3 Comments
User's avatar
Eli Finer's avatar

Exquisite as always! Totally agree and it seems that for now the best combo is a human + AI duo - the duo can perhaps do 10x of what the human can do, but it doesn't mean the agent can be fully independent - even one "section chief", let alone a CEO with multiple subordinates. We're not quite there yet.

Pawel Jozefiak's avatar

The simulation here is useful. CEO agent -> department agents -> autonomous decisions. The gap between what a workflow looks like in a diagram and what happens when feedback loops break is real. My version: spent month one trying to build a "Jarvis" - general purpose, fully autonomous. Month two I had one agent that did one thing reliably.

That one agent taught me more. The workflow automation framing isn't a demotion - it's accurate. The gap between reliable workflow automation and actual autonomous reasoning is where most of the real work is, and most people underestimate it.

Daniel Blank's avatar

100%, there is value here. Also, paradoxically, workflow automation teaches you more about your business than you thought. It forces you to think in systems, even in places where you didn't do so before