Transcription
If you're currently active on YouTube, you'll hear the same thing everywhere. ChatGBT is back, Cloud is struggling, and the hype might be over. I've been observing this for weeks and have come across two things that I'd like to share with you today. One of them is a trick that allows you to completely disregard which AI app is currently leading, making you completely autonomous from trends. But first, a brief context. So, factor number 1: A few weeks ago, many Germans hadn't really heard of the company Entropic at all. Then Claud became number 1 in the App Store. Many users switched from Chat GPT, and Claud subsequently scored one hat-trick after another, adding a little extra everywhere. So many cool new features were released, for example, capabilities that you simply retrieve via command. Cowork, with which Cloud independently processes tasks on your computer. Cloud Code, which independently programs things for you, and you can even control everything from your phone. Wow. This finally forced OpenAI to deliver, and they have released a few things in recent weeks that have really, really hit the mark. GPT 5.5 as a new model is significantly better and is on par with Clord Opus. Images 2 is a real leap forward in image generation, and especially text is now displayed frighteningly cleanly. And precisely that was always the problem across the industry before. And then came Codex. This is a standalone app that exists autonomously from the classic Chat GPT app and is essentially OpenAI's answer to Cloud Code and Cloud Cowork. And many advanced users have now become fans of Codex because the app is really, really good. That's why so many YouTubers are currently jumping on this hype train, but one must remain nuanced, and there are also other factors that have nothing to do with it at all. Look, Claud was, in a way, a bit of a victim of its own success and therefore quickly hit its capacity limits. Everyone knows these annoying limits that pop up in the middle of work. The app sometimes says: "Hey, the quota is exhausted, come back in four hours." which is super, super annoying. But why is all of this happening? So, it's not artificial scarcity or even greed, but rather something else. Look, Entropic is the significantly smaller company, and the demand is currently greater than what the data centers can provide. So, imagine you open a pizzeria, it goes viral overnight, and suddenly half the city center is at your door. But you can't bake pizzas that quickly, you can't keep up, and in the end, a few people go home hungry and are probably a bit frustrated. Okay, exactly the same thing is happening now. Shopic is trying to fix it. Elon Musk has apparently reacted and lent data center capacity. But data centers aren't scaled up just like that in two weeks. So, before we get to the solution, I wanted to talk about probably the most important point in the entire video. What the story shows is that the AI race isn't decided in a single month. What's ahead today can easily fall back next week and potentially lead again the week after. Of course, you'll always have a favorite, and that's perfectly okay, but betting everything on one app, having all your workflows there, will probably backfire within the next two years. And knowing the other apps at least a little bit will become increasingly important. And that's why I want to report on all the important apps on this channel and not limit myself to just a few. Okay? And honestly, regardless of performance, features, and limits, I still really like Cloud because the way it works feels like an AI that truly understands me and consistently delivers good results. But yes, Cloud has changed how we think about AI and how we work with AI, and introduced new ideas that have completely shifted our mindset regarding AI workflows. And it will also influence other apps in their DNA. Which brings us to the next point: hybrid setups. And this is where it gets concrete. What many people are doing right now, especially in the coding scene, are hybrid setups, and there's a very popular example. Many use the interface of Cloud Code, but the tokens from Codex run in the background. How does this work? You get the so-called API key, and for those who don't know the term, an API is essentially an interface, and an API key is like a digital password or a key that an application uses to identify itself to another software. This means you generate a key from the model of your choice, manually insert it into Cloud Code, and then the engine from OpenAI runs in Tropic's cockpit. In terms of technical effort, I wouldn't call it rocket science, but you should be a little tech-savvy. And if that's too fiddly, there's also a much simpler option, which leads us to the next point: my secret tip for today's video. I received a tip from a user in the comments, and the app is called Mammoth AI, so "Mammoth" in German. And at its core, it's an aggregator that runs APIs from all the major models in the background, so Cloud, Chat GPT, Gemini, whatever you want. But you get a unified interface and a token quota that you can distribute freely among all models. So, if, for example, Cloud is the best and leads in a given month, you simply activate it. If another app performs better the following month, you can easily switch. I think it's a super cool concept, and it's also a bit cheaper than paying for consumer versions individually. And for someone who is open to working with multiple models, this is probably the cleanest and easiest solution right now. If you're interested in the details, please write it in the comments. Maybe I'll make a separate video about it where I test this tool. And finally, if you take one thing away from this video, don't become dependent on a single app, no matter which one. Always have a plan B in reserve, and I'm really interested to know how you're doing with this. Did you also recently switch to Cloud? Are the limits annoying you so much that you're thinking about switching back? And finally, what's your knowledge level? So, would you classify yourself as a professional, or have you only been using AI recently? Please write it in the comments. As always, thank you very much for watching, please subscribe, and see you next time.