Engineer.ai’s AI Was Just Humans in a Trench Coat

Remember Engineer.ai? The Indian startup that promised an AI assistant could whip up 80% of your mobile app in about an hour? Yeah, turns out the only thing artificial about it was the hype.

According to a Wall Street Journal report, Engineer.ai wasn’t using any real AI to assemble code. Instead, it was relying on human engineers in India and elsewhere to stitch apps together manually. The company had raked in nearly $30 million from SoftBank and other investors, much of it riding on the promise of a futuristic automation platform. But the actual tech? A decision tree for task assignment and some natural language processing for estimating timelines. That’s not even close to the kind of AI that powers modern machine translation or image recognition.

The company’s founder, Sachin Dev Duggal—who also calls himself “Chief Wizard”—was out there on stage claiming you could build most of an app from scratch in an hour. But the company’s own chief business officer, Robert Holdheim, sued earlier this year, alleging that Duggal told investors the product was 80% done when it was barely started. That’s not a stretch, that’s a lie.

This isn’t an isolated case. A 2019 study from UK investment firm MMC Ventures found that startups claiming an AI component could attract up to 50% more funding than other software companies. They also estimated that 40% of those companies don’t actually use any real AI. Engineer.ai was just the most egregious example of “AI washing”—slapping the label on traditional software to grab attention and cash.

What really gets me is how predictable this is. The startup landscape is saturated. Investors are desperate for the next big thing. So you take a boring app development platform, add “AI” to the pitch deck, and suddenly you’re a disruptor. It’s the same trick that’s been pulled with blockchain, VR, and every other buzzword cycle. The difference is that AI is particularly easy to fake because it’s so vaguely defined. A decision tree? That’s been around since the 1960s. Calling it AI is like calling a toaster a smart appliance.

And let’s be honest, this reveals an uncomfortable truth about a lot of “AI” products: they barely exist. Much like content moderation at Facebook and YouTube, which uses some AI but mostly relies on armies of contractors, many AI startups are really just humans behind a curtain. The tech isn’t there yet, but the marketing is.

Engineer.ai isn’t alone. The number of companies using the .ai top-level domain has doubled in recent years. Japanese conglomerate SoftBank has pledged hundreds of billions in AI investments. The pressure to look like an AI company is enormous, and the incentives are all wrong. Investors want to fund AI, so startups pretend to be AI. It’s a vicious cycle.

The real tragedy is that this kind of fraud hurts the entire field. When a startup like Engineer.ai gets caught, it erodes trust in legitimate AI research and development. Genuine breakthroughs get lumped in with vaporware. And the next time a real AI startup pitches a genuinely useful tool, investors might hesitate.

I’ve seen this pattern before. Back in the 80s, it was expert systems. In the 90s, it was neural networks. Every AI winter was preceded by a wave of overpromising and underdelivering. Engineer.ai is just the latest reminder that the technology is still hard, and the hype is always easier.

So what’s the takeaway? Don’t believe the hype. If a startup claims AI can do something that sounds too good to be true, it probably is. Ask for specifics. Look for published papers, open-source code, or at least a demo that isn’t a slideshow. And if the CEO calls himself “Chief Wizard,” run the other way.