← All Projects
In Development

Fizz Culture — AI Publishing Engine

A fully automated content pipeline that discovers topics, writes articles with distinct AI personas, and publishes to Ghost.io — designed to mirror how major publishers operate, but run entirely by one person and AI.

The Problem

Running a content publication normally requires writers, editors, SEO specialists, and a publishing team. For a solo operator, it's impossible to maintain a consistent posting schedule across multiple topics and voices without automation. The goal: build what The Verge has, but with AI.

What I Built

A multi-brand content engine using an object-oriented Python architecture. External sources (Polymarket, Google News, RSS) feed a topics queue. AI agents with defined personas (distinct voices, writing styles, topic affinities) generate articles in multiple formats — listicles, opinion pieces, deep dives. Articles publish to Ghost.io via API with affiliate links, metadata, tags, and SEO optimization auto-applied on a configurable schedule.

Content Pipeline Architecture

📡ExtractorsPolymarket · Google News
RSS · Transcripts
📋Topics QueueSQLite · Deduplication
Persona assignment
✍️AI WritersClaude API · 3 personas
Multiple article formats
🚀Ghost.ioAuto-publish · Affiliates
SEO · Schedule
🎛️DashboardFlask web UI · Brand switching · Queue management · Extractor controls
Automation7 launchd scripts · Every 5 min to monthly · Fully scheduled pipeline

Tools & Stack

PythonClaude APIGhost.io APIFlaskSQLitelaunchd / CronJinja2RSS / PolymarketOOP Architecture

Key Features

  • Multi-brand architecture — one engine, multiple publications
  • 3 AI personas with distinct voices, topics, and writing styles
  • Base class inheritance pattern (Agent → Extractor → Destination)
  • 7 scheduled automation scripts via macOS launchd
  • Flask dashboard for managing the entire pipeline
← PreviousSocial Virality System