Case study
Everything Claude Fable
An autonomous content pipeline — with a human still holding the approve button.
Live
everythingclaudefable.com ↗ · Shipped
The challenge
Content-directory sites live or die on one unglamorous problem: someone has to keep finding, vetting, and publishing new material, continuously, or the site goes stale. Doing that by hand doesn't scale; doing it fully automated risks publishing garbage.
The approach (built with Claude Code)
Enable built an automated ingestion-to-publish pipeline: the system continuously pulls candidate content from multiple public sources, uses AI to summarize and classify each item, and holds everything in a review queue for a human to approve before it goes live. Nothing publishes without that approval step. The public-facing side is a fast, SEO-structured Next.js site — category pages, detail pages, a directory-style front end — built around the ingestion engine rather than bolted onto a generic CMS.
What we shipped
A production Next.js/TypeScript site on Railway, fronted by Cloudflare, with a Prisma/MySQL backend and an AI-assisted content-review workflow (/admin/review, access-gated). Shipped Lighthouse scores: Performance 90, Accessibility 100, Best Practices 100, SEO 100. Built across roughly two weeks (10 build phases, 38 commits).
Outcome
Live in production. One spot-check of Cloudflare Web Analytics showed 27 page views / 16 visits in a 24-hour window (2026-06-25) — a single snapshot, not a trend, so it isn't published as a standing traffic figure without a current pull.
Screenshots
Tech notes
Next.js (App Router) + TypeScript, Tailwind + shadcn/ui, Prisma + MySQL on Railway, Claude Haiku for summarization/classification.
