AI-washing
Builder.ai and the label that outran the build
Builder.ai · Software · United Kingdom
A startup sold an AI that would build most of your app, and raised on that story to a reported $1.5 billion valuation. When the cash was seized, what remained looked more like a large engineering team than an autonomous one.
Peak valuation, reported: $1.5B · Raised ~$445–500M
~300% overstated (reported)
The record
- Founded in 2016 as Engineer.ai, the London company raised roughly $445–500 million — investors included Microsoft and the Qatar Investment Authority — at a peak valuation reported near $1.5 billion. (The Register; eWEEK; Rest of World, 2025)
- Its “Natasha” assistant was marketed as AI that could build much of an app; reporting describes roughly 700 human engineers, largely in India, doing much of the work. A 2019 Wall Street Journal report had already found the company relied on human engineers rather than AI for most coding.
- In 2025 a reported audit put revenue at roughly a third of the ~$220 million claimed; lender Viola Credit seized about $37 million, and the company filed for insolvency on 20 May 2025, laying off close to 1,000 staff. (Bloomberg; The Register, 2025)
The story that raised
Builder.ai’s pitch was one of the cleanest in the category: an AI assistant named “Natasha” that would build most of your software, making app-development “as easy as ordering a pizza.” The story raised on it. Over its life the company took in a reported $445–500 million from investors including Microsoft and the Qatar Investment Authority, reaching a peak valuation reported near $1.5 billion.
The premium the label commanded was the point of the exercise. A firm that assembles software with a large offshore engineering team is priced like a services business; a firm whose AI does the assembling is priced like a platform. The gap between those two multiples is the AI-washing premium — and it is borrowed money.
What the reporting found
According to reporting from Bloomberg, The Register and eWEEK, much of the building was done not by Natasha but by roughly 700 human engineers, largely in India. This was not a late discovery: a 2019 Wall Street Journal investigation had already found that the company — then called Engineer.ai — relied on human engineers rather than AI for most of its coding work. The label had outrun the build for years.
The through-line matters more than any single number. A capability was marketed as autonomous; the reporting describes it as substantially human. That gap is survivable while the money keeps flowing and no one audits it too closely. It is not survivable when either of those changes.
The collapse
Both changed in 2025. A reported audit put actual revenue at roughly a third of the ~$220 million the company had claimed. Its lender, Viola Credit, seized about $37 million from its accounts, leaving only a few million in restricted funds. On 20 May 2025 Builder.ai filed for insolvency; close to a thousand employees were laid off.
Separately, reporting alleged “round-tripping” — companies billing each other for services to inflate revenue — between Builder.ai and the Indian firm VerSe Innovation between 2021 and 2024, a matter reported to be under US investigation. VerSe has denied the round-tripping allegation. We treat those allegations as unproven. What is not in dispute is the shape of the ending: a company that marketed AI, relied heavily on human engineers, and ran out of cash.
The premium, called in
Builder.ai is the AI-washing pattern at full scale, and its lesson is about timing. The premium a label commands is not free money; it is a claim against a product that has to catch up. As long as it goes unexamined, the borrower looks like a platform. When it is examined — by an auditor, a lender, a journalist — the correction is not surgical.
The repricing takes the honest parts down with the rest, because by then no figure from the company can be trusted on its own. That is the specific danger of building on the label rather than the capability: you don’t get to choose which of your numbers the market stops believing.
The lesson
The AI-washing premium is borrowed against a product that has to catch up. When the gap between the label and the build is finally priced, the repricing is not surgical — it takes the honest parts down with the rest.
How we’re reading this
The revenue-overstatement and “round-tripping” claims are reported allegations, some reported to be under US investigation; VerSe Innovation has denied the round-tripping allegation, and we treat the fraud allegations as unproven. Figures are attributed to the outlets that reported them. What is not in dispute: the company marketed AI, relied heavily on human engineers, and became insolvent in May 2025.
Sources
- 01Builder.ai coded itself into a corner — now it’s bankrupt
The Register · May 21, 2025
- 02700 Engineers — Not AI — Developed Code at $1.5B Disruptor Filing Bankruptcy
eWEEK · Jun 13, 2025
- 03What was Builder.ai and why did it shut down?
Rest of World · Aug 1, 2025
- 04Builder.ai faked revenue with India’s VerSe Innovation through round-tripping
DealStreetAsia · Jun 13, 2025
- 05VerSe Innovation refutes round-tripping charges
Exchange4media · Jun 13, 2025
Compiled from public filings, court records, company statements and reputable press. Figures are attributed to their sources; allegations are labeled as such. Not legal or investment advice.