Vibe Coding Boom Fuels Demand for AI Code Cleanup

In February 2025, OpenAI co-founder Andrej Karpathy coined the term “vibe coding” to describe using natural-language instructions to rapidly generate code with AI. While the approach can speed up product launches, it can also create technical debt in the form of duplicated logic, broken features and hard-to-maintain code. That has created new demand for code governance and refactoring services.

On July 8, 2026, Slopfix, a three-person Polish team, attracted attention for its service. Its three senior engineers charge a benchmark rate of $10,000 for one week of work. They first analyze a codebase for free and then commit to a code-reduction target. A case study on the company’s website shows a codebase cut from 100,000 lines to 35,000. Fees are based on the percentage of the target achieved, with proportional charges if the target is missed, and the service includes a two-week warranty.

Vibe coding is a development approach in which AI rapidly generates software from natural-language instructions. Former OpenAI researcher Andrej Karpathy coined the term in February 2025. The approach lowers the barrier to building apps, but it can also lead people without engineering training to overlook critical issues such as access controls, data protection and system architecture. Responsibility for cybersecurity therefore cannot be handed over to AI.

An AI-assisted app called “惜食獵人” recently came under scrutiny after vulnerabilities, including the exposure of GPS coordinates, were disclosed, prompting online debate over the security of vibe coding. The incident shows that while AI can shorten development and product-launch timelines, developers must still conduct code reviews, security testing and risk assessments themselves. Those most at risk of being replaced are people unable to distinguish good code from bad.