Notes archive

Why I Built a Goshuin Gallery

A reflection on turning a personal goshuin collection into a small gallery site, and using it as practice for AI-assisted product development.

In the spring of 2022, I started collecting goshuin: sacred stamps and calligraphy received from shrines and temples in Japan. Each one records a real visit, but it also carries something less formal: the weather that day, the neighborhood around the temple, the feeling of arriving somewhere quiet, or the memory of a short trip.

At first, the collection was private and physical. That is part of the charm. A goshuincho is meant to be held, opened, and revisited slowly. But over time I also wanted a digital place where the collection could be seen as a map of experience rather than just a set of images.

Why a Gallery Felt Right

When I looked around online, I found many goshuin-related pages that were useful but not always pleasant to browse. Some felt visually crowded. Others felt like older database pages: functional, but not very close to the atmosphere of visiting a shrine or temple.

That gap made the idea interesting. Goshuin are visual, personal, and location-based. They deserve a presentation that is calm, readable, and respectful. A gallery format felt like the right starting point because it can keep the focus on the stamps while still leaving room for place, date, and memory.

I spent a meaningful amount of effort on the styling because the visual tone matters for this kind of collection. The site needed to feel quiet and deliberate, not like a generic image grid. Typography, spacing, color, and navigation all had to support the same mood: respectful of the subject, simple enough to browse, and personal enough to feel like my own archive.

The result is Nicole no Goshuincho, a personal sacred stamps gallery at goshuin.nicolemay.me. The current site presents the collection with a Japanese-first identity, simple navigation, and sections for browsing the gallery, viewing places on a map, checking collection statistics, and reading more about the project.

Building With an AI Coding Agent

The project also became a practical exercise in AI-assisted development. I wanted to see how far I could go by describing product direction, visual mood, and interaction goals clearly, then using an AI coding agent to help turn those decisions into a working site.

That process is useful, but only when the human side stays active. The agent can produce code quickly, but it cannot decide what the site should feel like, what details should be included, or what should remain simple. Those decisions still need taste, review, and restraint.

For this project, the useful part was not speed alone. It was the repeated loop of writing a requirement, reviewing the result, adjusting the content or layout, and deciding whether the change made the site clearer. That loop made the gallery a good practice ground for working with AI without treating it as a replacement for engineering judgment.

What I Wanted the Site to Express

I did not want the gallery to feel like a travel checklist or a casual photo dump. I wanted it to feel closer to a personal archive: quiet, visual, and structured enough to grow over time.

The homepage phrase, "The journey itself became the destination," captures the reason I kept working on it. Goshuin collecting is not only about completing a list of shrines and temples. It is also about the route, the visit, the small memory attached to each place, and the habit of paying attention.

That is also why the site includes more than a simple image grid. A gallery, map, and statistics view each tell the collection from a different angle. The gallery shows the objects. The map gives them geography. The statistics make the accumulation visible without making it the main point.

What I Learned

This project reminded me that small personal software can still benefit from product thinking. Even a simple gallery needs decisions about audience, tone, navigation, data structure, visual hierarchy, styling, and what not to build yet.

It also gave me a better sense of where AI-assisted development fits my workflow. It is strongest when the project has a clear shape and the next step is specific. It is weaker when the direction is vague or when taste matters more than implementation.

For me, the Goshuin Gallery sits between a personal collection and an engineering exercise. It is a way to preserve visits that matter to me, and it is also a place to practice building software with clearer intent.