Transcription
Your Cloud limit was used up at 9:41 AM. What can you still do now? Nothing. And the worst part is, you didn't work too much. You simply used Cloud incorrectly. Today, I'll show you six habits that will always change that. And I'd say we're starting right away. Let's go. My name is Lisa, and this channel is all about AI and how you can use it in your business and everyday life. Do you know this? You sit at your desk in the morning, are about to start, and after a few hours, the message appears: "The limit with Cloud has been reached." And you think to yourself, how can that already be? The problem isn't that you're using Cloud too much. The problem is rather how Cloud works in the background. Most people treat Cloud like a search engine. Quickly type something in, get an answer, and then the next question. But Cloud works completely differently, also differently than ChatGPT, because with Cloud, your message or your messages don't count. With Cloud, tokens count. What are tokens? Exactly, we'll look at that now too. And tokens burn faster than you think, often without you even noticing. Every follow-up message, every newly uploaded PDF, every long chat, everything costs, and it happens very, very often unconsciously. And when the limit is gone, then it's gone. Therefore, I'd say let's look at the six tips now on how to avoid this from now on and never reach the token limit again. Tip number 1: Stop sending follow-ups. Edit instead. What do I mean by that? This is probably the habit with the biggest impact and also the easiest. Imagine you send Cloud a prompt. The answer is okay, but not quite what you wanted. What do you do then? Probably you send a follow-up prompt, meaning a follow-up message. Can you make that shorter? Or that wasn't quite what I meant. And the problem is, every new message makes Cloud re-read the entire conversation from the beginning. Everything. So message 30 costs 31 times more than message 1, simply because a lot has happened in between. Not because the question is more complex, but because Cloud carries so much more history along, meaning what has happened in the meantime. Therefore, my tip: click on the edit icon, you always see it below your prompt, and fix the whole thing. So, change the prompt again. So, if you now notice, for example, that Cloud has given you a wrong answer or not the answer you want, then it's usually because your prompt wasn't precise enough. So, regenerate it or edit it here again. Over several rounds, this habit alone saves up to 80% of your tokens, and that's something to be proud of, of course. Tip number 2: Reset the chat every 15 to 20 messages. And here it definitely gets interesting, and here I actually find myself more often not always adhering to this tip. So, we looked at tip number 1, and this tip is also related to it. So, even if you edit the prompt instead of sending it as a follow-up, at some point your chat has simply become super, super long. Of course, it has the advantage that a lot of context is already present, meaning Cloud already knows what it's all about, but long chats also burn a lot, a lot, a lot of tokens for context that nobody needs anymore. So, not for the day you sent the request. Tip: every 15 to 20 messages, you can have Cloud summarize the chat. Then you copy that and paste it into the new chat. Then you have a fresh start, zero waste, and it takes 20 to 30 seconds. You can, of course, also do it so that if the topics are already separated from each other, you then start a new chat. So, assuming you've now drafted an email and then have a new email or something completely different, then start with a new chat instead of doing everything again in the old chat. Tip number 3: Send all questions in one message. This is actually something that is often recommended, that you ask things step by step. But with Cloud, it's the case that with three separate messages, it means three full context loads. So, it's loaded three times. Cloud re-reads everything with each message, and that three times. My tip therefore: summarize this text, extract the main points, and suggest three headings for me. One message, one context load. The answers will often even be better. By the way, a small note, as soon as this video comes out, definitely check out my Instagram profile too, because I've summarized it all there again, and you'll also find it here again as a PDF for free download. But let's continue with tip number 4: Load recurring files into projects. What do I mean by that? I've brought an example here. You're working with a briefing or a file that you always need, and you upload it anew each time. What happens then? Cloud uses tokens again each time because it reads through this document again. The briefing, which costs, for example, x tokens, you pay for it not just once, but every time again. Therefore, my tip: the project function. In the sidebar, there is the project function. There, as the name suggests, you can start projects. And here you can also include or have documents read in. Upload it once, and every chat in this project will have this information. That means you don't have to upload it again each time, but it's stored here. Tip number 5: Use the right model for the right task. And this is also something very, very, very important again. I can also make a separate video about this if you're interested. So, with Cloud, you have different models to choose from. And this is a decision with probably the biggest impact, and most people don't make it actively. There are cheaper models like, for example, Haiku. It's perfect for grammar checks, quick questions, simple formatting. Then there's Sonell, and that's for real work, I'll call it that, for analyses, for strategy. And there's Opus, and that's for really, I'll just say, heavy stuff. My tip: if you perform simpler tasks with a heavier model or with a higher model like Sonell, you pay three to five times more than necessary. And if you never change that, then it's of course a daily drain. Before you start a task, think about which model you really need here. As I said, you can find more information in my newsletter. We summarize everything there in detail, and you'll have it in black and white there again. Tip number 6, and this is a bit reminiscent of tip number 4, namely Memory and User Preferences. What is meant by that? So, Memory is the memory, and User is the user. User settings, I'll call it that. And this is the tip with the lowest effort. And if you implement only one thing here, then I would recommend you implement it after this video. So, after you've liked the video and ideally also commented, of course. So, the effort is very, very low, as I said, but very effective. So, every new chat without saved context burns three to five messages. Just by explaining to Cloud who you are, what your business is, what tone or in what tone you write, and so on and so forth. Because these are usually things that remain. You [clears throat] won't change your job, your name, and whatever else there is, or even your whole personality per chat. And these are things that you can store in the settings. I'll show you where it all works. So, this is definitely a setup that I highly recommend. So, the tip is: go to settings, then you have Memory and User Preferences here, and here you can simply write down everything about who you are, your target audience, your style. Cloud will automatically add it to every new chat from now on. When I work with freelancers and companies on an AI Efficiency Audit, we almost always look at this first. Not because the tools are bad, but because most people use Cloud like a search field. Ask in, then get the answer out. Next question. And that is, as we've learned here, not 100% ideal. The difference between using Cloud and building a Cloud system, that's exactly what it's all about. And these six habits are the first step towards that. So, let's summarize everything again. Edit the prompt instead of sending follow-ups. Especially if you notice that things are going in the wrong direction. Then tip number 2: reset the chat every 15 to 20 messages, or batch the questions. Then load files into projects, if they are files you work with frequently. Choose the right model and set up the Memory, meaning the information about you or what is important. With these six habits, you will definitely save quite a few tokens. I've summarized it all again as a PDF. You'll find it down in the info box. Otherwise, also check out Instagram, you'll find it there too, and then I'll send it to you for free. Otherwise, I'm currently working on the AI Community, I think I've already mentioned that, and you'll find all the information there too, down in the info box. I hope this video has helped you to save your tokens in the future, so that you don't have to wait until 9:50 to be able to send new requests again. And here I'll link other videos relevant to the topic. I wish you a nice day, I definitely wish you a lot of fun with Cloud, and I look forward to seeing you again in the next video. Best regards and see you then.