About

How this site is structured, built, and operated.

https://the-shashi.com/about/ / Last updated: 2026-07

About This Site

The社史は、日本の上場企業を中心とした企業史・経営判断をまとめた個人サイトです。「ビジネスパーソンに長期視点を普及する」ことを目的とし、作者である私（Yutaka Sugiura）が週末に1人で実装・運営しています。サーバー代を1日100円未満に抑えつつ、10年来の趣味としてコツコツと楽しく作っています。

Development Approach

Built AI-agent-first, with the human-facing experience optimized for desktop (PC).

Page access breakdown for this site (a day in July 2026)
  • Human (PC) 44.1%
  • Human (SP) 6.6%
Human total 50.7%
  • ChatGPT 37.5%
  • Meta AI 7.7%
  • Perplexity 3.1%
  • Claude 1.0%
AI total 49.3%
Based on AWS CloudFront log analysis. The share of accesses to target pages (static assets excluded) from humans (PC / SP) and the major AI crawlers (ChatGPT / Claude / Meta AI / Perplexity). Bots are identified by UA plus ASN/IP; search and SEO bots (Googlebot / bingbot / Semrush, etc.) and stealth scrapers from data-center ASNs (M247 / Oxylabs / HostRoyale, etc.) are excluded. Unidentified bots and hostile traffic such as vulnerability scans are also separated out of the tally.

Automation Pipeline Infrastructure

  • Research and analysis are driven by large-scale quantitative and qualitative data.
  • For quantitative data, annual securities reports going back 20 years are covered, with the most recent decade parsed and structured from XBRL across roughly 3,600 companies.
  • For qualitative data in particular, a private research corpus—centered on paper materials I have personally collected over the past decade and digitized via OCR—informs how analytical questions are framed and how the historical narratives are written.
  • Routine work is implemented in Python to keep token consumption low, so the human can focus on framing the issues and writing by hand.
  • A self-built access-log platform analyzes crawl behavior — bots included — to guide key editorial decisions (discontinuing mobile-specific development, expanding to an English edition, and so on).
Overview of the corporate-history writing pipeline (5 domains × 36 steps)
Overview of the research & analysis pipeline
Example of the XBRL extraction tools (Python batch)
XBRL extraction tools (Python batch)
Access-log analytics dashboard (per-page stats and daily access trends)
Log analysis & automated improvement-report generation
Working screen for analysis, research, and writing (Ghostty terminal)
The analysis / research / writing workspace

Delivery Infrastructure

Built entirely around static delivery — zero EC2/ECS/RDS — to cut costs while still handling heavy, large-scale bot traffic.

← Back to Home

© 2026 Yutaka Sugiura. All rights reserved.