It used to be deepfakes and janky marketing copy. But now people are raising money and selling dreams on the lack of practical knowledge of ‘AI.’

Devin (Cognition’s recent product announcement) pushed it a bit. It was billed as an AI assistant capable of replacing a programmer in all major areas, with a demo to prove it. It heralded a $21M Series A investment … but once people had a week to watch their demo a few times, they could easily tell it was mostly fiction.

Now, the “smoke and mirrors” tech demo has been a grey area and a pitch staple forever. But they were called prototypes — and on the investor side, there was still ostensibly emphasis on ‘unfair advantages’ and differentiators.

What’s problematic about Devin and other pitches is the tech is reproducible for less than $21M (or even $21K) using off-the-shelf tools. It’s a great demo, but it’s shaky ground when you lead the market to believe it’s here today. I believe in the promise of GenAI but this “fake it till you make it” trend is really hurting the credibility of truly useful work.

Consider some warning signs:

1: “We have AI Now!” Sometimes, they’re not even referring to Generative AI or just tacked on somebody else’s plug-in to say, ‘We have one of those’. If you read the fine print, it’s usually billed as a beta feature that’s rolling out slowly, or the pricing is an add-on at a suspiciously high-cost (to discourage use).

2: “Nobody Else Does This!” Check if it’s something the person down the hall can already do with a GPT or Copilot. You’d be surprised how many can. You can’t build a business around something that has zero barriers to entry. These pitches ride on the fact 98% of people don’t know they’re already capable of writing prompts themselves.

3: “Eliminate a Job!” Our tool ‘entirely replaces’ [something] like a job or a product, especially something with an autonomous aspect. Devin claimed this, creating an impressive demo now widely known as hilariously hand-cranked. Nobody uses GenAI for huge chains of process unless low-quality output is acceptable (marketing and email outreach copy) or unless there are several human review steps.

4. “The Biggest Use Us!” Our Tech Used by the Fortune [x]. No F100 companies are doing strategic, transformational, generative-AI-based deployments. Not the F500 either. Very few of them are even planning for it, these are unpaid proofs of concept at best. Letting people use a chatbot on a FAQ or feeding the employee manual into Claude doesn’t count. I talk to people at these firms, and there is still a strong wait-and-see attitude.

Have you seen other warning signs that a pitch isn’t what it seems? And to be fair, has anyone seen companies deliver hard on their promise? They have to be out there…

Follow and comment!