Kendi Yapay Zeka Ekibimi Kurdum. Ben Kod Yazmıyorum.

Anyone who wants to advance in artificial intelligence eventually has to learn to play the orchestra. So, it needs to understand how to manage multiple AI models simultaneously as a single unit. In this video, I

wanted to show how I run multiple AI agents on the same project, what I pay attention to, and what I don't pay attention to while doing so. We will also return to cloud code from time to time throughout the video. I will demonstrate this with examples in the cloud code . Let's start slowly. This is

actually the most important point for me. I'll put an agent in charge, guys. I'll give you the other agents underneath. The main agent I'm talking to doesn't write code. Generally, the agent I'm talking to manages the other agents below them. This could be the cloud, it could be Codex. These two are included in my diet. Generally, the

main agent I choose is usually the cloud model . Especially with Opus 5.5, this phenomenon reached a completely different level. The agents listed below were normally codex agents. To balance it with my limit in the Codex. But to be honest, I now generally make them my opus as well. Depending on my limits. Look, he's actually

orchestrating it. He's voting for all of them. This is what we mean by orchestration, folks. We don't write code. I have given jobs to a total of 2000 agents. They created 1653 sub-agents. These sub-agents actually did their job. Look, he's a sub-agent working for Codex. So I'm not actually writing code. I speak to the person above as if they were my boss.

He manages those below him. Let me give you an example. Look, I'm copying the command. I'll tell you to open it in the terminal now. Now, look, after you log in, I'm copying the prompt. Prompt is actually a simple thing. I've included a sample file. I tell myself not to read any of them.