Most People Have No Idea ChatGPT/Claude Projects Can Do This

If you're still creating and using custom GPTs to handle your workflows, then you're behind. The top 1% of Power users are actually using chains of prompts or prompt chaining to get this work done in a more efficient and powerful way. And in this video, I'm going to show you how to create your own agentic prompts that have context, predictable outcomes, and also memory, which every Power user knows is one of the most important and most desired features of any AI. And we're going to do this using chat GPT projects and you could do this with Cloud projects as well. And you can also use this method with Gemini projects when they release soon if the rumors are true. Custom GPTs

are great, but the biggest problem with them is that they're severely limited in their scope. You only get 8,000 characters to tell it what to do. And while that's enough to actually create a prompt chain, it's not nearly enough for most power users. and you end up truncating your prompt, deleting things, taking out whole section and trying to make it fit inside of this very narrow window or you end up creating three to five different custom GPTs just to handle one workflow. Because the truth of the matter is real work doesn't happen with AI with a single prompt. It's a multi-layer and a very complex process. many times when we're doing our human work and even when we're working with AI trying to replicate that workflow. Another problem that's not so obvious with custom GPTs is the fact that you're not able to organize your chats with custom GPTs in any way. Whether it's your custom GPT or one that someone else built, those chats remain in the sidebar and you have to use the search function to find that earlier conversation. And we all know how big of a headache that can be. And the solution is simple. We need a single workspace where we can carry out a multi-step in-depth workflow that can carry on for days or even weeks and maintain memory and create a consistent output over and over again without fail. And

that's what I'm about to show you how to do with your Chat GPT account. Inside of your Chat GPT account, on the lefth hand side, come down to where it says new project. And we're going to create a name this project business brain. We're going to select an appropriate emoji for this project. And then we're going to change the color to green to represent money. But before we click create project, I want to show you one more thing which is another benefit of projects over custom GPTs. And don't worry, I'm going to show you how projects replace custom GPTs in just a moment. But if you click this cog

right here, there are two types of memory that you can activate with your chat GPT project. The first one is default. This simply means that if you're inside of a project, chat GPT has the ability to access information from outside of this project. And also, if you're having a conversation outside of a project, then you're able to tell chat GPT to access memories inside of this project. Now, the problem with that is this. This is in essence going to make this project simply be a folder for conversations because these conversations are going to access the same reference past chat history that the other conversations do. But the second memory option is the one that we want to choose. We want this space to be a private and a focused space within our own account so that whatever memories chat GPT uses or references, we know that it's reaching back as far as it can within a single project. Theoretically, this should extend the reach of its memory and increase the performance and accuracy of recall and everything else that we want when we're using chat GPT because it's not reaching back over irrelevant conversations and every conversation inside of a project should most likely be relevant. Now, we're going to click create project. And now

there are two things that you need to know about any project. Number one is that you are able to upload 10 files with a ChatGpt Plus account. And I believe that the number is unlimited for Pro. I'm not sure, but I do know that at a certain point, no matter which account you have, performance begins to fall off when the files sizes get too big. And this is the exact moment where chat GPT projects almost make custom GPTs obsolete. There's still a very limited number of use cases where they're very practical. For instance, if you create a custom GPT and then you want to give access to this uh workflow to a certain number of people without necessarily passing off your files that you use to create it, then you would use a custom GPT probably. So right now I have seven files uploaded. And what makes these files different from how people typically use chat GPT projects is that these are not knowledge graphs. These are not files that contain data or anything from chat GPT to draw from. Each of these files is a set of instructions that could stand alone as its own custom GPT. And in the past, they have stood alone as their own custom GPTs. And each of these files most likely goes far beyond the 8,000 character limit. I haven't counted them, but they're very detailed. They're very long, and I'm pretty sure they probably go past the 8,000 character limit. So that in essence, we probably have somewhere over 60,000 characters as far as the instructions contained in these files. And that brings me to the

instructions because this is where projects really shine. We're able to upload several different files which each contain several thousand characters of custom instructions with multi-steps in each of them. Instead of using our custom instructions to say how chat GPT can best help with this project or telling chat GPT to focus on certain topics or ask it to use a certain tone or format for responses, we give it a very detailed set of instructions that cause chat GPT to function like a router. But let me give you a brief overview of this router. So the purpose is a deterministic router only operating standard for this project. It opens a fixed list of prompt files in order, emits each file verbatim into chat, executes the files own instructions to completion with approvals, then advances. So that was just an overview of the instructions, but everything else contained there is basically forcing chat GBT to open one file at a time in a very particular order and then to print that file in the chat and execute it. Now the reason I do this is because of the old mantra trust but verify. So it's easy for chat GPT to say that okay I opened it up but then not to actually execute it. And so I want to make certain that it actually does what I'm requesting now that it's outputed the entire file. Now what I would simply say is run the prompt. And so now it's going to give me the first question like it's supposed to. Now chat GPT is going to take me through this very detailed process of intaking and gathering information for me for this very particular workflow that I've created which used to be seven different custom GPTs but now they live in one single conversation and they work together handing off information one to the other without switching tabs and I can always come back and have a conversation about what we created decisions we made about my business etc because it's all there in a single place. And that's not the only benefit of replacing custom GPTs with chat GPT projects in this particular manner. Now, not only can I

do everything that I just showed you, but I'm also able to perform deep research within the workflow of a custom GPT's instructions because I'm no longer using custom GPTs. I've just taken the instructions and placed them inside of a project and now I can run deep research. So, if I have a marketing custom GPT, I can now use the benefits of deep research, even agent mode and all of the other tools that you typically can't use with custom GPTs. Yes, you have web search. Yes, you have data analysis, but to my knowledge, you don't have all of the other tools that you do with projects. You also have the ability to virompt, as I call it. I like to walk around sometimes and use chat GPT like Tony starts talking to Jarvis. And so now I can open chat GPT advanced voice mode, go to this project, start a conversation, and I can just vibe and talk about whatever this process is and work through the workflow, create the documents, create the outputs, and finish a very complex task without sitting at my laptop because of the way I've structured my chat GPT projects. Also, because this isn't a project, I can share this conversation and make it public. So if this is a project where I'm creating something that can be beneficial to everyone, I have the option to create a public share link and share that so that other people can walk through this as well and see the outputs. And we already touched on this before as far as the memory. But another benefit of chat GPT projects is my own personal long-term memory protocol that I use to store my personal memories on my own virtual private server. and I did a video about it that I'll put a link to

right here. So, should you stop using custom GPTs? I think so. The benefits of using them are very small because everything you can do with them, projects. Projects are where a gentic prompts come alive. There's just so much that you can do there. And then when you factor in the memory choices that you have and the ability to introduce your own long-term memory protocol to that project, I think it's a very clear choice that you should be using chat GPT projects over anything and everything right now inside of your ChatGpt account. If you got value out of the video, make sure you hit the like button and subscribe to the channel. And also, if you have the ability, hit the hype button so that this video can reach more people on YouTube. And as always, take care, have a good day, and I'll see you in the next