Theranos for Code: Inside Builder.ai’s $1.5 Billion AI-Washing Scam

When Microsoft, Qatar’s sovereign wealth fund, and SoftBank poured hundreds of millions of dollars into Builder.ai, the London-based start-up looked like a Cinderella story for the generative-software boom. Its pitch was irresistible: an AI assistant named “Natasha” could snap together mobile apps “like Lego bricks” in days, slashing development costs and putting coding within reach of anyone with an idea. Investors duly valued the company at $1.5 billion. But five years after Wall Street Journal reporters first noticed something fishy, the fairy tale has collapsed into one of tech’s most brazen frauds—an object lesson for every AI-flavoured startup chasing easy money.

Cybernews has now confirmed that “Natasha” never existed. Instead, Builder.ai ran on sweat, not silicon: about 700 human engineers stationed in India turned customer specs into code while company marketing touted a neural network that “writes 88 percent of the software for you.” Inside dev chat rooms, the gag became: “It’s not AI, it’s Indians.” That is classic “AI-washing”—relabeling plain old labor as machine magic to lure deep-pocketed backers. 

The masquerade might have endured if the money had matched the myth. Yet an audit revealed that 2024 revenue was just $50 million, miles below the $220 million Builder.ai boasted to investors. Investigators also uncovered a round-tripping scheme with Indian unicorn VerSe Innovation, apparently designed to inflate sales on paper—book the cash as income, quietly send it back, and voilà: hockey-stick growth. Once creditors smelled the ruse, leading lender Viola Credit yanked $37 million from company accounts, freezing payroll in five countries and forcing an abrupt bankruptcy petition in late May. 

Employees learned the truth in a grim all-hands call: funds were gone, 80 percent of the workforce would be pink-slipped, and interim CEO Manpreet Ratia had no path to keep the lights on. For long-time staff the only surprise was how long the charade lasted. Posts on Reddit dating back to 2019—just months after a Wall Street Journal exposé suggested humans, not algorithms, were doing the heavy lifting—warned that Builder.ai was “Theranos for code.” Nobody listened.

Read More: We Need A New Way For Startups Valuation And We Need It Fast

The fallout stretches far beyond one failed company. Microsoft planned to fold Builder.ai’s tech into Azure and give Natasha a starring role in its Copilot suite; those integration road maps are now scrapped. Venture firms that chased frothy AI valuations are writing down investments, and regulators on both sides of the Atlantic are investigating whether investor updates crossed the line from puffery to securities fraud. The debacle lands as lawmakers debate guardrails for generative AI, reinforcing the argument that “trust us” isn’t enough when billions—and critical software supply chains—are at stake. 

So what should early-stage founders and backers learn from the wreckage? First, buzzwords don’t build products. If you’re genuinely using AI, show your work—publish benchmarks, open your model cards, and invite third-party audits. Second, growth targets must track real usage, not creative bookkeeping. Round-tripping inflates nothing but lawsuits. Third, corporate governance matters: Builder.ai’s board let founder Sachin Dev Duggal retain multiple seats and the title of “chief wizard” even after he resigned as CEO, leaving no independent watchdog in the room. 

Finally, investors need to stop grading companies on press-release prose. Demand proof of automation efficiency before wiring funds. Call random customers. Interview rank-and-file engineers, not just C-suite evangelists. If a startup claims its AI writes most of the code, ask to see a commit log that isn’t full of human fingerprints.

Builder.ai’s implosion will be remembered as a milestone—the moment the hype cycle hit a wall of human reality. The tragedy is that the underlying idea, low-code tooling to democratize software, remains powerful. But trust is fragile in tech; once broken, it drags the entire ecosystem down. For the next generation of AI entrepreneurs, the message is brutally simple: ship real technology or someone—an auditor, a journalist, or a Reddit whistle-blower—will eventually call your bluff. Better to build the hard way than to let 700 hidden engineers pretend to be the future.