Experimenting with vibe coding? Nithin Kamath says ‘If I traded and had Claude…’
Nithin Kamath, co-founder and CEO of Zerodha, stated that there has been a rise in the number of users engaging in jitter coding of financial instruments using Kite APIs.
In a March 10 post, the tech leader posted a screenshot of a Reddit user’s post about creating a “panic button” to cash out his portfolio during a market crash, noting: “If I were trading today and I had Code Claude, I’d probably vibrate code that.” He also provided a link to the Kite API page (https://www.kite.trade/).
In the original Reddit post, user Many-Virus4840 said he “created a little ‘panic selling’ script that quickly dumped my entire portfolio if the market started to crash.”
“The goal was to find a way to react quickly during market dips without having to open apps or take multiple steps. It’s a simple tool I built for fun while experimenting with automation,” the user explained.
Kamath notes the trend of individuals using vibrational coding
In a post on LinkedIn last week, Kamath noted a similar trend. “My print feed is full of people coding all kinds of financial tools using Kite APIs, such as portfolio dashboards, backtesting setups, market insight tools, and even automated trading setups. Most of these are useless, but there are some really well-thought-out tools designed to solve very specific problems that traders and investors have,” he said.
He added that this is an influx of people who have “never written a line of code in their lives” but are using “plain English and AI coding tools like Claude Code, Codex, etc.”
He also encouraged others to try it: “If you have FOMO to try these tools but don’t know where to start, try our MCP plugin first. Connect it to your Zerodha account and start asking things to it. You’ll get a real feel for what’s possible before you start using AI coding tools.”
What is jitter coding?
Vibe coding is a software development practice in which users describe a task or project to an artificial intelligence (AI) chatbot, which then generates a source code. Implementation typically involves accepting automated AI-generated code that is refined through follow-up prompts.
Some AI coding assistants automatically complete code written by human programmers, similar to “autocorrect” features that suggest next lines of an email or text. More advanced tools, known as AI agents, are being given more autonomy to access computer systems and do the work themselves.
The big language models behind generative AI chatbots like Anthropic’s Claude, OpenAI’s ChatGPT and Google’s Gemini can do many things from homework help to organizing meal plans, but the “best use case” for most businesses is coding and software engineering, Gartner analyst Philip Walsh told the AP last September.
In particular, experts agree that while jitter coding tools are great for prototyping, they fall short of enterprise-level adoption. Integration into complex architectures, compatibility requirements, and sustainability concerns will likely limit large-scale production deployments in the near term.
But the market is still large. Nasscom has predicted that the low-code/no-code (LCNC) market in India will grow tenfold, i.e. from $400 million in FY21 to $4 billion in FY25.
