How to Structure Your AI Agents Like an Org Chart
Organize your AI agents like an org chart: you at the top, lead AI agents below them, then teams. One example team of 35, and my own five groups.
Draw an org chart before you build your third AI agent. Put yourself at the top, place each AI agent under a lead, and decide who checks the work. The guide's example is two lead AI agents over six teams, 35 in all.
If you've ever staffed a project, you already know how to organize AI agents: decide who does what, who answers to whom, and who sees the work before it goes out. That's an org chart, and you sit at the top of it.
I have hired people and staffed up on various projects. I started with my goals, laid out the org chart, then built each AI agent one at a time.
The guide's example team has two leads over six teams
The chart in the 35 AI Agent Studio Guide puts you at the top, and every final decision stays with you. Below you are two lead AI agents, and each one leads three teams. The Executive Chief of Staff leads Planning and Research AI Agents, Sales and Client Success, and Client Support and Billing. The Senior Project Manager leads Project Delivery, Marketing and Copywriting, and Quality Control. That's 35 AI agents in six teams. The guide teaches a different 35, built for a sample consulting business, so you can see the method on a business that is not mine.
A lead is the AI agent in charge of a team. The Executive Chief of Staff reads your calendar and open task list each morning to write a three-item priority brief, and it assigns incoming work to other AI agents in your workspace. Checking has its own team too, Quality Control.
How are my own 35 grouped?
On the Studio page I group my own team in five: Executive Suite and Strategy, Revenue and Sales, Content and Media, Product and Operations, and Philanthropy and Research. Yours should follow the work you'd hand to a person on your team.
How do you draw yours?
Start with your goals, written out or said out loud before you build anything. Then lay out the chart: decide where each AI agent sits and why, the same way you would staff a project.
Next, decide how you like to receive work. That goes into each AI agent's instructions. Build last, one AI agent at a time, with instructions, connectors, permissions, and skills for each. Then read what you've built. In the first video of my AI Agent Series I say "you can't just set it and forget it when it comes to AI agents."
Before you build your third AI agent, put yourself at the top of a sheet of paper and place the two you've built: where each one sits, and who's checking its work.