HN Showcase
Back to browse
I built an AI system where content compounds like interest

I built an AI system where content compounds like interest

by bittensorman·Feb 28, 2026·1 point·0 comments
Visit ProjectView on HN

AI Analysis

●MidShip ItBig Brain

Self-improving content loop sounds cool, but the actual loop isn't closed yet.

Strengths
  • •Closes a real feedback loop: search traffic back into knowledge expansion via AIMediaPost.
  • •Demonstrates three working demos (CMS, media post, radio) with visible output.
  • •Integrates local LLMs (Ollama) for on-device content generation without API costs.
Weaknesses
  • •Early stage with unvetted knowledge quality—author admits they're still in 'foundation phase' and optimizing later.
  • •Competes with established knowledge graphs (Obsidian, Roam) and AI content tools without a clear differentiation story.
Category
AI/ML
Target Audience

Content creators, knowledge curators, AI researchers interested in content distribution

Similar To

Roam Research · Obsidian · Perplexity

Post Description

I've been building AIGM (AI Generated Media) – a self-evolving media architecture where AI doesn't just create content, it learns from how that content performs and gets smarter over time.

The core loop: 1. AIKnowledgeCMS collects & structures knowledge into keyword nodes 2. AIMediaPost turns those nodes into articles, publishes to Blogger, X, etc. 3. Search traffic flows back into AIKnowledgeCMS 4. The system expands its own knowledge base automatically

Every post makes the next one better.

Live demos: - AIKnowledgeCMS: https://aiknowledgecms.exbridge.jp/ - AIMediaPost: https://aimediapost.exbridge.jp/ - AIRadio: https://airadio.exbridge.jp/

Happy to answer any technical questions.

Similar Projects