# Stop Overthinking AI: Use This 4-Step Framework ## Метаданные - **Канал:** Tiago Forte - **YouTube:** https://www.youtube.com/watch?v=tyUChp9yrqw - **Дата:** 21.04.2026 - **Длительность:** 9:20 - **Просмотры:** 13,406 - **Источник:** https://ekstraktznaniy.ru/video/49372 ## Описание 🧠 Want to participate in the next Second Brain AI cohort in the fall? Join the waitlist here: https://bit.ly/4t2974w Most AI frustration I see isn't about the tool but about the setup around it. In this video, I walk you through the four decisions that shape how much you get out of AI: which platform to use, which harness to run it in, where to interact with it, and which model to pick for which task. Get these right, and you'll notice the difference immediately. ✉️ Join my newsletter for weekly productivity and AI insights: https://bit.ly/4ttx4SK 0:00 The common mistake 0:54 Decision 1 2:54 Decision 2 4:28 Decision 3 6:30 Decision 4 8:26 Your AI stack is personal ## Транскрипт ### The common mistake [] Most of the AI frustration I see comes from one specific mistake. People tend to choose, let's say, chat GPT or Claude. They use it for a few days or a few weeks using the most basic vanilla offtheshelf setup and then it doesn't go too well. They think it's not very useful and they set it aside and don't come back. The real issue is not the tool. It's the stack underneath it. The environment and the tools that you build around it. You don't just think, "Oh, I want to get a Toyota and then buy a Toyota. " No, you choose the make, then you choose the model, trim. There's so many small decisions to get to the exact configuration of car that fits your needs. The same applies to AI. Here's what you need to consider when choosing your AI stack. It's not just picking an LLM. There are three other choices to make. Here's the first decision you'll ### Decision 1 [0:54] need to make. Which AI is better? Claude, Gemini, or Chat GPT? These tools weren't just built differently. They were designed for completely different kinds of people, different markets. OpenAI is very clearly pursuing the classic consumer playbook. Make it feel magical. Make it sticky so you can't turn away. Have zero setup so you can get started just like that. Make memory automatic so it's not something that you have to think about. It's designed to delight that novice user that arrives at AI really not knowing anything. Anthropic is really targeting professionals. It has opt-in memory for those who are privacy conscious. It's highly reliable. It doesn't have features such as image generation that are a little bit wonky. It's built for more complex knowledge work and business use cases. It's designed really to be trusted by professionals. And then you have the third major player, Google's Gemini, which has massive context windows, which it can do because it has all that big Google infrastructure. It's highly multimodal, so it's really the best at translating between different formats like text, video, audio, software. Gemini is really best in my opinion if you really live inside Google's ecosystem and you want the integrations with all the other Google products. So decision number one isn't which eye is best at a kind of universal level. Really it should be which AI tool was built for someone like you. Do you value convenience and accessibility which would be more open AI or control and reliability that would be more claude? Right there you can understand why I've really chosen claude from the very beginning. My usage of AI is primarily in my work and I can see over really years many decisions that Anthropic has made that really align with that focus on doing better, faster, higher quality work. Once you've made your choice, that's decision number one ### Decision 2 [2:54] done. But now there's a second decision. And this one is in a way even more important for how much the AI tool can do for you. Decision number two, which harness? See, the same underlying AI model can operate in completely different modes. Think about the difference between an assistant that can just answer your questions and a colleague, a collaborator who can actually move forward projects on your behalf. For Claude, the AI tool that I use the most, there are three harness levels. Chat, co-work, and code. Chat, GPT, and Gemini have direct analoges of each. The names change product to product, but the three categories are the same. From here on, I'm focusing on Claude. For day-to-day small questions and tasks, or especially if I'm on my iPad or my phone, I just do the standard chat. It's the easiest to use. It's the fastest. It loads instantly. I get my response right away. That is fine for probably 60 or 70% of my interactions. If I have something a little more complex, I'll often move from chat to co-work. Co-work is kind of an intermediate step. I have more power, more control, more reliability, more thoroughess, higher quality, I can trust the results more. And then when I do have something that is really challenging, really subtle and sensitive, really high stakes, and I really need every bit of capability that cloud has to offer, then of course I do go to cloud code. ### Decision 3 [4:28] Decision number three is where you interact with AI. Let's look at the options. The first one and by far the most common is inside the web browser. You go to, for example, claw. ai or chatgbt. com. And this is a great place to start. It's where all of us started. It works anywhere on any device. There's no installation necessary. It's fantastic for that occasional casual use. But you'll find it's a really big step up. I really recommend everyone who uses any LLM to download the desktop app. They're now available for all the major AI platforms. And this is what I use every day, all day. The desktop app 4:00 on my Mac. It actually has three separate tabs across the top. So with one click, I can switch between chat mode, co-work mode, and code mode. And there's still other options. A lot of people don't realize this, but there's something called claude code on the web. And I know this is a bit confusing because cloud code on the web is not referring to using it in a browser tab. What cloud code on the web allows you to do is to open up your phone and basically fire off requests, fire off tasks using the full power of cloud code, which then are completed on anthropic servers and then the result, the output is sent back to you on your phone or on your computer or really any device. Returning to the options for interfaces, there's still a couple more. There's one called the CLI which stands for command line interface. This is a much more advanced mode of using cloud code or any of the LLMs. It's really designed for software developers or people in technical roles. And finally, there's a fifth interface which is the which stands for integrated development environment. Overall, my take is you should really be using at the very least the desktop app. Moving from the browser, which again is what the vast majority of people do, to the desktop app, the native app on your computer is an absolute game changer. ### Decision 4 [6:30] Decision number four, which model? I want to show you something that I see constantly. Someone asks an important question, nuanced, high stakes, that needs careful reasoning. They get back a mediocre or even just a wrong answer. But here's usually the mistake they're making. They sent a complex important request to the fastest, cheapest model. Within any major AI platform, there are multiple models that you can choose from. For Claude, there's three main options. There's Opus, Sonnet, and Haiku. Each one of them is built for different situations. The Opus model is the most powerful, but that also means it's the slowest, it's the most expensive, and it burns through your tokens the fastest. Sometimes that's worth it for high stakes decisions, for complex analysis, for important, nuanced writing, big decisions that have a big impact. Sonnet is kind of the middle ground. It's a good balance of performance, speed, and cost. This is what I typically tend to use on a day-to-day basis for most tasks. And then you have Haiku, the fastest, cheapest, least powerful model. But for things like looking up simple information, short replies to questions, low stakes tasks, where speed is really what you care about, it's perfect. For most use cases, I use Sonnet. Like I said, it's a nice balance of performance and speed and cost. If I am, say, on the go on my phone and I just need a really fast answer, I might switch over to Haiku. And then if I have something that's really big or complex or high stakes, I'll often use Opus for that. Even though it's more expensive and uses more tokens, if I'm say creating a new app or I'm getting feedback on some of my more complex writing or I'm making a strategic decision about our product portfolio, I really want every bit of juice that it has to offer. It's almost like we now need to ### Your AI stack is personal [8:26] configure our own AI exoskeleton and it looks different for every person depending on their temperament, their needs, their goals, their strengths and weaknesses. That's really what I hope you take away from this. That the level of customization is both needed, it's necessary, but it's also tremendously empowering once you do it. And if you want to see what Claude Code specifically looks like in action, I really think it's the most powerful harness in the stack, I made a beginner's guide that walks you through everything. That one's up next.