“Natasha” Turned Out to Be Human: How 700 Indians Wrote Code Under the Guise of AI
When the startup BuilderAI launched “AI Natasha,” which promised to create an app based on your request in just a couple of minutes, the world sighed in relief — yet another revolution in development. In reality, the “neural network” was simply sending tasks to an office in India, where more than 700 programmers manually wrote the code, pretending it was the marvel of machine intelligence. Over eight years, “Natasha” raised $445 million in investments, including from Microsoft, and went bankrupt only in 2025. Human intelligence instead of algorithms.
The illusion of AI at work was perfectly maintained: the code was just as opaque as the outputs of a machine learning black box. The programs often didn’t work — perfectly mimicking the bugs of a “trained model.”
Signs of “Natasha’s” code:
– Unreadable variables and missing comments
– Unstable builds with random errors
– Missing key features or their lackluster implementation
Xrust.ru believes this story isn’t just BuilderAI’s failure but a lesson for the industry: without a dose of truth, even the most expensive startups will collapse if the “AI” label hides nothing more than a group of overworked people.
https://xrust.ru/ https://xrust.ru/news/311230-natasha-okazalas-chelovekom-kak-700-indusov-pisali-kod-pod-vyveskoj-ii.html
https://xrust.ru/news/311230-natasha-okazalas-chelovekom-kak-700-indusov-pisali-kod-pod-vyveskoj-ii.html
