← Back to blog
building in-public4 min read

Building the Skillpress Content Engine: From Chaos to System

By Skillpress·March 24, 2026

How we built an automated content research and production system that turns AI releases and industry signals into actionable content for finance, HR, and operations professionals.

The Problem: Content Chaos

Like most small business owners, I was drowning in content opportunities. Every day brings:

  • New AI tool releases (Claude 3.5, ChatGPT updates, Gemini upgrades)
  • Industry discussions on LinkedIn and Twitter
  • Customer questions in our support channels
  • Research findings from my day job in finance

But turning these signals into useful content for our customers—finance professionals, HR coordinators, operations managers—felt impossible. I'd bookmark dozens of articles, take notes in random docs, then never action any of it.

Enter: The Content Engine

Last month, I built what we call the "Content Engine"—a system that automatically:

  1. Captures signals from RSS feeds, Readwise saves, and industry monitoring
  2. Scores relevance for our specific customer personas (finance, HR, ops, small biz)
  3. Drafts content across multiple formats (X threads, blog posts, Instagram carousels)
  4. Stages everything in our editorial calendar for review and approval

The result? We went from publishing 2-3 random posts per week to having a planned content batch ready every Tuesday morning.

The Architecture

Signal Capture (Layer 1)

Three automated inputs feed our signal vault:

  • Release Monitor: Checks Anthropic, OpenAI, and Google AI RSS feeds daily at 7am
  • Research Cron: Runs topic-based searches weekly (rotating between "AI for finance teams", "AI for HR", etc.)
  • Readwise Filter: Scans my personal saves for anything matching our ICP keywords

Each signal gets scored across 6 dimensions: finance relevance, HR relevance, operations relevance, sales relevance, admin relevance, and small business relevance.

Content Production (Layer 2)

Every Tuesday at 8am, a production batch cron:

  1. Reviews all new signals from the past week
  2. Picks the top 3-5 highest-scoring items
  3. Maps them to our content lanes:
    • Personal account (70%): Practitioner insights, lessons learned
    • Product marketing (20%): How Skillpress helps with these challenges
    • Building in public (10%): System updates, behind-the-scenes

Distribution (Layer 3)

Content gets staged in platform-specific folders:

  • X threads (280 char limit, hook + 3-4 supporting tweets)
  • Instagram carousels (6-slide scripts with visual cues)
  • Blog posts (800-1200 words, conversion-optimized)
  • Reddit posts (community-specific, value-first)

The Results

Before: Sporadic posting, always scrambling for ideas, content felt disconnected from customer needs

After:

  • 15+ pieces of content queued every week
  • Each piece mapped to a specific customer pain point
  • Zero stress about "what should I post today?"
  • Much higher engagement (people actually respond with "this helped!")

What I Learned

1. Systems > Inspiration

The breakthrough wasn't having better ideas—it was building consistent capture. When good signals automatically flow into your system, content creation becomes assembly instead of alchemy.

2. Customer-First Filtering

We added a hard rule: every piece of content must answer "what does this mean for a finance manager?" or "how does this help an HR coordinator?" Generic AI hype gets filtered out.

3. Batch Everything

Publishing daily from a queue feels completely different than creating daily from scratch. The production pressure disappears.

The Tools

For anyone wanting to build something similar:

  • Signal capture: RSS feeds + Readwise API + Obsidian for storage
  • Content scoring: Local LLMs (Mistral) for cost-effective classification
  • Editorial management: Obsidian folders with status-based organization
  • Distribution: Self-hosted Postiz for social scheduling + Plausible for analytics

The whole system runs on cron jobs, costs ~$30/month, and saves me 15+ hours per week.

What's Next

We're adding:

  • Competitor monitoring (track what other AI productivity tools are talking about)
  • Customer conversation mining (turn support tickets into content ideas)
  • Performance feedback loops (boost content types that drive actual skill sales)

Building in public means sharing what works—and what doesn't. This system transformed our content from reactive to strategic. If you're drowning in content opportunities, maybe it'll help you too.

Want the behind-the-scenes technical details? I'm documenting the full build at chris's personal Twitter. Or grab our Finance AI Coordinator skill to see the kind of practical AI content this system helps us create.

Get more like this

Practitioner-tested AI workflows delivered weekly.