LinkedIn AI Content Automation System

An AI-powered LinkedIn content engine built with Claude Code that transforms manual content creation into a fully automated pipeline.

---

System Architecture

LinkedIn Posts
      ↓
Content Analysis Agent
      ↓
Knowledge Base
      ↓
Idea Generation Agent
      ↓
Draft Writing Agent
      ↓
Hook Optimization Agent
      ↓
Content Scheduler
      ↓
Publishing Workflow
      ↓
Analytics Collection
      ↓
Feedback Loop

---

Folder Structure

linkedin-ai-system/

├── prompts/
│   ├── hooks.md          # Hook generation prompts & formulas
│   ├── drafts.md         # Full post draft writing prompts
│   ├── ideas.md          # Content idea generation prompts
│   └── repurpose.md      # Content repurposing prompts

├── workflows/
│   ├── content-analysis.md     # Workflow 1: Analyze top posts
│   ├── content-generation.md   # Workflow 2 & 3: Idea + Draft pipeline
│   ├── engagement.md           # Workflow 6: Comment & reply automation
│   └── analytics.md            # Workflow 7: Performance tracking

├── data/
│   ├── posts/            # Raw LinkedIn post exports (CSV/JSON)
│   └── analytics/        # Engagement metrics exports

├── outputs/
│   ├── ideas/            # Generated content idea batches
│   ├── drafts/           # Completed post drafts
│   ├── hooks/            # Hook variation batches
│   └── reports/          # Weekly analytics reports

└── README.md

---

Quick Start

Phase 1 — Content Generation

1. Export your top LinkedIn posts into data/posts/
2. Run the Content Analysis workflow (workflows/content-analysis.md)
3. Use the Idea Generator prompt (prompts/ideas.md) to generate 50 ideas
4. Use the Draft Writer prompt (prompts/drafts.md) to convert ideas to posts
5. Save outputs to outputs/drafts/

Phase 2 — Content Repurposing

1. Feed existing blogs, newsletters, or videos into the repurpose prompt (prompts/repurpose.md)
2. Generate LinkedIn posts, carousel outlines, and thread formats
3. Save to outputs/drafts/

Phase 3 — Engagement Automation

1. Export comment data into data/analytics/
2. Run the Engagement workflow (workflows/engagement.md)
3. Get categorized reply suggestions

Phase 4 — Analytics Feedback Loop

1. Export weekly metrics into data/analytics/
2. Run the Analytics workflow (workflows/analytics.md)
3. Reports saved to outputs/reports/

---

Tech Stack

Layer	Tools
Core AI	Claude Code, Anthropic API, OpenAI API
Automation	n8n, Make, Zapier
Data Storage	Airtable, Notion, PostgreSQL
Analytics	Google Sheets, Looker Studio

---

LinkedIn Post Anatomy

HOOK (First 2 lines — visible before "see more")
↓
Blank line
↓
STORY/CONTENT (3–5 short paragraphs)
↓
LESSON/INSIGHT
↓
CALL-TO-ACTION
↓
#hashtags (3–5)

Character Limits

Section	Limit
Post	3,000 characters
Hook (before "see more")	First ~140 characters
Headline	220 characters
About section	2,600 characters

---

Scaling Roadmap

Phase	Focus
1	Content generation
2	Content repurposing
3	Engagement automation
4	Analytics feedback loop
5	Multi-agent orchestration

---

Results Benchmark

• 327 posts analyzed
• 14 workflows automated
• 30 days of content generated in minutes
• 1,000+ hours saved annually
• Repeatable LinkedIn growth system

The goal is not to use AI to write posts.
The goal is to build the machine that creates the posts.