
There's a 3,900-word piece making the rounds on X about how 99% of AI agent content is fake. The core argument: people post cinematic screenshots of ten browser agents running simultaneously, but never show what those agents actually produced. No metrics, no output, no ROI. Just vibes and engagement bait.
The author is right. And the piece is doing exactly the thing it describes.
The opening line — "If you need pictures to stay engaged, bookmark this and come back when your attention span recovers" — is engagement bait in a trench coat. Positioning yourself as the lone honest voice in a dishonest space is one of the oldest plays on social media. It triggers the same emotions as the hype posts: curiosity, flattery, tribal identification. The packaging is different. The machinery is identical.
To be fair, there's a genuine signal here. The three red flags — no specific output, no before-and-after metrics, vague use cases — are a useful filter for evaluating any AI demo, not just agents. And the author's admission that they spent two full days getting five agents to fight over git credentials before shipping nothing is the single most useful paragraph in the piece. That's the failure report the space actually needs. More of that, less of the 3,900-word frame around it.
But the "real" examples deserve the same scrutiny the author demands of others. A TikTok pipeline that earned $4,000 in a day. A car negotiation that saved $4,200 overnight. These are presented as receipts, but a dollar figure isn't evidence — it's an anecdote with a number attached. One person's success tells you nothing about the base rate. How many people tried the same setup and got nowhere? We don't know, because nobody posts that.
The deeper problem the piece won't touch: the author frames the current state as "we're not fully there yet" — never as "this might be narrower than we think." That's still a bullish position wearing a realist costume. Honest about the hype, sure. Honest about the ceiling? Not so much.
Every hype cycle produces its class of honest brokers — people who build credibility by calling out the noise while staying carefully long on the underlying asset. It happened with crypto. It happened with NFTs. The honest broker gets to have it both ways: the credibility of the sceptic and the upside of the believer. It's a comfortable position. It's also, by definition, not the full picture.
The most valuable thing in the piece is a line buried near the end: "I spent 2 days and shipped nothing." If the entire post had been 400 words expanding on that sentence — what went wrong, specifically, and what it taught him about where agents actually break — it would have been more useful than the 3,500 words of scaffolding around it.
But that post wouldn't have gone viral. Which is, of course, the whole point.
Stay current weekly
Get new commentary and weekly AI updates in the AI Primer Briefing.