Hook

Their other posts in the index, biggest breakout first.
I'm going to take you from Claude AI beginner to expert in three minutes. Claude is one of ChatGPT's biggest competitors. Claude and ChatGPT are both language models, meaning they generate text based on the prompt they receive. When you open up Claude, it'll look like this with the chat bar at the bottom. If you click on the blue globe, that'll enable web browsing, allowing Claude to search the web and provide you with up-to-date answers. On the right-hand side, you can see the microphone, which allows you to speak to Claude. You click the plus sign on the left-hand side, and this will pop up. You can click on camera to open your camera and take a picture of something. You can upload a photo and have Claude analyze it, or you can upload a file and have Claude analyze that. You can also add a project. Projects allow users to give Claude centralized knowledge that it can continue to refer back to. As an example here, we're say we're working on an AI food recipe app. Once we click create project, we then have the ability to add project knowledge and custom instructions. This allows us to continue to customize our experience with Claude. You could upload a PDF that breaks down your food recipe app, then give custom instructions to Claude about how you want it to respond every single time you interact with it. You can tell it to act as something like your chief sales officer or your social media manager. After finishing your project, you can see that the name of it appears under the dropdown. This means that Claude is referring to that exact project during this chat. We've covered projects, so now let's move on to style. One of the really cool features of Claude is you can choose the style that it responds in. You can choose from normal, concise, explanatory, or formal. If you're using Claude on your laptop, you can actually create a custom response style. So if we click that button in the bottom left corner, this will pop up. You can add examples of your own writing, and that'll give Claude the ability to respond in writing that sounds just like you. You can upload text, articles, emails, whatever you want to give Claude that'll give it the best idea of how you communicate. You can also turn on extended thinking, which is Claude's reasoning capabilities. So if you turn this on, you're going to have to wait a little bit longer to get answers, but you will get better answers. Lastly, we can click on manage tools at the bottom. This is where you can control if web search, analysis, and artifacts are flipped on or off. I like to leave all of these on so Claude can use the feature if it feels the need. Now that we've covered all of the features, we can get into prompting Claude. Your output is only as good as your input, so we have to make sure that we're uploading high-quality prompts to Claude. That way we can get high-quality output in return. This is a super simple prompting framework that I've been using for a while, and I think works really well. You're going to assign a role, give context, give a command, and then tell Claude to ask us questions for more context. You can also add things like examples or potential pitfalls or things that you want Claude to avoid, but I don't think those are totally necessary. As an example here, I told Claude to act as a social media manager. I find assigning a role super helpful because it allows the AI to think from the perspective of the role you gave it. We'll say that we're the owner of a coffee shop and need a social media content calendar for next month. We're going to have Claude conduct some research on Kansas City so it's familiar. We're going to be posting on LinkedIn, TikTok, Instagram, and Facebook. And of course, we're going to finish by telling Claude it must ask us questions before generating its output. If you wanted to do examples, we could have included a successful content calendar from a month that we liked. And for the potential pitfalls or things we wanted Claude to steer clear of, we could say avoid repetitive content ideas and make sure you don't talk down on local competition. After we click submit, we are going to get Claude's questions. We're going to go ahead and answer all of these questions so Claude will have all of the context it needs to carry out the task that we gave it. After you answer all of these questions, you'll get the first draft of your output. And from there, you can refine the output with follow-up prompts, or you can copy and paste whatever it gave you into a document and make changes from there. Follow for more AI hacks.