O jeito certo de construir um agente de IA

I'm going to tell you something, and I'm sure you'll like it. Look , everywhere I go, every person I talk to, every news story I read, or even my own Instagram, which should be a place for recreation, gives me anxiety attacks because I'm constantly being bombarded by AI. I was going this way, I was going that way

, folks. SAS is dead, OpenCLA is dead, Claudio is dead, the Portuguese chat is dead, my God, everyone is dying. And I feel like I don't know what to do with all this information, with so many new things happening, so many companies dying, and with this crazy speed that leaves us with constant gastritis from anxiety. But I have

good news for you. So, what happens in practice is that, amidst the comings and goings of hundreds of companies and tools, one thing persists, which is what you'll learn in this video: about agents. So, in today's video, I want to share my complete vision of what an agent is in 2026. I'll explain

what an agent is, how an agent works, and the three pillars of an agent, which are basically routine, context, and skills. I'll also talk to you a little about the mistake most people make when building their agents. I'll also explain the loop you need to know to test with your agents, and I'll give you some

small tips throughout this video to give you a good foundation to build your first agent. So it does n't matter if you do it with OpenCLA, with Cloud, with GPT Chat, it doesn't matter . What matters is that you understand the logic behind having an agent and how to operate one so you can fully appreciate the

technology in its purest essence, okay? So, I've broken up today's video. Well, if you're new here on my channel, first of all, welcome! I'm Bruno Camoto, and it's an indescribable pleasure to have you here with me. Here on my channel I talk a lot about SAS, microSAS, AI, entrepreneurship in general, and I really