Why organizing AI models by role (CEO, PM, designer) wastes money
The idea of assigning a dedicated AI agent to every organizational role — a senior product manager who manages a design intern, who manages the CTO — is understandable for people who think about work in those terms. But there's a real inefficiency in it, especially around cost. If you need a YouTube thumbnail, you shouldn't have to go to a 'CEO' agent who delegates to a 'PM' who delegates to a 'graphic designer' who delegates to an 'intern.' Every hop looks at context, soul, and skills, burning tokens along the way. You could have just gone straight to a capable model like Opus and said, 'Build me an image.' It doesn't have to be that deep. Everyone has their own workflows, and you should do whatever works for you — but for maximum cost savings and efficiency, keeping it model-by-model with custom profiles rather than role-by-role is the better approach.
Go deeper with Greg Isenberg
Ask anything about this, and Greg Isenberg answers from everything they've shared.
