← Back to portfolio ILLUSTRATIVE SAMPLE
AI MARKETING SYSTEM · ARCHITECTURE

The AI Content Machine

Architecture sample AI-assisted content operations for a lean B2B marketing team
System Overview
Pipeline Flow
Architecture
Skills & Automation
Roles & Next Steps

The Situation

Lisa's departure left a critical execution gap. Content quality, volume, and speed have been persistent issues. Planned roadmap items from Q4 2025 remain unbuilt. Meanwhile, buyer intelligence from 841 call transcripts shows we've been telling our story instead of the story buyers need to hear. The Content Machine addresses all of this simultaneously.

20
Claude Skills
~70%
AI-Assisted Tasks
7
Pipeline Stages
4
Scheduled Automations

What We Built

An AI-powered content system that flows from intelligence to published assets, with Claude handling the heavy production work and the team focused on strategy and quality assurance.

The Core Idea

Notion is the brain, the single source of truth for all content planning, assets, context, and workflows.

Claude is the engine, generating briefs, drafting content, running brand QA, repurposing assets, monitoring competitors, and producing performance reports.

The team is quality assurance, reviewing, approving, providing strategic direction, and making the final call on what ships.

This inverts the traditional model. Instead of a person doing 80% of the work and AI assisting with 20%, the machine does 70% and the team focuses on the 30% that requires human judgment.

Content Pipeline

Every piece of content flows through 7 stages with two human checkpoints. Everything between checkpoints is automated.

💡
Idea
AI seeds
HUMAN CHECK
Approved
Team vets
⚙️
In Production
AI produces
📄
Drafts Ready
Staged in Notion
HUMAN CHECK
🔍
In QA
Team reviews
✔️
QA Approved
Cleared
📢
Published
Team ships

Stage Details

Stage What Happens Who
💡 Idea Idea Engine scans market signals, threat intel, customer data, and competitive moves, seeds pipeline with prioritized ideas AI
✅ Approved Team reviews ideas in Notion, approves the ones worth pursuing Human
⚙️ In Production Campaign Manager plans the campaign → Content Executor produces all assets → Brand QA runs automatically AI
📄 Drafts Ready Anchor + derivative drafts staged as nested pages in Notion with QA results AI
🔍 In QA Team reviews drafts, provides editorial feedback Human
✔️ QA Approved Content cleared for publishing Human
📢 Published Team publishes to channels (blog, LinkedIn, email, etc.) Human

Three Layers

The system runs across three layers, each with a distinct role.

🗄️

Notion

Everything the team sees
  • Content Pipeline database
  • Pipeline Board view (Kanban)
  • Draft pages with QA results
  • Campaign grouping
  • Implementation Plan docs
💬

Cowork

Brian's control panel
  • "Run the idea engine"
  • "Execute approved campaigns"
  • "Run brand QA on drafts"
  • "Scan competitive landscape"
  • Any ad-hoc content requests

Scheduled Tasks

Runs automatically
  • Idea Engine: Mon/Wed/Fri 8 AM
  • Pipeline Watcher: 4x daily
  • Competitive Monitor: Weekly
  • Weekly Big Rocks: Monday 7 AM

Scheduled Automations

Task Schedule What It Does
Idea Engine Mon / Wed / Fri, 8 AM Scans market + intel → seeds Content Pipeline with ideas
Pipeline Watcher 4x daily, weekdays Monitors for Approved items → triggers Campaign Manager + Executor
Competitive Monitor Weekly (Monday) Scans competitor activity → updates Competitive Landscape database
Weekly Big Rocks Monday, 7 AM Generates Brian's 6 weekly priorities → posts to Marketing Team Meeting page

Skills Arsenal

20 specialized Claude skills organized in 5 categories, each trained on the company's brand voice, terminology, buyer personas, and competitive landscape.

Intelligence
📡 Pulse of the Market
🔎 Intel Gathering (Clickhouse)
👥 Customer Research Intel
🏆 Competitive Intel
Orchestration
💡 Idea Engine
📋 Campaign Manager
🔨 Content Executor
🔄 Content Repurposer
Writing
📚 Thought Leadership Writer
📱 Social Media Writer
🌐 Web Writer
🎯 Demand Gen Copywriter
Quality & Strategy
Brand Terminology QA
🔍 SEO / AEO / GEO Optimizer
📝 Content Brief Generator
🎨 Product Marketing
Knowledge Base
🏢 About the Company
🔐 FinSec Expert
📊 Content Performance Analyst
⚔️ Competitive Battlecard Updater

How Skills Chain Together

Idea Engine calls → Pulse of the Market + Intel Gathering + Customer Research Intel + Competitive Intel → synthesizes 3-7 ideas → seeds Content Pipeline

Campaign Manager reads approved idea → selects anchor asset type → maps 4-6 derivative pieces across channels → assigns Writer Skills

Content Executor reads campaign plan → routes each asset to its Writer Skill → runs Brand Terminology QA on every piece → stages all drafts in Notion under the anchor item

Human vs. AI Roles

The machine handles volume and consistency. The team handles judgment and direction.

🤖 Claude Handles

  • Market scanning & idea generation
  • Campaign planning & asset mapping
  • First-draft production (all formats)
  • Brand QA & terminology checks
  • Competitive monitoring
  • Content repurposing
  • SEO / AEO optimization

👤 Team Handles

  • Approving / rejecting ideas
  • Reviewing & editing drafts
  • Strategic direction & prioritization
  • Final publish to channels
  • Customer relationships & voice
  • Brand evolution & positioning calls
  • Stakeholder alignment

What's Live

Component Status
Content Pipeline (Notion database + Board view)✅ Live
Idea Engine (skill + scheduled task)✅ Live
Campaign Manager (skill + pipeline watcher)✅ Live
Content Executor (skill)✅ Live
All 4 Writer Skills✅ Live
Brand Terminology QA✅ Live
Competitive Monitor (scheduled)✅ Live
Weekly Big Rocks (scheduled)✅ Live
Clickhouse MCP (Gong, ChurnZero, Pendo)✅ Connected

What's Next

Enhancement Priority
Slack notifications when drafts hit "Drafts Ready"High, pending channel confirmation
Verify Clickhouse intel-gathering with real dataHigh
Hootsuite integration for automated social postingMedium, future phase
GA / Pardot data for Monthly Performance DigestMedium, future phase
Salesforce MCP for attribution trackingLow, future phase