305 | Mastering AI Automation: Custom GPTs to Agents with Isar Meitis

0:02Hello, and welcome to the Leveraging AI podcast, the podcast that shares practical, ethical ways to leverage AI to improve efficiency, grow your business, and advance your career. This is Isar Meitis, your host, and I have an interesting episode for you today. There have been growing rumors in the past few months that OpenAI will cancel custom GPTs. Now, those of you who don't know what custom GPTs are, well, then you're not missing much. But if you did not use any custom GPTs so far, then you were missing maybe the most capable automation tool that AI gave us until the existence of scales and agents.

0:39And many, many, many people, myself included, has multiple custom GPTs, which they were regularly using to run multiple things in their business. I switched completely to scales and agents, but I know many people who have not. Now, I haven't seen any real information from OpenAI about custom GPTs going away. But if you're following what's happening on X or on Reddit, then you would see that there are many conversations talking about custom GPTs going away. And if you are somebody like me, who until not too long ago, were running a lot of aspects of my business through custom GPTs, then it is something that you need to be aware of and start preparing for.

1:19Either way, whether you have built custom GPTs that you're using regularly or you don't, and you want to know how to build different kinds of automations with AI that can run more or less everything in your business, this would be a great episode for you to learn the basic concepts because they're the same across the board. So let's get started. I'm going to be sharing my screen, but those of you who are just listening and not watching this, I will explain everything that is on the screen. So if you are driving or walking your dog or doing the dishes or something like that, yoga and you can't watch the screen, then that's perfectly fine.

1:51But if you do want to watch the screen and you are able to do this now, you can either do this on Spotify or sharing the videos on Spotify, as well as on our YouTube channel. And there's a link to the YouTube channel on the show notes. But let's get started. So there are several different levels on the ways you can work with AI, whether it's ChatGPT or any of the others. The very basic one is chat, right? You can just go to the chat window of Gemini or Claude or ChatGPT, etc., and just chat with it, and then you give it instructions or questions, and then it does what it needs to do. You are heavily involved in the process.

2:26Option number two is to create custom GPTs or projects. These are a consistent setup with a bunch of instructions and additional supporting information that becomes reusable and allows the AI to follow an existing set of instructions time and time again without you having to give it any additional instructions. We're going to see examples shortly. The third level, which is the newest of the three, and if you want from an evolution perspective, we had custom GPTs first. That was something that OpenAI invented. Then we got projects from Anthropic, which were then copied by OpenAI to have projects inside of OpenAI.

3:00And then we received gems from Gemini, which is a very similar concept. And then we got scales and agents and so on. So scales was long after we have projects, but Scales allows you to make it portable and reusable automation components, building blocks that you can use in multiple places. You can use Scales in other platforms. So if you are using, as an example, ChatGPT in Excel, there is an extension for that, or ChatGPT in PowerPoint, there's an extension for that. You can take your Scales with you. So if you have created Scales for specific processes in financial analysis, you can bring them from ChatGPT into ChatGPT in Excel.

3:41And if you have a specific branding or process or style that you're using and you created skills for that, you can bring it into PowerPoint as well. And then on top of that, there are obviously agents where you can build agents. And to be fair, I'll do a little distinction on what agents are and what they're not because there's a lot of chatter and everybody calls everything agents right now when it's not really the case. So agents are employees, right? It's an entity that can do a full, complete workflow and not just one simple specific task again and again. And they can do it end to end and they can do it in a consistent way while taking into account information they have and they do not have.

4:22I'm not going to cover this today. We're going to stop at scales because I want to show you the basic building blocks that you can use. Now, a scale is basically a very simple agent if you want, and that's why a lot of people call them the same. but an agent could be a lot more sophisticated, have access to more tools, be connected to more systems than just a simple skill. So these are the different options that we have in front of us today. Now, before we dive into the different options, I want to kind of show you the process that I went through, the involvement that I went through, through this process, through the lens of writing proposals.

