The Slopfix development team has launched a flat-rate service for refactoring code generated by artificial intelligence tools. The startup, which itself uses AI agents, is asking for $10,000 for one week of work, but the final payment amount directly depends on the amount of redundant code successfully removed.
The Slopfix development team has launched a flat-rate service for refactoring code generated by artificial intelligence tools. The startup, which itself uses AI agents, is asking for $10,000 for one week of work, but the final payment amount directly depends on the amount of redundant code successfully removed.
According to Tom’s Hardware, the developers of Slopfix are focused on cleaning up cluttered repositories from what they call «AI slop.» The most ironic detail is that the team uses its own AI agents to combat the excessive amount of code, although they keep them under strict control of engineers.
Before starting work, three startup engineers analyze the client’s repository for free and agree on a target for reducing the number of rows. As a guideline, they cite optimizing the project from 100,000 to 35,000 rows with full preservation of working functionality.
Before making changes, the team creates a detailed screen-by-screen walkthrough of the application, which acts as a checklist for regression tests. Customers receive a clean codebase, linter rules, CI configurations, and a CLAUDE.md instruction file to prevent future code growth, as well as a two-week bug fix guarantee.
The project founder, known on the Hacker News platform under the nickname zie1ony, explained the pricing logic: «We get paid to remove code.» According to him, if the team promised to reduce the amount of code by half, but managed to do so by only 20%, the client will pay a proportional share — $4,000. At the same time, the engineers emphasize that although they use AI assistants to detect duplications, they keep them «on a very short leash,» not allowing the tools to make architectural decisions on their own.
The need for such services is confirmed by recent global studies. According to the GitClear «Maintainability Gap» report for 2026, the number of duplicate code blocks in the world increased by 81% compared to 2023. At the same time, the share of refactoring has critically decreased: if in 2022 it was 21% of all changed lines, now this figure has fallen below 4%.
Previously, the author of the term «vibecoding», co-founder of OpenAI and former director of AI at Tesla Andrey Karpaty predicted the onset of a «slopocalypse» (uncontrolled accumulation of AI slop) in 2026. The specialist, who himself switched to writing code using AI agents by 80%, noted that the ease of creating programs using LLM will inevitably lead to an oversaturation of the industry with low-quality and redundant code, which will later have to be cleaned up en masse.
