Engineer.ai claimed AI built your app. It was just people the whole time.

Remember when every startup slapped “AI” on its pitch deck and VCs threw money at it like confetti? Engineer.ai was a textbook case. The Indian company claimed it had built an AI platform that could knock out 80% of a mobile app in about an hour. Investors, including a SoftBank-owned firm, poured in nearly $30 million.

Except the AI was mostly humans.

The Wall Street Journal broke the story, and the details are deliciously predictable. Engineer.ai wasn’t using machine learning to generate code. It was using human engineers in India and elsewhere to manually assemble apps. The company’s “Chief Wizard” (yes, that’s a real title) Sachin Dev Duggal had been telling audiences that the AI was “human-assisted” — which turned out to mean “humans do everything, we just call it AI.”

The company’s own chief business officer, Robert Holdheim, sued earlier this year, claiming Duggal told investors the product was 80% done when it barely existed. That’s a bold move for a guy whose business card probably says “Chief Wizard” too.

Engineer.ai did try to defend itself. They told the WSJ they use natural language processing to estimate pricing and timelines, and a “decision tree” to assign tasks. Neither of those is modern AI. That’s basic software. No AI agent is compiling code. No neural network is designing your login screen. It’s people.

This isn’t an isolated case. The MMC Ventures study cited in the article found that 40% of European AI startups don’t actually use AI in any meaningful way. Yet those same startups attract 50% more funding than traditional software companies. The .ai domain from Anguilla has doubled in registrations. Everyone wants a piece of the hype.

What gets me is how familiar this all feels. It’s the same pattern as content moderation at Facebook and YouTube — they claim AI is doing the heavy lifting, but it’s actually thousands of underpaid contractors in Manila or Austin reviewing the garbage. AI is often just a thin layer of PR over a human workforce.

Engineer.ai isn’t even the worst offender. At least they eventually plan to build real AI, according to their story. Plenty of startups never even get that far. They raise money, burn through it, and pivot to something else before anyone notices the emperor has no clothes.

The uncomfortable truth is that real AI is hard. Training models requires massive datasets, expensive compute, and teams of PhDs. It’s much easier to hire a few developers in Bangalore, build a nice website, and tell investors you’re “leveraging machine learning.” And as long as VCs keep funding buzzwords over substance, this will keep happening.