4:55So today I use what some of you see on the screen, which is a very sophisticated proposal pipeline agent that has multiple skills, that is connected to multiple systems, in my ecosystem, and that creates amazing proposals every time I finish a conversation that is relevant in which a proposal was requested or discussed. It does all of that automatically. It updates my CRM. It updates my Google Drive. It updates my outbox with a draft email while attaching the proposal. It does research in the back end on the customer. It does a lot of other great stuff. But this is not how I started. How I started was a custom GPT.

5:28So now let's look at what a custom GPT is. If I go to chat GPT, initially, and this is one of the reasons why

5:35Isar Meitis:people thinking that custom GPTs are going away, custom GPTs lived on the left menu right above where the conversations are. And now you can see it's not there. Like if you look for custom GPTs, you won't see them. So where are there? If you look on the left side menu, you have where you have new chat and search chats and library and so on. There's the three dots that says more. If you click on more, you will see GPTs. And if you click on that, you will see all your GPTs over there. So they made it less intuitive to get to or create. And if you don't know they exist, they just will disappear on you and you won't even know that they're there.

6:09Isar Meitis:That gives you a hint that this is not their focus and not the focus of their product. Now, how does this work? You can see that this is called Multiply AI Training and Education Proposal. That's the name of this custom GPT. All I have to do is I have to go to add a file and drag in a file of a conversation I had with a potential client. And once it loads the file, all I have to do is click go. I won't type anything. I won't say anything. I literally just upload the transcript of a conversation that I had with a client that requested a proposal. And it is going to work and it is going to write a great proposal.

6:43Isar Meitis:And you can see it's already started writing and it has an introduction and the objectives and why AI training is urgent and why to work with multiply, which is my company where I provide training and education and training formats recommended for this particular company based on their needs and so on and so forth and pricing and the whole thing. It writes about a 8 to 12 page proposal, well-structured, and I'm winning a lot of business, so these proposals actually work pretty well. Now, this is how I started. It worked very, very well. When projects started arriving, I switched it to projects.

7:14Isar Meitis:And we're going to talk about later on why, but at first I want to show you just so that you see the different options. So now if we go in this particular case, this is a Claude project, but inside the Claude project, what you can see is the same kind of thing. I uploaded the transcript. I asked, please create a proposal based on my brand guidelines, and it created the proposal. One thing you can see immediately inside the project that we're seeing right now is that it has my logo on top. It is using my brand colors. It is using my fonts. It is from a structure perspective, has a much better structure from a document perspective.

7:46Isar Meitis:The benefit of doing this in the old school way without this is that I can edit the file. Here I cannot edit the file because it is a clawed artifact, which is not editable, which is really annoying, but that's the case. But it does come as a fully branded output, which now you can do with skills inside of custom GPTs as well. So what we have right now is we saw that I can do this in a custom GPT. I can do the same exact thing in a project, whether a project in ChatGPT or a project in Cloud or a gem in Gemini, all of them will do exactly the same thing and they will write solid proposals. And then like I said, then I switched to the full skill-based agentic pipeline that has all the other functions as well.

8:32Isar Meitis:So now I do not use the custom GPT or the projects, but it doesn't mean they're not good. It doesn't mean that I couldn't use them. It just saves me more time and does more things in an automated way. And so it saves me more time while creating a better output. And so this is why I'm doing this. But what I want to show you today is different options so you can pick the one that is best for you. So first of all, let's talk about what they are. All of these things, custom GPT's projects and skills have the same exact DNA. And that same exact DNA is they have two components. They have a set of instructions that tell them what to do.

9:06Isar Meitis:And there is context, meaning knowledge files and additional information that the AI may need in order to create the output that it needs to create. So in the proposal example, this could be a proposal draft, or if you want a template that it can use. It could include a winning proposal or a few that has won me business to show it what good looks like. It could include descriptions of the services that I provide or the products that I sell so the AI understands what it can pick from when it writes the proposal. It could include links to my website or testimonial websites or testimonials on LinkedIn that it can pull from in order to use when it's writing proposals.

