Red faces and recriminations are flying thick and fast after revelations that UK AI company builder.ai has been tricking customers and investors for eight years – selling an advanced code-writing AI that, it turns out, is actually an Indian software farm employing 700 human developers.
Founded in 2016, the company, also known as Engineer.ai, became a darling of the AI space and marketed itself so well that it attracted $700 million ($US444.5 million) investment from the likes of SoftBank, Qatar, and Microsoft – which integrated Builder.ai into its Teams collaboration suite.
Builder.ai CEO Sachin Dev Duggal built the house of cards based on the purported capabilities of the company’s AI assistant ‘Natasha’, which was said to be a no-code tool that could build apps six times faster than traditional development processes and be 70 per cent cheaper.
Duggal became an AI industry stalwart, giving himself the title of ‘chief wizard’ at the company and appearing as an expert on podcasts and news outlets that lauded him as a captain of the emerging AI industry.
Yet it all came undone after revelations the company – which appointed new CEO Manpreet Ratia in February – will declare bankruptcy after major backer Viola Credit demanded immediate repayment of the $77 million ($US50 million) loan it extended to Builder.ai in 2023, when it was valued at $2.3 billion ($US1.5 billion).
“Despite the tireless efforts of our current team and exploring every possible option,” the company said, “the business has been unable to recover from historic challenges and past decisions that placed significant strain on its financial position.”
Pulling back the curtain
The bankruptcy came after a Bloomberg investigation found that the company had been working with Indian social media startup VerSe Innovation to undertake questionable financial practices – including using ‘roundtripping’ to overstate revenues for 2023 and 2024 by a factor of four.
Natasha offered to build the app of your desire. Image: Supplied
As well as its financial shenanigans, the company’s questionable business has been outed after confirmation that Natasha was not, in fact, an AI at all – but a team of 700 very human Indian developers who were not only writing customers’ software, but tasked with behaving like bots.
Rumours to this effect had circulated since a 2019 Wall Street Journal report questioned Engineer.ai’s bona fides, but the truth was finally exposed this month with confirmation that the entire company was a lie – with former employees describing the company as “all engineer, no AI.”
Although the developers used a range of software tools in their work, coding was performed manually, meaning that while Builder.ai did eventually deliver apps to its customers, it was simply another player in an Indian offshoring industry attracting $27 billion ($US17.7 billion) annually.
That puts the company in a completely different market segment than the one that propelled AI-hungry investors through four funding rounds before and after the debut of OpenAI’s ChatGPT turned the global tech industry on its head.
The typo in the screen grab was probably a red flag. Image: Supplied
A reckoning for the AI industry
The combination of financial malfeasance and outright deception has reportedly attracted the attention of US federal prosecutors, with suggestions that a formal criminal investigation has kicked off as Builder.ai careens into the annals of AI’s worst ever disasters.
Reports suggest that around 90 per cent of AI startups fail within a year, whether due to a lack of market demand, financial instability, operational challenges, or AI’s technological complexity – and the demise of the likes of Artifact, Shyp, Tally, Eaze, and Ghost Autonomy are prima facie evidence.
Even major firms have struggled to get AI right, with McDonalds, Air Canada, Sports Illustrated, iTutor Group, and Zillow among those that have backtracked after wrongfooted AI implementations.
Gaps between AI hype and AI reality are regularly proving disastrous – as when the surprise debut of Chinese generative AI venture DeepSeek sent share markets tumbling in January, or when Microsoft and AWS announced in May that they would pause their expansion of AI data centres.
As of press time, Builder.AI’s website still claims that “AI means we can build more cost effectively and at speed”, describing an AI-based workflow in which Natasha is an “AI project manager” that helps customers “order an app, like you would a pizza.”