Stop! Watch This Before You Use Agent Builder!

Everyone's talking about OpenAI's agent builder right now, like it's the golden ticket. And the truth is, it might not be the best solution for you or your business if you're looking for AI agents to handle some of the workflow for you. So, in this video, I'm going to show you when it's the perfect time to use OpenAI's agent builder. And also, at the end of this video, I'm going to show you how to build your own AI agent. And I'm also going to give you a very quick comparison between OpenAI's agent builder against NA so that you can see exactly what OpenAI has released to the public and which one is the best one for

your business. Just in case you're new to this discussion, an agent is simply an AI that does work for you. Now, in order to build an AI agent, you need to understand your work and your workflow. And you need to have a way of communicating to the AI how do you actually do your work. Now, of course, you can write this as a prompt. You can write your instructions in a step-by-step format. I do it all of the time and build AI agents inside Chat GPT. But for a lot of people, it's just easier if you can do it visually. And

this is where platforms like OpenAI's agent builder come in. If you look here on the screen, you can see the entire workflow. This is where it starts. Then it goes to a query rewrite. And you can see that there are instructions written for each node. Then it classifies the rewrite. Then there's an if else that determines what to do based on the classification and how this query is supposed to be handled. And if you're not a beginner and you're somebody who's already familiar with this conversation, then you know that agent builder is probably a response to n because they have a chokeold on the market right now when it comes to AI automation. But

let's do a very rapid fire comparison. So if you're looking for control and ownership, then NAN is going to be the platform that you want to go with because you have full ownership and control over the workflow. Whereas when you're building with agent builder, everything that you build is built on top of the OpenAI stack and any changes they make immediately impact your business and your ability to do what you've normally done. Sometimes those changes might help your business and sometimes they might completely destroy your business. Now let's look at cost predictability. When you use OpenAI's agent builder, everything that you do is metered. It's based on usage. It's based on tokens and so forth. It's almost like the light bill or the water bill at your home. It can fluctuate from month to month based on usage. Whereas with N8 and especially if you set up your own virtual private server with hosting, which I'm going to talk to you more about at the end of this video, then it's just like paying your mortgage. It's set every single month for the most part. And this next one is a dealbreaker for most people, but that's model freedom. Now, OpenAI chat GPT great model. I have no problems with it. It's absolutely wonderful. I use it every single day. But there are a lot of people who prefer different models for different reasons. But there are other less expensive options on the market to handle some really mundane tasks. But you don't really have that option per se with OpenAI unless you just choose one of the smaller models. For whatever reason, you may prefer different models for different workflows. Or maybe here's a thought, you want different models to collaborate to reach a certain end. You can't do that with agent builder, but you can do it within an end. Then there's the speed to prototype. It's so easy to build out a workflow with agent builder. And on OpenAI dev day, we actually saw them build out an agent in real time. And it took about six to seven minutes to do so and deploy it. That was pretty fast. But as fast as agent builder is, this is another major problem and something I think is an oversight on the OpenAI dev team, but you don't have the ability to share your workflows inside of N. If I click on these three dots right here, I can come down to import from file and I can upload a JSON file that will automatically install a completed workflow right here. And depending on the workflow, I may need to authorize different applications, but for the most part, all of the work is done. That means that not only can I share this workflow with people, I can sell it to other people, or I can have an AI write and create all of my workflows for me. But that's not even really an issue because if I come over here and I click on templates, there is a community that has over 6,000 workflows that you can use for absolutely free if you use NN. But over here in agent builder, you're absolutely on your own. I mean, OpenAI does give you six templates to start off with, but for the most part, you have to build from scratch. Finally, the last thing, and this is kind of like me just plugging myself right here, but there's no way for you to tie this agent that you build inside agent builder back to your chat GPT account. But inside N, I'm able to build my own long-term memory protocol with only two simple nodes. And you can see here that it's executed successfully two times, which I'll show you in just a second. So, here's the bottom line. Open AAI wants to be the apple of the AI industry, and they're creating their wall garden. If you're invested in their large language models and the quality of outputs you get from chat GPT, then Agent Builder is going to be something that you're seriously considering. But if you want flexibility with your business, and if you like the fact that there are thousands of templates already designed that you can simply download and upload to quickly install in your business, then NAN is going to be something that you're looking for. And now let me show you how

