Vibe Coding: Now, Coding is More Accessible

Vibe Coding: Now, Coding is More Accessible

VIBE coding, the term which was coined in February this year, has now emerged as a trend in the software development space. Rather than manually writing code, developers rely on Large Language Models (LLMs) to generate code. Ganesh Sahai, CTO, Nagarro said there is significant investment going into the development of Al-assisted tooling to cater to all stages of the Software Development Lifecycle (SDLC), such as requirement gathering, design, coding, testing, delivery, production, and maintenance.

“The traditional ways of working in these stages are being augmented with AI tools. These tools enable more conversational ways of achieving goals,” he said.

For example, a coder or tester may now use a tool like Cursor to speak and give prompt-based instructions, which are converted to text and can be further edited to specify the functionality to be implemented or tested. The code thus generated—previously written manually—can be improved further through refined prompts or direct edits. This approach reduces grunt work, makes software development more enjoyable, and adds a cool vibe.

This new AI-assisted method is sometimes referred to as vibe coding, he added. Vibe coding helps developers focus on prompting, reviewing and iterating the code, essentially letting the process be driven by “vibes.” This involves embracing more intuitive, high-level conceptualization and iterative refinement guided by prompts, explained Praveen Ojha, Chief Technologist, EPAM Systems, India.

How useful is it for clients?

This would mean that the time lag from feature idea to production is going to reduce significantly. Hence, the pace at which clients can experiment and create business impact is going to move faster. “As we move into the future, developers will need to be more skilled at participating in conversations that help shape features. This will require a deeper understanding of business requirements and, as a result, the development of a product manager—like mindset. However, it is also important that the auto-generated code is well understood by the developer and, if needed, properly debugged. Therefore, strong coding skills will still be essential,” Sahai added.

The roles of product manager and developer may begin to merge. “In fact, the essence of various conversations among team members may increasingly be fed directly into AI tools to develop features and push them to production. This will unleash cross-functional collaboration to the next level, as has long been envisioned in an ideal agile team. In a way, these AI tools are evolving into fellow developers—engaging in meaningful conversations and taking over the grunt work,” he further added.

Praveen Ojha said from an efficiency standpoint, vibe coding can result in faster turnaround cycles for developers, reduced manual work and increased focus on strategic thinking. However, the idea of letting an AI write production grade code can stir both fascination and doubt. According to him, some see the promise of near-instant productivity, while others worry about unleashing legions of unmaintainable scripts into our codebases.

“Being the frontliners of modern engineering practices, we see this trend as an opportunity using a lens mixed with curiosity and caution. While traditional copilots can provide effective code suggestions and auto-completions, vibe coding focuses on a more collaborative and context-aware approach. It involves the use of AI tools through natural language prompts to modify and manage code, introducing a symbiotic relationship between developers and AI,” he explained.

Learn more about how EPAM enables the future of business with modern engineering: https://www.epam.com/services/engineering