9:46Isar Meitis:So all of these things are included as reference material, as context for each and every one of the three. It doesn't matter if it's a custom GPT, a project or a skill. So they're very, very similar. And that's why when you think about, let's say you do have 20 or 30 custom GPTs and you're afraid maybe OpenAI will take them away, converting them into projects or skills is a very small effort because they use exactly the same things. Before we dive further, let's talk about instructions. AI reads instructions from multiple places every single time you talk to it, especially in these kind of places.

10:21Isar Meitis:So first of all, there is the system-wide, if you want the account-wide, always-on instructions. In Claude, you can change your Claude.md, which is the file that Claude reads at the beginning of every single conversation, and you can tell it exactly how to work with you across the board in every single interaction it has with you. In ChatGPT, there is the custom instructions that live inside the settings on the bottom left corner, and you can go and tell OpenAI how to work with you, again, on every single conversation. Then inside a project, if you created a project, there are project level instructions.

10:54Isar Meitis:Same thing with a custom GPT. So you can give it instructions. And again, I'll show you in a minute where exactly you do this in the system itself. Then there in the scale, there is scale details, which is where it writes the instructions for the scale. And there are two levels of memory. In addition, one of them is the general memory, which remember things about you. And you probably noticed that Chachapiti or Claude already knows stuff about you. It knows where you work. It knows what you do. It knows your hobbies. It knows everything you talk to it about, but it's not at a very granular level.

11:24Isar Meitis:On the granular level, it is connected to the specific project. So every project is a little bubble of context that remembers things about that particular project. And a project could be, well, a project, that kind of makes sense, but it also could be a specific client. So you can keep a specific client information in a project, have a separate project for every client, and have all your conversations about that client in that project. And the project is going to remember more and more details about that particular client, about the people who work there about proposal you sent them and so on and so forth.

11:54Isar Meitis:And so there's these two levels of memory. There's the general memory and the project specific memory. So all of these things are places that the AI will reach out to depending on what kind of conversation you're having in order to get additional information to be used. Now, what I want you to remember that applies to each and every one of those things here is that AI is an intern. It's not just an intern. It is the best intern on the planet, right? It will do amazing things. But just like you won't bring an intern into your room and say, hey, I want you to write me this report, or I want you to create this proposal, or I want you to create a summary of one, two, and three, because it will probably fail.

12:33Isar Meitis:And if it will fail, you will know it is your fault, because you didn't tell the intern exactly what to do. And the intern doesn't know you, it doesn't know the company, it does not know the industry, he or she is an intern. And AI is exactly the same thing. So the instructions you're going to give it, regardless whether you give them in a custom GPT or a project or a skill, etc., are the SOP, the standard operating procedure on how to do a specific process. The knowledge files that you're going to give it is the binder you're going to give the intern in order to know everything they need to know.

13:03Isar Meitis:So when you sit with the intern, you're going to show them, okay, here are the previous proposal we gave to this client. Here's information you can find about this client. Here's where on SharePoint, you can find information about the services that we provide. Here on these Excel files, you can run the pricing and get to the, like, that's what you're going to do. And you need to do the same thing for the AI. And then it's going to do an amazing, amazing work for you every single time. Now, the project, if you want a Chachupiti or a club project, is a dedicated workstation, right? It's the table that has the computer and all the files and everything you need in order to get access to the right information.

13:38Isar Meitis:And the skill, if you want, is a laminated SOP card. It's something you can take with you if you want a suitcase with everything you need so you can go to other offices and do the work over there. And you'll see why I'm saying that once we talk more about what a skill is. But it is basically a project or a custom GPT that is fully portable and can work from everywhere without going specifically to that folder. So let's dive a little deeper to the first two options of a custom GPT versus a project. And again, you'll see that they're very, very similar and that you can switch between the two very, very quickly.