to build your own AI agent that has a perfect memory about your business. And personally, this is my favorite reason for using N because it allows me to connect Chat GPT to an unlimited memory of my data, my experiences, my frameworks, and so forth, whatever I choose. So that my chat GPT conversations are hyperpersonalized in a way that other people's aren't. And I don't have to wait another five or 10 years for OpenAI to give us unlimited memory. If they ever give us unlimited memory, I already have it. So, quick pricing heads up. And depending on when you're watching this video, the early Black Friday pricing is going to be $6. 49 a month. So, you can get your own virtual private server for 2 years for less than 150 bucks. If you use the link in the description and you use code Corey at checkout for an additional 10% off. If you're watching this outside of the early Black Friday time frame, then prices may be different, but you can still get 10% off by using my name, Corey, as a coupon at checkout for 10% off. So, make sure you use the link pinned at the top of the description if you decide to set up your own hosting or VPS long-term memory protocol after you watch this short demo. And if you like what you see and you want to set up your own long-term memory protocol, then send me a screenshot of your receipt and I'll send you over my long-term memory protocol guide that's going to give you all of the commands so that all you have to do is copy and paste and it's going to walk you through setting everything up. And I'm also going to send you a copy of the JSON file so that all you have to do is upload it to NAN, change the form options like I'm going to show you. And you're going to have your own long-term memory protocol set up so that ChatGpt has an infinite understanding and unlimited knowledge about your business or about your life or anything else you want to share with ChatGpt. Now, let me

show you how to build your own AI agent inside of your ChatGpt account that knows and understands everything about you. The first thing you want to do is click the link in the description and you're going to come to this landing page. And then you want to choose a plan. The most popular plan that's probably going to cover everything you need is this KVM2 plan. Depending on when you're watching this video, it's $6. 49 or $6. 99. So if it's $649, that means you get 64% off the price. And if you want an additional 10% off, then all you have to do is choose this plan. And when you get to checkout, right here where it says have a coupon code, type in my name, click apply, and it's going to take an additional 10% off. And now it's only $150 for 2 years for your own virtual private server. With this plan, you get unlimited workflows, unlimited concurrent executions, you get community nodes, you get n with cumo, you get 100 plus pre-made workflows, oneclick n installation. You simply choose it when you first purchase your account and it's going to walk you through everything and set it up for you. You get Hostinger API NAN community node the AI assistant powered by MCP and you get one free. cloud domain for one year. But once you have your account set up, your dashboard is going to look something like this. And if it doesn't, just make sure you click on the overview button right here. And then I want you to click on browser terminal. This is going to

