I Got Tired of AI Slop so I Built 20 Prompts to Fix It—They Save My Teams 100’s of Hours

Source: Nate’s Newsletter Date: 2025-10-10 URL: https://natesnewsletter.substack.com/p/i-built-a-20-prompt-set-to-kill-ai

Summary

Zero-cost AI content generation has flooded organizational workflows with low-quality output that requires expensive human review, creating a paradox where AI productivity gains are eaten by AI quality management. Nate’s solution is AI-as-reviewer: 20 specialized prompts targeting different content categories act as quality gates, routing scarce human attention to only the work that genuinely requires it.

Implications

Agent-product positioning thread. The “AI reviews AI” pattern is an early form of multi-agent workflow — and the fact that it’s being packaged as a 20-prompt kit rather than a product suggests the practitioner market is building these pipelines manually. That’s a product gap.

AI economics thread. Karpathy’s framing — “99.9% of attention is about to be LLM attention” — reframes human attention as the scarce resource in an AI-abundant world. If that framing holds, products that allocate human attention efficiently (rather than just automating tasks) have a durable value proposition.

Watch: Whether “AI slop” becomes a recognized enterprise budget line (quality assurance tooling for AI-generated content), and which vendors build dedicated products around the review-gate pattern rather than prompt kits.