Here was a pitch almost too good to question. You did not need engineers, or a technical co-founder, or even a fully formed idea. You just told an AI assistant named Natasha what you wanted, and Builder.ai would assemble your app the way you might order a pizza: pick the toppings, wait a bit, done. Investors loved it. Microsoft loved it enough to put money in. By 2023 the London startup carried a $1.5 billion valuation and a spot on every “future of software” list. There was just one awkward detail about the AI that built your app on demand. A lot of it was people.
Key Takeaways
- Builder.ai was a $1.5 billion startup backed by Microsoft and the Qatar Investment Authority, selling an AI assistant called Natasha that promised to build apps for anyone.
- It collapsed into bankruptcy in May 2025, after both its finances and its “AI” came apart in the same few weeks.
- Much of the “AI” was human. Reports described a network of more than 700 engineers, mostly in India, doing the building by hand behind the assistant.
- The trigger was money. An audit put 2024 revenue near $50 to $55 million, not the roughly $220 million shown to investors, and a lender seized about $37 million from the company’s accounts.
- US prosecutors got involved. The Southern District of New York subpoenaed Builder.ai, turning a startup failure into a possible fraud case.
What Happened to Builder.ai?
Builder.ai, one of Europe’s most hyped artificial intelligence startups, filed for bankruptcy in May 2025 after its revenue and its technology both turned out to be far less real than advertised. It was not a slow fade. It was a cliff.
The company began in 2016 as Engineer.ai, founded by Sachin Dev Duggal, a promoter with a gift for a headline who later gave himself the title “Chief Wizard.” Its promise was seductive in the exact way the moment wanted: democratize software, let anyone build an app, and let an AI do the hard part. Money poured in. Microsoft took a strategic stake in 2023, the Qatar Investment Authority and others joined, and total funding climbed to as much as $445 million. The valuation hit $1.5 billion, the magic word: unicorn.
Then, in the space of a single spring, it all unwound. Duggal was pushed out as CEO in February 2025, a new chief executive arrived, and within months the company admitted it had run out of money and was insolvent. The same pattern that runs through the Big Tech collapse files was here too, just faster: a story that outran the thing underneath it.
Was Builder.ai Actually an AI Company?
Not in the way it sold itself. Customers typed their requests to Natasha, the friendly AI assistant, and expected an algorithm to turn their words into working software. Behind that assistant, according to multiple reports, the real work was done largely by more than 700 human engineers, most of them in India, assembling apps by hand. The “artificial” intelligence had an awful lot of the human kind stitched into it.
Builder.ai always pushed back on the blunt version of this. It insisted Natasha was a genuine tool and that human engineers were a normal part of software delivery, not a secret. But the doubt was not new. As far back as 2019, when the company was still Engineer.ai, a Wall Street Journal report questioned how much of its “AI” was automated at all. Builder.ai survived that round of skepticism. What it could not survive was the moment the money told the same story the critics had.
| Year | Moment |
|---|---|
| 2016 | Founded as Engineer.ai by Sachin Dev Duggal, promising AI-built apps |
| 2019 | A Wall Street Journal report questions how much of the “AI” is real |
| 2023 | Microsoft takes a stake; valuation reaches $1.5 billion, unicorn status |
| Feb 2025 | Duggal is replaced as CEO but keeps a board seat and “Chief Wizard” title |
| May 2025 | Reports of inflated revenue; a lender seizes about $37 million; SDNY subpoena |
| May 2025 | Builder.ai announces insolvency and begins bankruptcy in the US, UK and India |
Why Did Builder.ai Collapse?
It collapsed because the money was as invented as the AI, and money is the one story you cannot keep telling once someone audits it. In May 2025, Bloomberg reported that Builder.ai and an Indian firm, VerSe, had used round-tripping, essentially selling services to each other and buying them back, to make revenue look bigger than it was. When a new team actually checked the books, the numbers cratered.
That gap was staggering. Builder.ai had shown investors 2024 revenue of roughly $220 million. Audited, the reality was closer to $50 to $55 million, an inflation of several times over. Worse, the company had reportedly raised emergency debt partly on the strength of those rosy forecasts. So when the real figures surfaced, its lender, a consortium led by Viola Credit, declared a default and seized around $37 million straight out of Builder.ai’s bank accounts. A company that had raised nearly half a billion dollars suddenly could not make payroll. Days after a subpoena landed from federal prosecutors in New York, it gave up and filed for insolvency.
The Critical Choice
The decision that doomed Builder.ai was made at the very beginning, in what the company chose to be: an AI company that sold the appearance of AI rather than the substance of it. From day one the pitch was magic, an assistant that conjures software, and the reality was a large, expensive room full of humans doing the conjuring. That is not automatically a fraud. Plenty of “AI” products lean on people early while the technology catches up. The fatal part was building the entire valuation, the entire investor story, on the magic, which set a bar the product could never actually clear.
Once you have sold a $1.5 billion story that your software runs on artificial intelligence, every quarter has to look like magic too. When the real growth is not there, you are left with a terrible choice: admit the gap, or paper over it. Builder.ai, according to the allegations, papered over it, and inflated revenue is just the financial version of the same original lie. For years the company insisted the curtain hid a machine. When it finally opened, investors found people, and an audit found a hole. A business built to look like AI could not survive being measured like a business. That is the same rot at the center of Theranos, which faked the technology until the money ran out, and it runs through half the scandal files: the story got too big for the thing it was standing on.
Where Things Stand Now
As of 2026, Builder.ai is being wound down through insolvency proceedings across the US, UK and India, and more than a thousand employees have lost their jobs. The Southern District of New York investigation is ongoing, and Duggal has faced allegations including money laundering, which he denies. What began as a flagship of the AI boom is now its most cited cautionary tale, the go-to example when investors warn each other about “AI washing,” the practice of slapping an AI label on ordinary work to raise money at magical prices.
Our take: the cruelest irony is the timing. By 2025, real AI had actually become good enough to build simple apps from a description, the very thing Builder.ai had promised and faked for years. The company was destroyed not because its vision was impossible, but because it could not wait for the technology to become real, and chose to fake the numbers in the meantime. It sold a future it could not yet deliver and financed the wait with fiction. In an era where every startup is suddenly an “AI company,” Builder.ai is the warning label: ask what is behind the curtain before you ask what it is worth. For the human cost of these hardware and software dreams that shipped before they worked, see how our Big Tech investigations keep finding the same gap between the demo and the product. The wizard, it turned out, was just a man with a very good microphone.
