Days after Google CEO Sundar Pichai praised the trend of ‘vibe coding’ as an emerging future that allows non-engineers and non-technical people to easily build software, Zoho’s Chief Scientist Sridhar Vembu took to social media to caution against oversimplifying the craft of software development, highlighting its deep complexity beneath the surface.
For context, Pichai, in the recent ‘Google for Developers’ podcast, stated that vibe coding has made the job of software development more fun and enjoyable. Pichai highlighted how crucial the AI-based coding tools are to reduce project development periods and make it a lot easier to focus on greater things.
Pichai hails vibe coding as accessible future
Vibe coding, a term coined by OpenAI co-founder Andrej Karpathy, describes the use of conversational AI assistants like ChatGPT 5 or Gemini 3 to generate, revise, and debug code using simple natural language prompts.
Speaking on a podcast, Sundar Pichai argued that vibe coding makes programming “more enjoyable” and “more approachable,” giving anyone with an idea the power to visualise and prototype apps or websites themselves. Vibe coding negates the need to master traditional languages like Python or JavaScript. He positioned Google’s latest model, Gemini 3, as their “best vibe-coding model ever,” signaling the company’s commitment to the trend.
As a result, Pichai is now ensuring that 25 per cent of the code written at Google is by AI. Microsoft’s CEO, Satya Nadella, stated that his firm advocates for a 30 per cent share of the code written by AI.
Vembu highlights coding’s fundamental complexity
While acknowledging the convenience of AI tools, Zoho founder Sridhar Vembu pushed back on the idea that coding could be reduced entirely to natural-language prompting. In a post on X (formerly Twitter), Vembu reminded the tech community of the intricate layers involved in software development, thus requiring proper skills to navigate the complex challenges of code development.
“All code is magic until it is lowered by the compiler to another form of code and that code is magic until magic all the way down,” Vembu wrote.
Vembu’s essentially states that behind every convenient AI interface, there are multiple, complex layers of code, compilers, optimisations, and abstractions, all of which remain fundamental to how computers function. While he did not directly criticise the AI tools, his commentary highlights the need for being good at coding in the software engineering ecosystem.
