The golden era of AI-pilled founder entitlement

Having spent the last month reading twitter, hacker news and other sites while being on holidays has left me raising eyebrows more often than not. Everyone seems to be doing the right thing and think they're unstoppable. They're preaching left and right about them being on the right path to stardom. But, are they?

Scroll far enough through founder Twitter right now and you will find screenshots of solo founder's "team" page with five AI agents instead of actual humans. Another few posts, and you run into someone claiming to have made it to 20k MRR in a couple of weeks with a few agents only. Further down, one person in Bali runs outbound sales for other companies using agents and she - allegedly - makes dozens of thousands of bucks a month. Another person handles millions of support tickets for companies with an army of agents. Another one is close to becoming the first - legit - one-man unicorn. If it's linkedin, these baity posts will end with something like "if you're interested in my secret sauce, reply with BUBBLEGUM and I'll dm you my playbook". Engagement baits, at best, truthful or not.

Here is the part nobody posts a screenshot of: the first time one of those agents is confidently, fluently wrong about something that actually matters, or the support agent promises a refund policy that does not exist or the coding agent sends a test email to all your subscribers (rings a bell? 👋). Worse: the code review agent approves a change that breaks billing, or SEO, or both!

In a normal company, someone in the process says "wait, that doesn't sound right", or asks for extra time to think things through. Sometimes, in the gap between a teammate finishing a task and a reviewer actually looking at it, someone catches that something's off - and the change gets fixed, or the deploy gets held, before it ships.

Two things have changed: time and alignment.

In the AI era, time acts differently. Between humans, time is a shared concept. We all perceive it in the same order of magnitude. An hour for me is an hour to everyone else. There's no scenario in which I experience a second and someone next to me experiences a week. We play by the same rules.

AI doesn't abide to these rules. As the fictional AI in HER said: It’s like I’m reading a book… and it’s a book I deeply love. But I’m reading it slowly now. So the words are really far apart and the spaces between the words are almost infinite.

This soul-crushing revelation explains partially why our relationship with AI agents is very asymmetrical. It doesn't operate at the same scale. Even if, fundamentally, computers run with clock cycles and those are time, in and of themselves, the pace at which AI and software executes, thinks and processes can't be compared to that of humans.

We can produce faster than ever, just like cars enabled humans to move faster than ever, but, in what direction?

The second thing that's changed is alignment. In a human environment, alignment between individuals fluctuates. Leadership determines whether they can keep a team alignment, or whether someone has to be given the boot. Leadership also means sitting down to talk to every member and listen to their feedback.

This doesn't happen with AI. AI doesn't have judgement, and if not steered in the right direction, AI agents are exceedingly sycophantic. They won't even contradict you. At most, they'll gently push back on something, but they won't interrupt the flow, because they're engineered to produce more and more work.

In a real company, you hire people who are better than you - if you do things right. One of their jobs is to question your decisions as CEO. If you hire people that can only execute, you're not a CEO, you're a primary school football coach.

These two factors combined make for an explosive cocktail: people producing at high speed with no one telling them not to.

When you can do everything, when you feel like you're unstoppable with superpowers, how can you not become entitled?

How can we avoid this?

We can't fight the tide, most likely. We will have to ride this wave, whether we like it or not, but there's hope for the resistance: we can stick to each other.

You do not need to abandon the agent trend, you just need to put a human back in the loop: one that supervises your project, counsels or just checks on you every now and then. This can be a paid advisor for two hours a month, a peer founder whatsapp group, or a part-time contractor whose entire job is to look at what shipped last week and ask why.

The point is not their expertise, it is that they have no stake in you being right, so they will actually say when it is not. They can be blunt, and they have to be blunt. They have to force you to think.

Go find this person. It's your next power-move unlock that can increase your chances of becoming successful in the mid-term, by placing fewer but better bets and thus avoiding burnout.

Agents are not your team: they are your hands - more hands, cheaper hands, faster hands than you could otherwise afford at this stage. They're minions. But judgment is still your job as a founder and/or CEO, and it is the one thing you cannot delegate to something that has no reason to tell you when you are wrong.

Outsource the execution as aggressively as you want but find a way to give you time to breathe and exercise your judgement. Your agents won't ever need that coffee break a coworker needs.