Everyday Explainer
Self-improving AI pipeline for short-form YouTube explainer content. Built with Claude Code subagents for script generation, fact-checking, quality review, and analytics-driven optimization.
EverydayExplainer is a self-improving AI content pipeline for producing short-form YouTube explainer videos. The system generates scripts, fact-checks claims through web research, reviews output against anti-slop quality rules, and continuously improves future episodes using analytics-based learnings. It was built as a modular automation workflow orchestrated with Claude Code subagents and triggered through a homelab cron setup.
The Challenge
Producing consistent short-form explainer content requires repeated work across ideation, scripting, fact-checking, quality control, and performance review. The challenge was to turn these separate steps into a repeatable system that improves over time without requiring a growing amount of historical context for each new run.
The Solution
I built a modular agentic workflow powered by Claude Code. Specialized subagents handle analytics reflection, script generation, fact-checking through web research, and anti-slop review. Instead of passing full episode history into every run, the system stores only stable project rules and curated learnings, creating a scalable self-improving pipeline with controlled context growth.
Results
- 01Automated script generation and review workflow for short-form YouTube explainers.
- 02Built-in fact-checking with web research and correction steps.
- 03Self-improving feedback loop based on episode analytics.
- 04Modular subagent architecture with clearly separated responsibilities.
- 05Foundation for expansion into voice-over, rendering, subtitles, and publishing workflows.