0:00
/
0:00
/
Preview

Your agent needs a SOUL.md you can't write from scratch. I built a 45-minute prompt that writes it for you.

Every agent product is solving the wrong problem — and the right one sits upstream of all of them.

The most expensive problem in the agent space right now isn’t infrastructure, model selection, or security. It’s that the people adopting these tools can’t describe what they actually do all day — not in the resolution an agent needs.

OpenClaw has 250,000 GitHub stars. Jensen Huang compared it to Linux. Meta bought Manus for $2 billion, Nvidia shipped NemoClaw, Anthropic shipped Dispatch, Perplexity built a dedicated hardware product around the concept. A thousand engineers lined up outside Tencent’s headquarters in Shenzhen to get it installed on their laptops. The install problem is solved — you can have a personal AI agent running on Telegram in the time it takes to make coffee. And the most common message in every agent community after a successful setup is three words long: “Okay... now what?”

That’s not a tech support question. It’s what happens when you sit down to delegate and realize you’ve never described your own work in the resolution anything else can act on — human or machine. I’ve been reading those threads for months. The install goes fine, the first task goes fine — and then the agent sits there waiting for instructions the human doesn’t know how to write. There’s a structural reason for that, and it matters for your career whether you ever touch an agent or not.

Here’s what’s inside:

  • The 40-hour wall. What actually happens when people try to spec their own work for an agent — and why every product in the space skips this step.

  • The expertise trap. The more senior you are, the less visible your own operating system becomes to you. That’s not a character flaw — it’s the intended outcome of how expertise develops.

  • The career problem hiding inside the agent problem. Delegation failures, promotion ceilings, and institutional knowledge loss all trace to the same root cause.

  • What to do about it. The first agent worth running isn’t an assistant. It’s an interviewer — and I’ve built one.

The agent revolution has a prerequisite nobody’s selling. This piece is about the prerequisite.

Subscribers get all posts like these!

Listen to this episode with a 7-day free trial

Subscribe to Nate’s Substack to listen to this post and get 7 days of free access to the full post archives.