Builder.ai promised to make building software "as easy as ordering pizza." Its AI assistant, Natasha, was supposed to be a breakthrough — a system that could create custom applications with minimal human input. Microsoft invested. The Qatar Investment Authority led a $250 million round. The valuation hit $1.5 billion. Then a Wall Street Journal investigation revealed the truth: Natasha wasn't AI. It was approximately 700 engineers in India writing code manually while the company told investors the work was machine-generated.
The deception wasn't subtle. Builder.ai claimed its apps were "80% built by AI" — legal filings later revealed the supporting technology was barely functional. Former employee Robert Holdheim sued for $5 million, alleging he was fired after flagging the deceptive practices. Internal documents reviewed by Bloomberg showed Builder.ai worked with VerSe, an India-based social media startup, to artificially inflate sales numbers through reciprocal billing between 2021 and 2024.
This wasn't a startup that overpromised on a roadmap. This was a company that sold a product as AI-powered while employing hundreds of humans to do the work manually, then fabricated revenue to keep the illusion alive. The word for that isn't "pivot." It's fraud.
The real story isn't that Builder.ai committed fraud. It's that Microsoft, SoftBank's DeepCore, the Qatar Investment Authority, and every other investor in the $450 million stack failed basic due diligence. The Wall Street Journal found evidence of manual coding in 2019 — six years before the bankruptcy filing. The company continued raising money for another five years after that reporting.
This is the environment that produces AI-washing at scale. When investors are so desperate to deploy capital into anything labeled "AI" that they don't verify whether the AI actually exists, you get Builder.ai. When the biggest names in tech are writing checks based on pitch decks instead of technical audits, you get a $1.5 billion valuation for a company whose core technology was a Slack channel where project managers assigned tickets to developers in India.
When Viola Credit, a major lender, seized $37 million from Builder.ai's accounts in early 2025, the company was left with $5 million in restricted funds. It couldn't make payroll. It filed for bankruptcy across five jurisdictions — the UK, the US, India, and others. Clients who had built their entire digital operations on Builder.ai's platform watched their applications go dark overnight. Customer data, transaction records, source code — all of it became inaccessible.
Builder.ai is not an isolated case. It's the most dramatic example of a pattern that runs through the entire AI bubble: companies that market human labor as artificial intelligence, inflate metrics to justify valuations, and rely on investor enthusiasm to outrun the truth. Industry experts estimate that 99% of AI startups will fail by 2026. Most of them won't fail because the technology doesn't work. They'll fail because the technology never existed.
When OpenAI raises $110 billion and projects a $14 billion loss in the same year, the legitimate AI companies are struggling. The illegitimate ones don't stand a chance. The only question is how many more Builder.ai-scale frauds are hiding in the $16.7 billion that VCs poured into proptech alone last year.