Vibe Coding: A Good Idea for Founders?

By Fadi Boulos, Founder & CEO of Supportful

Vibe coding is making waves in Silicon Valley. The term was coined by Andrej Karpathy, a founding engineer at OpenAI and former AI Director at Tesla.

Its premise is simple yet revolutionary: instead of writing code line by line, you “give in to the AI vibes”. you describe – in plain English – the features you want, and Large Language Models (LLMs) handle the rest. They generate the code and build the product for you.

Tools like Replit and Cursor are today’s flagship vibe coding kit. Replit requires no configuration or setup effort at all. Cursor is an Integrated Development Environment (IDE) forked from VS Code and developers’ favorite AI coding assistant.

This new approach to coding has divided the tech world. Some see it as a threat to sound software engineering practices due to concerns about the quality of AI-generated code, while others hail it as a game-changing innovation that accelerates product development.

The Promise of Vibe Coding

For startup founders, vibe coding feels like a dream come true. It’s a tool that allows them to build products without needing technical expertise or large engineering teams. This is particularly appealing for non-technical founders who want to bring their ideas to life quickly.

Y Combinator dedicated an entire podcast episode to vibe coding, discussing (but mostly praising) its potential to revolutionize how startups build and launch products.

Andrew Ng, a leading AI expert, has embraced this trend by creating a free "Vibe Coding 101" course on his DeepLearning.AI platform. The course teaches both technical and non-technical individuals how to use tools like Replit to create functional products.

I’ve taken this course myself and I can confirm that it’s straightforward and effective. I applied its principles to a small project I’m working on, and the process was surprisingly smooth.

The Concerns Around Vibe Coding

However, not everyone is sold on vibe coding. Many tech leads and software engineers are skeptical about the quality of code generated by AI.

LLMs are not actual coders; they generate code based on patterns in their training data, which may include suboptimal or outdated practices.

Another significant limitation is that LLMs lack the broader problem-solving skills of software engineers. While coding is an essential part of software engineering, it’s not the whole picture. Engineers translate business requirements into scalable, maintainable solutions – something LLMs currently struggle with.

Debugging is another weak spot for AI-generated code. LLMs often introduce new problems when tasked with fixing existing ones. Even during the Replit demo as part of the Vibe Coding 101 course, there was a 15-minute segment where the AI went off track while building the product.

What Founders Should Consider

Despite these concerns, startup founders – especially those without technical backgrounds – should still consider AI as a valuable technical co-founder. Vibe coding enables rapid prototyping and allows founders to deploy Minimum Viable Products (MVPs) with minimal technical knowledge.

Once the concept gains traction or attracts paying customers, founders can transition to building dedicated engineering teams. These teams can then collaborate with AI tools (again) to create scalable and reliable products.

While vibe coding may not be suited for long-term scalability or complex systems – at least today, it’s a great strategy for founders looking to get to market quickly.

Who knows though? By the time current startups grow, AI might evolve enough to build reliable products for mass adoption. Only AI can tell 🤷🏼‍♂️

Supportful is a software engineering consulting company that enables tech startups to scale their teams with remote engineers from Lebanon. Whether you want to build your MVP or grow your engineering team, Supportful provides solid software engineers who are good communicators. Supportful is on a mission to reduce the brain drain in Lebanon and help its small communities preserve their human capital.