More and more voices are speaking up about the extraordinary gains they’re seeing from AI in writing code.
The latest comes from Bipin Preet Singh, co-founder and CEO of MobiKwik, one of India’s largest digital financial platforms. In a post on X addressed directly to Claude’s official account, Singh described a moment that would have been unthinkable just a few years ago: “Claude built an entire fraud detection and prevention system for us with 3000 lines of code and 40 files in less than 30 minutes. On a weekend morning.” He ended the post with three words that capture the anxiety now rippling through the developer community: “Dev is cooked!”, and added a picture for good measure.
MobiKwik, founded in 2009 by Singh and co-founder Upasana Taku, is a Gurugram-based fintech company that began as a mobile wallet and has since grown into a full-stack financial services platform serving over 160 million registered users and more than 4 million merchants across India. The company offers a wide range of products spanning consumer payments, digital credit, buy-now-pay-later services, mutual funds, insurance, and UPI-based transactions. Having gone public in December 2024 in a successful IPO, MobiKwik is one of the most prominent names in Indian fintech — and a company for which fraud detection isn’t a side feature but a core operational necessity given the volume of financial transactions it processes daily.
That makes Singh’s post all the more striking. A production-grade fraud detection system — the kind that would traditionally require a team of engineers, weeks of planning, code reviews, and testing cycles — was reportedly assembled by an AI in the time it takes to watch a movie. Whether or not every line is production-ready, the claim points to something the industry is increasingly grappling with: AI coding tools are now capable of taking on substantial, domain-specific engineering tasks with minimal human input.
Singh is far from alone in making such observations. Spotify co-CEO Gustav Söderström revealed during the company’s most recent earnings call that some of Spotify’s best engineers have not written a single line of code manually since December, with AI tools handling development while engineers manage and review output. Söderström said the company shipped over 50 new features in 2025 using these AI-driven workflows. Microsoft and Google stated in April 2025 that AI was generating roughly 30% of the code across Microsoft, a figure that has likely climbed since. Salesforce CEO Marc Benioff cited a similar proportion for his company. And Anthropic’s own Boris Cherny, who leads Claude Code development, noted that he personally hasn’t written any code in over two months, with 22 pull requests shipped in a single day — every one of them AI-generated.
The broader pattern is hard to ignore. What Singh’s post adds to this conversation is a perspective from a fast-growing emerging market fintech, where engineering resources are precious and the stakes around financial security are high. If AI can spin up a fraud detection system in 30 minutes on a weekend, the calculus around developer hiring, system architecture, and product velocity is fundamentally changing — not just in Silicon Valley, but across the global tech industry.
Whether “Dev is cooked” turns out to be hyperbole or prophecy, one thing is increasingly clear: the bar for what a single person with AI can build in an afternoon keeps moving.