AI writing quality tool. Keep content human and direct.
Internal tool that scans text for common AI writing patterns and either flags or auto-removes them. Built because AI-generated content has predictable tells: em dashes, hollow superlatives, corporate jargon, and filler openers. Unslop catches them before they go out.
Tool
Type
Developer tool
Writing QA
Use case
Content quality control
Active
Status
In daily use
Context
AI models have consistent writing fingerprints. They overuse certain phrases, structures, and punctuation in ways that trained readers recognize immediately. The em dash is the most obvious. Corporate jargon is another. Filler openers like "Certainly!" and "Absolutely!" are a dead giveaway.
MGT uses AI heavily for content generation, code documentation, and client-facing copy. Without a quality pass, the output sounds like every other AI-written piece on the internet. That is not the voice MGT wants in front of clients or on social.
Unslop was built to automate the quality pass. It runs fast from the command line, flags every pattern, and where safe to do so, removes them automatically. The result reads like a person wrote it.
What We Built
Core engine that runs text through a library of AI writing patterns. Outputs a score and a list of flagged phrases with line numbers and suggested fixes.
Safe patterns get removed automatically. Em dashes, filler openers, and hollow superlatives are stripped without needing a manual review step.
Patterns that need human judgment are flagged for review rather than auto-removed. Passive voice clusters and bullet overuse fall in this category.
Runs as a command-line tool. Pipe in a file, get flagged output back. Fits into any existing writing or deployment workflow without extra setup.
Patterns detected
Tech Stack
Quality gates, AI rewrite pipelines, structured output enforcement: these are solved problems. Book a 15-minute call and tell me where your content breaks down.