Back to Work Samples
AIContentWordPress

AI Content Pipeline

A 12-stage automated content creation system that discovers trending topics, researches them deeply, writes articles with genuine personality, and publishes to WordPress — all without manual intervention.

The Problem

  • Researching and writing quality articles takes 4-8 hours each
  • AI-generated content often sounds robotic and generic
  • Capturing screenshots manually is tedious and inconsistent
  • Publishing workflow has multiple manual touchpoints

The Solution

  • End-to-end automation from discovery to publish
  • 12 emotional arcs create authentic-sounding content
  • Puppeteer captures contextual screenshots automatically
  • WordPress integration with proper metadata

12-Stage Pipeline

Each stage is modular and can run independently. Failures in one stage don't block others.

1

Discovery

Monitor Hacker News, Product Hunt, GitHub for trending AI tools

2

Trend Analysis

Claude scores topics 0-100 based on popularity signals

3

Auto-Approval

High-scoring topics (80+) automatically queue for processing

4

Deep Research

Fetch GitHub README, website content, documentation

5

Emotional Arc

Select from 12 narrative styles to avoid generic AI voice

6

Competitive Intro

Generate intros with multiple AI models, pick the best

7

Article Writing

Claude generates full article using template and style guide

8

Humanization

Inject personality based on emotional arc phases

9

AI Review

Multi-model review for clarity, accuracy, and tone

10

Screenshots

Puppeteer captures contextual screenshots automatically

11

WordPress Publish

REST API uploads draft with metadata and categories

12

Validation

Check formatting, sanitize code blocks, verify images

The Secret Sauce

Emotional Arc System

Instead of generating identical-sounding articles, each piece gets a unique narrative trajectory. This is how we avoid the “AI voice” problem.

Skeptic Converted

Start doubtful, end impressed

Best for: Overhyped tools that actually deliver

Pleasant Surprise

Casual curiosity becomes enthusiasm

Best for: Lesser-known gems

Frustration to Relief

Struggle leads to breakthrough

Best for: Complex but rewarding tools

Technical Deep-Dive

Analytical discovery of quality

Best for: Developer-focused tools

Mixed Feelings

Nuanced pros and cons

Best for: Tools with clear trade-offs

Instant Fan

Love it from the start

Best for: Genuinely exceptional tools

How Arc Selection Works

  • Claude analyzes research data to match the optimal arc
  • Tracks recent articles to enforce variety (min 3 articles before repeating an arc)
  • Adds 20% randomness to avoid predictability
  • Each arc has banned phrases to prevent robotic patterns

Article Templates

Different content types get different structures optimized for their purpose.

Overview

5-6 sections

Broad introduction to tool capabilities

Tutorial

7-8 sections

Step-by-step usage guide with screenshots

Comparison

9 sections

Side-by-side analysis with competitors

Troubleshooting

8 sections

Problem/solution format for common issues

Multi-Model AI Strategy

Different models excel at different tasks. We use each where it's strongest.

Claude Sonnet

  • Deep research synthesis
  • Article generation
  • Emotional arc selection
  • Humanization passes

GPT-4o

  • Competitive intro generation
  • Creative alternatives
  • Extended thinking analysis

Gemini Pro

  • Multi-model review
  • Image generation prompts
  • Final quality checks

Results

~$0.75
Cost per article
AI API costs only
3-4
Articles per week
Consistent output
12
Emotional arcs
For authentic voice
0
Manual writing
Fully automated

Tech Stack

Node.jsClaude APIGPT-4o APIGemini APIPuppeteerGitHub ActionsWordPress REST APIGray MatterSharpRSS ParserCheerioYAML

Need content automation?

I can build custom content pipelines that research, write, and publish at scale — while maintaining quality and authenticity.