AI is no longer just helping developers. It’s actually writing big chunks of the code for companies now.
Across global and Indian tech companies, AI has quietly moved from an experimental tool to a part of everyday workflows, including coding, testing, and deployment. This is no longer just a behind-the-scenes change; it is now being reflected in what company leaders are openly reporting.
Global companies
Google: Last month, CEO Sundar Pichai said that nearly 75% of all new code at Google is now AI-generated and then approved by engineers, up from 50% the previous fall, according to ET. He also said the figure had jumped from just 15% in 2023.
Microsoft: Last year, at Meta's LlamaCon conference, Microsoft CEO Satya Nadella said that AI was already writing about 20% to 30% of some code at the company. Meanwhile, CTO Kevin Scott predicted that AI will generate 95% of all code within the next five years.
“Very little is going to be — line by line — human-written code,” he said. But he made it clear that this does not mean human programmers will be sidelined. While AI will take over routine coding tasks, people will still be crucial for the creative and problem-solving parts of software development.
Meta: In March, Business Insider reported that Meta set internal targets varying by division. Its creation org, which builds products like Messenger, WhatsApp, and Facebook, aims for 65% of engineers to write over 75% of their code using AI tools by early 2026. The Scalable Machine Learning team has a 2026 target of 50%–80% AI-assisted code.
“I think sometime in the next 12 to 18 months, we will reach he point where most of these codes that are going towards these efforts will be written by AI. And I don’t mean like autocomplete,” Meta CEO Mark Zuckerberg said in a podcast interview with Dwarkesh Patel last year.
Anthropic: In February this year, Anthropic’s chief product officer Mike Krieger said the company’s own AI, Claude, has become the primary author of its development.
He added that engineers now ship AI-generated pull requests of 2,000 to 3,000 lines. “Dario predicted 90%... and today it's effectively 100%,” he said, referring to CEO Dario Amodei’s earlier view that 90% of code could eventually be written by AI.
Indian companies
Meesho: In May, CEO Vidit Aatrey wrote in a shareholder letter that over 70% of Meesho’s code is now AI-generated. The same report said Meesho is using AI not just for writing code, but also for testing, code reviews, production monitoring, and deployment fixes.
The report also said Meesho had not yet seen meaningful savings in people costs, which is important because it suggests the current gain is more about productivity than immediate headcount reduction.
Aatrey added, “75%+ of orders on Meesho come not from users searching, but from personalised feeds powered by PRISM (Personalised Ranking and Intent Signal Module) - our AI recommendation engine that infers what a consumer is looking for before they've articulated it themselves.”
Freshworks: When Freshworks announced layoffs of about 11% of its workforce, CEO Dennis Woodside said that more than 50% of the company’s code is now generated by AI.
Speaking to Reuters, he added that the job cuts were partly driven by increased AI use in product and engineering teams, along with automation of routine work across the business.
Credgenics: In April, cofounder Rishabh Goel told ET that AI can get 70 to 80% of the backend work done today. The comment came in the context of AI-led credit and collections workflows, showing that AI coding is spreading beyond consumer internet firms into fintech.
Paisabazaar: Chief technology officer Mukesh Sharma told ET the company was using AI to build a deeper understanding of borrowers’ credit histories, create fully AI-generated videos to explain the challenges to customers, and offer an AI-led chat service to assist them.
Banksathi: Cofounder Dhaka said currently 80% of the work is being done by AI, while the remaining 20% will be digitised only when banks create standard processes around resolutions.
What this means
These numbers don’t mean AI is replacing engineers. It mainly means AI is helping with first drafts, simple code, tests, and routine fixes, while humans still review and approve everything.
Earlier, the challenge was writing enough code fast enough; now, it is about making sure AI-generated code is correct, secure, and aligned with product goals.
The bigger change is that software teams are becoming smaller and faster. Developers are shifting from writing every line of code to checking AI-generated code, fixing issues, and focusing more on design, complex problems, and improving quality.
Challenges
The job-cut angle is now central. When companies say AI is writing half or more of the code, it often goes along with smaller teams, slower hiring, and sometimes layoffs, especially in roles that involve repetitive or backend work.
At the same time, while AI can improve speed and productivity, it also brings new concerns. More AI-generated code can help companies ship faster, but it raises questions about bugs, security, responsibility, and how much engineering teams are changing behind the scenes.
