Until generative artificial intelligence (AI), Zerodha’s new feature launches were a quiet marathon.  Senior developers measured each line, made it exact, and frequently rewrote it until everything seemed correct.

Kailash Nadh, the company’s chief technology officer, recalls the days when lengthy hours were spent developing code.  The process was slow yet deliberate.  Releases took time, which was wonderful since so did perfection.

Then AI showed up.

Nadh explained that complex engineering challenges that used to take days of trial and error are now resolved in hours.  Tasks that would normally take several hours now take only a few minutes.  Fixing esoteric technical issues has become much faster.

Another firm founder described it as a “opportunity for a massive productivity boost.”

There is one caveat: These benefits are best realized by experienced senior engineers “who know what they are doing.”

Zerodha isn’t alone.  AI agents are creating huge amounts of code, up to 40-80%, for Indian firms using technologies like ChatGPT, Gemini, and Claude, creators told.  From early prototypes to production-ready systems, these AI bots generate precise, error-free code, saving teams time and increasing productivity.

This mirrors a larger worldwide trend.

Related: How Technology is Changing IPL and Playing Major Role? List of Tech Involved in IPL 2025

Tech experts such as OpenAI’s Sam Altman, Salesforce CEO Marc Benioff, and Anthropic CEO Dario Amodei have projected that AI will generate the majority of the world’s code.  In March, the CEO of Y Combinator stated that over 25% of the accelerator’s existing businesses were already employing AI to write 95% or more of their code.

Nadh said that instead of employing AI tools to generate broad swaths of code, Zerodha uses platforms such as Aider, GitHub Copilot, DeepSeek, Claude, and Gemini for more specific tasks such as auto-completing boilerplate code, sanity-checking, and making code more concise.

Saurabh Karn, a founder member of Sarvam AI, recognized the trend, stating that some of the AI startup’s projects generate 80-90% of their code using various agents.  “AI may be coming into the picture, and changing the role of a coder, but the responsibility remains the same: build highly-compliant, controlled code and models,” he told me.

This change is visible across industries. Sudip Ghose, creator of luggage company Uppercase, claimed AI currently accounts for 50-55% of its software, up from 5-10% a year ago.

Inmobi founder and CEO Naveen Tewari took to social networking site X to warn computer professionals to upskill, as the business plans to write almost 80% of its code using AI technologies by the end of this year.

Reltio, a data management startup that just launched an R&D centre in Bengaluru, uses GenAI to generate around 15% of its code, according to CEO Manish Sood.  According to SA, AI is responsible for writing a third of data security unicorn Druva’s code.

This dramatically improves productivity.  According to Ghose, since this change, initial coding tasks have taken about 70% less time than previously.  Sood agreed, stating that the use of AI had increased productivity in several areas by approximately 10%.

Despite the obvious benefits, AI coding has posed some problems.  At e-visa company Atlys, where 25-30% of the codebase is either directly generated or greatly aided by AI agents, an AI agent once sought to simplify a script for cleaning up old logs. This resulted in a significant error: because of a missing configuration, the script wiped a complete working directory instead of only the intended log files, wiping out critical system data and causing a service to fail continuously, remarked the founder, Mohak Nahta.

To be on the safe side, online self-publishing and audiobook portal Pratilipi does not utilize AI to write code, but rather for basic structure chores.  “We avoid taking any existential risks.  The standard is high; code that goes into production must be error-free.  So, we utilize AI to structure and review code, but not to write it,” CEO Ranjeet Pratap Singh.

AI’s ability to generate boilerplate code and automate repetitive activities raises the entry-level bar for younger coders.  “Entry-level and early-career software engineers will have to accept and acknowledge AI as a significant and capable competitor on the floor,” stated Prasadh MS, head of workforce research at hiring firm Xpheno.

“AI’s ability to function on a scale and pace that humans cannot match is a clear advantage that businesses will pay for.  It’s normal for businesses to expect young engineers to arrive fully trained and ready to collaborate with AI agents as digital colleagues,” he said.

Neeti Sharma, CEO of TeamLease Digital, believes companies will eventually stop looking for software developers in large numbers.  “Companies will not hire coders, but they will look for someone who knows how to read the code and evaluate its efficacy,” she told me.

Each year, over 1.5 million STEM (science, technology, engineering, and mathematics) graduates enter the ecosystem.

“As witnessed over the last few quarters, the net number of employment in data security, AI, and machine learning has increased rapidly.  While coding jobs may be slowing, new opportunities are emerging. “Founders will now judge a candidate based on their prompt library rather than their code,” Sharma explained.