Is OpenClaw’s Hype Real? Are AI Agents Replacing Jobs? What’s the Anti-AI Thesis?

The Napkin Math duo is back with another episode. 

We spend most of this one trying to make sense of AI agents, OpenClaw - the new kid on the block, and why this wave feels messier but more real than the last few. 

We talk through hype vs reality, what agents can actually do today, and how work is quietly shifting from roles and seats to tasks and systems. We also discuss AI fluency levels, how people are using Claude Code in real marketing workflows, one-person companies, and end with the Anti-AI portfolio.

Timestamps

00:00 Intro

01:15 OpenClaw & the agent breakout moment

06:52 Hype vs. reality: shiny tools, boring winners

09:45 The Dropbox lesson: abstraction beats raw tech

12:43 AI agents, security & where this breaks

19:07 AI vs SaaS, leverage & outcome-based work

24:58 Are AI agents actually killing jobs?

28:23 Levels of AI fluency (and how people keep jobs)

37:26 Using AI to save your job, not lose it

45:18 Installing skills: marketing, SEO, CRO, execution

47:12 When agents go wrong: prompt hijacking & trust

52:03 The anti-AI portfolio (what AI won’t replace)

57:18 Why live experiences still win