People vs Algorithms Humans in the Loop
66 snips
Apr 17, 2026 They explain a vibe-coded AI system that turns a personal media diet into a daily briefing and living memory. They discuss multi-model prompt loops, human-in-the-loop editing, and building a wiki-like editorial brain. They debate clip-driven fame, meme warfare and how AI pivots reshape brands and publisher strategies.
AI Snips
Chapters
Transcript
Episode notes
Build An Editorial Memory To Win With AI
- Troy built an editorial ‘memory’ wiki that aggregates authenticated subscriptions, transcripts, and pinned stories to create a dynamic institutional brain.
- He uses multiple LLMs for extraction, writing, and fact-checking, plus human editing to preserve voice and accuracy.
PIMP Aggregates Troy’s Whole Media Diet
- Troy created PIMP to aggregate his entire media diet and authenticate subscriptions so the AI summarizes paid content legally.
- He used it daily to compress 400 stories into five trends and short blurbs for a briefing.
Put Humans In The Right Loop Positions
- Separate model roles and insert humans where judgement matters: use one model for extraction, another for fact-checking, and a human editor to choose and craft headlines.
- Troy made the system produce 20 candidate stories, then edited blurbs and pinned the final five for Substack publishing.