Software Engineering stands at a pivotal moment as AI rapidly evolves from assistant to primary code author. While there is anticipation of AI soon writing the majority of new code, developers are finding themselves moving from author to architects, reviewers, and orchestrators of AI-generated systems.
As vibe coding moves from experimentation to standard practice, our inaugural State of Vibe Coding Report surveyed 1,000 software engineers using AI to uncover who's adopting these tools, how they're being used, and the benefits and challenges developers are experiencing. The findings provide a data-driven snapshot of our industry in transition, highlighting both the tremendous potential and the hurdles we must overcome to fully realize the promise of AI-driven development.
Enterprise momentum is driving early vibe coding adoption
Most respondents work at companies with more than 1,000 employees, suggesting that large organizations are leading the way in experimenting with and adopting vibe coding tools.
Software engineers are leading the vibe coding movement
The majority of respondents identified as software engineers or developers, reinforcing the idea that vibe coding resonates most with those closest to the code and implementation.
Personal projects help drive vibe coding
While vibe coding originated in personal projects, it has rapidly gained traction in professional environments, with half of all usage now occurring in work-related projects. This shift demonstrates how developers are increasingly bringing these tools into enterprise settings after initial experimentation.
Daily use is common among vibe coders
Among users of vibe coding tools, a significant portion report using them several times a week or daily, indicating that LLM-assisted development is becoming a regular part of the workflow for many.
Data and analytics drives vibe coding adoption
Most users report using vibe coding tools for data processing and analytics workflows—contexts where speed and creativity are more important than production polish.
Vibe coding is a habit for many developers
Nearly half of respondents use vibe coding tools regularly, with only 7% having never tried them—suggesting strong early adoption and interest across the developer landscape.
These tools shape the vibe coding stack
GitHub Copilot, ChatGPT, and Claude Code are the most widely used tools, but a long tail of emerging platforms like Cursor, Firebase Studio, and Vercel v0 show a growing ecosystem of specialized solutions.
Javascript and Python dominate the vibe coding landscape
JavaScript/TypeScript and Python are the top languages for vibe coding, reflecting the tools’ strong alignment with web development, scripting, and rapid prototyping.
Most vibe-coded output still needs work before shipping
While some users report production-ready output, most say generated code requires moderate cleanup or serves as a rough draft—highlighting current limitations in quality and reliability.
Developers want smart tools, not just faster ones
The top areas for improvement are understanding developer intent, improving code quality, and better debugging support—suggesting a shift in priorities from speed to trust and refinement.
Developers are optimistic about the future of vibe coding
Most respondents believe vibe coding will go mainstream within three years, reflecting growing trust in the tools and their long-term relevance to everyday software development.
About
This report explores how developers are adopting vibe coding, a style of working with LLM-powered tools to generate and modify code through natural language. It captures how the practice is evolving across roles, project types, and tool stacks, with insights into what developers want next.
Methodology
We conducted an online survey in May, 2025, collecting responses from hundreds of participants across a range of roles, company sizes, and experience levels. The survey included a mix of multiple-choice and open-text questions focused on behavior, perception, and future expectations related to vibe coding.
Attribution
This report was created by the Codura team. Data was collected and analyzed independently. For questions, feedback, or reuse inquiries, contact us at vibes@codura.ai.