma (negative space) and Kanso (simplicity)—wasn’t just design philosophy. It became a filter for every subsequent decision.
When I was tempted to add another gradient, another animation, another decorative element, Ma asked: “Does this space need to be filled?” When complexity crept in, Kanso asked: “What can be removed while keeping the essence?”
Beyond philosophy, Day 1 established the technical foundation: Astro 5 with TypeScript, Tailwind CSS 4, Lighthouse CI from the very first commit, and property-based testing for design system compliance. These weren’t premature optimizations—they were guardrails that prevented drift over 26 days of rapid iteration.
3. Day 15: Comprehensive E2E Testing with Playwright Agents
This day represented the project’s commitment to quality through innovation. Using AI agents in a Planner → Generator → Healer loop to expand test coverage wasn’t just efficient—it was a proof of concept for how AI-assisted development can raise the bar rather than lower it.
The Planner agent identified gaps in coverage. The Generator agent wrote tests to fill them. The Healer agent fixed tests when they broke. The result: comprehensive E2E coverage that caught real bugs and prevented regressions, with a workflow that could be repeated and refined.
This day embodied the 26to26 philosophy: use every tool available to ship something better than you could alone, but never let the tools replace judgment.
What I Shipped
The numbers tell part of the story:
- 26 days of documented progress
- 4 custom results pages with personalized recommendations
- A multi-step assessment form with branching logic
- Case study infrastructure with content collections
- Comprehensive test coverage: unit, property-based, E2E, and visual regression
- 100% Lighthouse accessibility score
- Custom analytics with 10+ tracked events
- Mobile optimization with self-hosted fonts and safe area support
But numbers miss the point. What I actually shipped was a site that reflects how I want to work: direct, honest, focused on substance over flash. The bold typography, the stripped-down design, the Japanese aesthetic principles—these aren’t decoration. They’re a statement about what kind of work I want to do and who I want to do it for.
Please take a look at https://26to26.shikadigital.co.jp for what we have done in the past 26 days.
What I Learned
Constraints create clarity. The 26-day deadline eliminated endless deliberation. When you can’t defer decisions, you make them—and most of the time, the first instinct was right.
Documentation forces understanding. Writing about each day’s work revealed gaps in my own reasoning. If I couldn’t explain why I made a choice, I probably hadn’t thought it through.
Accessibility is non-negotiable. Day 20 proved that even careful work misses crucial details. Building for everyone isn’t a nice-to-have; it’s the baseline.
AI assistance amplifies, not replaces. Claude helped me move faster, test more thoroughly, and catch mistakes earlier. But every decision still required human judgment about what mattered.
Simplicity requires discipline. The hardest work was removing things, not adding them. Every round of simplification made the site better.
What’s Next
Tomorrow, this site goes live. Not as a finished product—no site ever is—but as a foundation that can evolve.
The 26to26 experiment ends, but the approach continues: show up every day, make one thing better, ship what you build, learn from what breaks.
Thank you for following along.
See you on the other side.
The site launches January 1, 2026.