This Indian Startup Claimed To Have AI That Built Apps, But Just Used Humans To Do Their Work

Artificial intelligence is the buzzword of the season, with every startup trying to include some level of the technology so they can market it like that.

For Indian startup Engineer.ai, it seemed they had the perfect product, an AI that develops apps. But the truth was different.

Needpix

Set up by Sachin Dev Duggal, the main idea behind Engineer.ai is that you can simply pay to have the artificial intelligence build your app in a third of the time needed, and most importantly without needing any coding knowledge yourself. On paper, that sounds like a flawless business plan. There's just one hitch.

Engineer.ai didn't actually have an AI to speak of.

According to a report from The Wall Street Journal, the company wasn't using an AI to build apps. Instead, it was using mostly human engineers to code apps, so they could attract enough customers and investors to keep the startup running until it could complete its automation platform.

Engineer.ai claimed to use "human assisted" AI tools to build 80 percent of a customer's app from scratch in an hour. At least that's what self-professed "Chief Wizard" Duggal said onstage at a conference last year. In reality though, they were using human engineers in India and elsewhere to speedily build apps.

Founder Sachin Dev Duggal - Sahin Dev Duggal/Instagram

The company was sued by its chief business officer Robert Holdheim earlier this year, who accused the company of exaggerating its AI abilities in order to secure investor funding to actually build that technology. He said Duggal "was telling investors that Engineer.ai was 80 percent done with developing a product that, in truth, he had barely even begun to develop."

And it apparently worked, because Engineer.ai secured nearly $30 million in funding from a SoftBank-owned firm, as well as others

In reality though, the startup admitted to WSJ that it uses natural language processing to estimate the pricing and timelines of requested app features, as well as a "decision tree" to assign tasks to engineers. In case you're still parsing that, it's technically a rudimentary form of machine learning being used. But rather than "human-assisted AI", it's more like AI-assisted human work being done. The AI itself had no part in the coding of the app.

Engineer.ai

And there are probably plenty of companies attempting to use this sort of trickery to get a leg up. WSJ reports that use of the .ai domain has doubled in the past few years. However it's likely very few of these startups are actually using AI in any meaningful way.

UK investment firm MMC Ventures told the publication that startups with some type of AI component can attract as much as 50 percent more funding than other software companies. More importantly, they suspect 40 percent or more of those companies don't actually use any form of real AI at all. After all, not only can it be expensive and time-consuming to develop, train, and test an AI, deploying it at scale is also a problem. And while tech giants like Google and Facebook have all the resources they would need to experiment with AI, startups do not.