open a command line interface that allows you to manage your virtual private server. And all you want to paste in is this very long and complicated command that I'm going to give you and it's going to automatically set up everything for your NAN account and for your memory system. And just to show you that this works, I wipe my server clean. So, I'm starting from scratch. And as soon as this finishes extracting, we're going to go back to that page, refresh it, and you're going to see that it instantly loads. And so, now let's come back over here. Let's refresh this page. And now we see that it loads, but we don't have an account set up. So now I'm just going to go through the process. It already knows everything. So I'm just going to click and add it. Going to click next. Get started. Skip. And as you can see, I have no workflows because once again, I wipe my server clean just for you. So I'm going to click on create workflow. the three dots and then I'm going to import from file. Then all I have to do is upload the file for my long-term memory protocol and automatically it sets up this very simple workflow where there is a form where I can add all of my memories and it automatically writes it to a file. So the first thing we want to do is open this form and down here where you see the field names there's a drop down then there's options. These are going to be the names of my projects inside of my chat GPT account. So, you can see that I have several different projects and different names. Hundred million dollar offers, Bible study, workflow, agent builder, business brain, and so forth. No matter what the name is, just make certain that you use all lowercases and no spaces. Use a simple dash to separate words. So, I'm going to delete these two because I'm not going to be needing these anymore. And then I'm going to add one. Then, I'm going to go back to the canvas and I'm going to activate this workflow. And now that it's active, I'm going to click on this form again. I'm going to go to the production URL. It's copied. And now we have our form set up. So if I click this, it's going to show me Bible study and $100 million offers. Depending on where I get my memories from, I'll choose my project. I'll paste my memories and then I'll submit. So now let's go to chat GPT and let's get some memories. What I've done now is pasted a memory extraction prompt that I created that's going to go through this conversation and distract as many meaningful memories as it possibly can. And you can run this prompt for any conversation you have with chat GPT. Personally, I prefer to do so inside of projects because it helps me keep things organized. Now, let me explain the structure of these memories so you can understand that this isn't just something we're throwing together and that there is an actual method to the madness. If you look right here, everything in this red is a category and the green is the actual data. So, there's a memory ID and this is the memory ID right here. This helps us actually catalog each individual memory so that the AI knows where it's pulling the memory from. This is going to reduce hallucination. Then there is a date time. Then there's a memory type. Is it a definition, an instruction, a fact? And there are other different types of memories that I store. But in certain conversations is going to be the same thing over and over. And then we have tags, memory builder, ro system prompt. Then we have the content. So it's going to capture the exact content from the conversation. So all we want to do is copy this code. Return to our nan form. Click here. Choose the project. Click inside memory and paste. And then we want to click submit. And our response has been recorded. And now that we're back at NAN, if you come to the top, you can see that there's an editor, executions, and evaluations. Let's click on executions. And we can see that it was just executed on October 12th at 4:51 p. m. and 22 seconds. Succeeded in 26 milliseconds. That's just how fast it recorded those memories. But that's not really why we're creating this agent. So, let's look at a more practical way of actually using this. Here's the space that was created for my file when we first set up the oneshot command. It's a subdomain memory server domain then 100 million offers-memory. json l and based on whatever your projects are, it's going to change this. So, it doesn't matter if your project names are different from mine. You just go inside of the NAN form, set up your projects the way I told you to, and it's going to foul your memories perfectly. And all you simply do is use this same format. But don't worry, if you sign up for this, I'll send you a free guide that walks you through everything and answers all of your questions. And if you have any problems, you can shoot me an email and I help you get it set up. So now, let's go back to ChatGpt, get some more memories, and then show you that this actually updates. Just to show you that this is working, we're going to grab some more memories and go back to our form. We're going to choose the same project. We're going to paste in our new memories. We're going to click submit. It's been recorded. We're going to go back to N. We're going to refresh this executions page because now there should be two executions. We see that there are two executions. And now let's look at the file or the space where all of this is being stored. And now I'm going to refresh this page. And when I do, I want you to watch how it doubles in size because those are the new memories that have been added. And now you can see them. And so you can see that all of these memories are stored here. And so all I have to do is copy the address to this file, open my chat GPT account, go to a project, edit the instructions, and then place a simple line here that says at the beginning of every conversation inside of this project, visit this file, read the memories, look for relevant memories based on our conversation, and then respond to whatever my query is. Or you can delay it, which is something I prefer to do, and ask chat GPT to wait for three or four queries to find out what the conversation is about, and then to visit the link, look for other memories, and contextualize your response based on what you found there that's related to what we're discussing. Now, like I said, I personally like to use projects to separate things that I'm discussing with chat GPT, and I like to use project memory only. But maybe you use your account differently. And if that's the case, you can go to your profile name in the bottom left corner. Click on personalization. And then right here, there's a space for you to add custom instructions. You can tell chat GPT at the beginning of every conversation to visit this link to learn more about you. Or you can add it here in the more about you section. and you shouldn't have any problems with chat GPT visiting this link, reading the information, and providing more contextually relevant responses to you. But you will need a tagging system to separate all of those ideas. And that's why I use projects because it just makes it so much simpler to keep everything separate so that chat GPT isn't bringing irrelevant information into the conversation, which is something that it's done in the past. So when you

factor in the idea of being able to store as much information, memories, and data as you want in your own virtual private space and give chat GPT access to it, as well as the ability to structure together several different prompts, very detailed and lengthy stepby-step prompts just like an agent, then you realize that you're able to create your own agent that has a perfect memory about your business. And this is valuable, especially if you're a soloreneur. So, if you got value out of this video, make sure you hit the like button, subscribe to the channel, and as always, take care, have a great day, and I'll see you in this next video where I show you how to turn your chat GPT projects into ChatGpt agents.