I hate LinkedIn. But this morning, I clicked on a link and infinite-scrolled through it for a few seconds. Skimming the headlines, I noticed a pattern. Then I quickly closed the tab - to me it's a giant firehose of garbage sold as career advancement advice.
It's all theater.
The pattern I noticed gave me something to think about. There were three posts with almost the same message: A team of tireless AI agents running a business overnight autonomously. A founder who thinks big thoughts while automation handles the rest. Twenty tools stitched together produce compounding gains. What is missing from every one of these posts is proof. A number.
That gap between fiction and reality is not incidental. I bet the truth resembles something close to this: Someone got fired from their previous job, applied to hundreds of positions using a $20 "pro" AI plan, but hasn't found a new position yet. They found out their AI plan also covers code generation, skills, MCP, and tool usage by watching a few youtube videos. So they started tinkering with an idea. Boom. Now they're suddenly a startup. No employees (yet!), but they automated the shit out of this cool new B2B SaaS idea. Yay - Let's go live! After two days, they notice they still haven't won any customers, so they plaster LinkedIn with AI-generated posts about how AI runs their non-existing startup. What a start, what a great way to win customers: With a bunch of AI-generated lies. Cheers to hustle culture 2.0!
It's all theater.
When it comes to larger companies, there's also a lot of theater surrounding autonomous AI agents. Teams claim to "enhance" their workforce. AI agents handling their customer support, sales, coding. Not just the dirty work, but also the oversight. There's always this one weird guy who brags about having automated Product Management in their team (you can count yourself lucky if it's not the Product Owner). But if these invisible teams actually existed, the operating costs would have moved, revenue per employee would have moved, or at the very least headcount would have moved. In most orgs, none of those numbers have moved in a way that would survive an auditor. What has moved is the theater around them.
Maybe it's time to start asking a very simple question at every AI update: what changed in the numbers you already track, and can you attribute it? Anything else is a performance. The absence of a P&L delta is not a measurement problem. It is proof for one thing, and one thing only:
It's all theater.
Don't get me wrong: I'm a heavy AI user myself, and it greatly advances what I'm capable of accomplishing in a single day at my day job. And at night, on my side projects. Thinking about it, I just realized I have launched four public projects in my spare time this year: Ultimam.com (2-Player game/Scorched Earth clone, together with my 8-year old son), Haushaltsloch.de (federal budget configurator for Germany), Fanlink.de (bet on soccer games with your friends), RitualDissent.com (human-agentic feedback loop), and this blog RefactoringOrgs.com. Wow - How cool is that?! But I'm not the one making bold claims here. Because that would be theater.
Thoughts? Find me on Bluesky.