Across global and Indian tech companies, AI has quietly moved from an experimental tool to a part of everyday workflows, including coding, testing, and deployment. This is no longer just a behind-the-scenes change; it is now being reflected in what company leaders are openly reporting.
Global companies
Google: Last month, CEO Sundar Pichai said that nearly 75% of all new code at Google is now AI-generated and then approved by engineers, up from 50% the previous fall, according to ET. He also said the figure had jumped from just 15% in 2023.
Microsoft: Last year, at Meta's LlamaCon conference, Microsoft CEO Satya Nadella said that AI was already writing about 20% to 30% of some code at the company. Meanwhile, CTO Kevin Scott predicted that AI will generate 95% of all code within the next five years.
“Very little is going to be — line by line — human-written code,” he said. But he made it clear that this does not mean human programmers will be sidelined. While AI will take over routine coding tasks, people will still be crucial for the creative and problem-solving parts of software development.
Meta: In March, Business Insider reported that Meta set internal targets varying by division. Its creation org, which builds products like Messenger, WhatsApp, and Facebook, aims for 65% of engineers to write over 75% of their code using AI tools by early 2026. The Scalable Machine Learning team has a 2026 target of 50%–80% AI-assisted code.
“I think sometime in the next 12 to 18 months, we will reach he point where most of these codes that are going towards these efforts will be written by AI. And I don’t mean like autocomplete,” Meta CEO Mark Zuckerberg said in a podcast interview with Dwarkesh Patel last year.
Anthropic: In February this year, Anthropic’s chief product officer Mike Krieger said the company’s own AI, Claude, has become the primary author of its development.
He added that engineers now ship AI-generated pull requests of 2,000 to 3,000 lines. “Dario predicted 90%... and today it's effectively 100%,” he said, referring to CEO Dario Amodei’s earlier view that 90% of code could eventually be written by AI.
Indian companies
Meesho: In May, CEO Vidit Aatrey wrote in a shareholder letter that over 70% of Meesho’s code is now AI-generated. The same report said Meesho is using AI not just for writing code, but also for testing, code reviews, production monitoring, and deployment fixes.
The report also said Meesho had not yet seen meaningful savings in people costs, which is important because it suggests the current gain is more about productivity than immediate headcount reduction.
Aatrey added, “75%+ of orders on Meesho come not from users searching, but from personalised feeds powered by PRISM (Personalised Ranking and Intent Signal Module) - our AI recommendation engine that infers what a consumer is looking for before they've articulated it themselves.”
Freshworks: When Freshworks announced layoffs of about 11% of its workforce, CEO Dennis Woodside said that more than 50% of the company’s code is now generated by AI.
Speaking to Reuters, he added that the job cuts were partly driven by increased AI use in product and engineering teams, along with automation of routine work across the business.
Credgenics: In April, cofounder Rishabh Goel told ET that AI can get 70 to 80% of the backend work done today. The comment came in the context of AI-led credit and collections workflows, showing that AI coding is spreading beyond consumer internet firms into fintech.
Paisabazaar: Chief technology officer Mukesh Sharma told ET the company was using AI to build a deeper understanding of borrowers’ credit histories, create fully AI-generated videos to explain the challenges to customers, and offer an AI-led chat service to assist them.
Banksathi: Cofounder Dhaka said currently 80% of the work is being done by AI, while the remaining 20% will be digitised only when banks create standard processes around resolutions.
What this means
These numbers don’t mean AI is replacing engineers. It mainly means AI is helping with first drafts, simple code, tests, and routine fixes, while humans still review and approve everything.
Earlier, the challenge was writing enough code fast enough; now, it is about making sure AI-generated code is correct, secure, and aligned with product goals.
The bigger change is that software teams are becoming smaller and faster. Developers are shifting from writing every line of code to checking AI-generated code, fixing issues, and focusing more on design, complex problems, and improving quality.
Challenges
The job-cut angle is now central. When companies say AI is writing half or more of the code, it often goes along with smaller teams, slower hiring, and sometimes layoffs, especially in roles that involve repetitive or backend work.
At the same time, while AI can improve speed and productivity, it also brings new concerns. More AI-generated code can help companies ship faster, but it raises questions about bugs, security, responsibility, and how much engineering teams are changing behind the scenes.