14:14Isar Meitis:So if custom GPTs will go away, your first immediate line of defense is switching your custom GPTs into projects. And as you will see in a minute, there are benefits to actually using projects over custom GPTs. But both things will do roughly the same thing. They will get an input, they will run through a process and will give you an output and they will give you a consistent output every single time. So we looked at the example before of the proposal, but what I want to show you right now is in addition to showing you the outcome, I want to show you how it actually works. So if I go back to the Custom GPT, on the top left corner, there is the name of the Custom GPT that has a drop-down menu, and you can click on Edit.

14:53Isar Meitis:If you click on Edit GPT, which is exactly the same screen you're going to see if you can show you in a minute how to do, you're going to see a similar screen. What you're going to see is on the left side, you're going to have the name, the description, the instructions, the inputs, the knowledge base, the actual graphs you attach, and some other stuff such as recommended model that you can pick and capabilities that you can choose. And if you really want, you can add code on the bottom. And there are conversation starters where you can add buttons to allow other people to understand how to use your custom GPT.

15:25Isar Meitis:But let's dive into the most important things. The name just gives it a name, tells it what it does. So in this particular case, it's Multiply AI Training and Education Proposal. The second thing is the description, writes proposals, blah, blah, blah. The description doesn't really matter. It's for you. And if you share it with others for the people you share it with, The most important part is the instructions. The instructions is where it tells you what it's actually doing. And you can see you're an expert proposal writer. Your goal is to write clear, easy to read, and follow an attractive AI training proposals.

15:51Isar Meitis:And then it explains what the inputs are going to be. And it's going to explain exactly what the process is to take those inputs and analyze them. And then it tells it exactly what the output needs to be. So this is how it runs. It references two separate files in the proposal, in the instructions. One is the Multiply AI Services brochure where it can take information about what are the services and describe what's the value in doing or taking these services from us. And the other is a master template AI for training and education of AI. Why does it need that? Because then it knows how to draft it from a formatting and flow and structure perspective.

16:26Isar Meitis:And this outline is really a very full comprehensive outline of a proposal I will never write. It describes every single thing that I deliver, which nobody ever orders all of them, at least not all at once. And so what it tells it in the instructions, it says each proposal that you write will include one or more of these training options. You need to only include the components that were discussed with the prospect based on the transcriptions and or emails I will provide you, right? So this is what it does. It gets access to a lot of different options and it knows how to pick just the ones that are relevant based on the conversation we had.

17:02Isar Meitis:And that's it. That's all you need. That's the entire magic. And it will know how to write proposals if you wrote the instructions correctly. We're going to talk about this in a minute on how to do that. In a very similar way, if you go to the project, the project has the same thing. So if we look at the project level, you will be able to see the files that are connected to it, as well as you can see brand guidelines and clarifying client proposal details and so on. And you can see here that it has instructions. so inside the instructions if i click on them it is the same exact instructions that we've seen before i literally copied and pasted the instructions so if we go back to the previous section you can see it looks it has the same exact thing the overview the knowledge base requirements the core workflow like all the different things that were there before are also here and then you can attach different files to here that you can attach so it can use as references for the proposal so very similar process different tools but instructions and knowledge files and then you can just drag in the transcript or emails or whatever you define for it as inputs and it will know how to do the work.

18:06Isar Meitis:By the way, to create new custom GPTs inside of ChatGPT, you go, as I mentioned before, to the three little dots where it says more. You click on GPTs and on the top right corner there's a create button. If you want to create a project inside of ChatGPT, they live right above your chats and you can see I have many, many, many of them. And in here next to where it says projects, if you put your mouse over it, there's a plus button that will allow you to create a new project. So on both platforms, it is relatively easy to know where to go. Again, other than custom GPTs that are now hidden under the more menu, because I think OpenAI wants you to now build projects and or agents or skills.

18:42Isar Meitis:So now let's continue with our flow. Another example, by the way, that we have here, which I'm not going to dive into, but I will talk about in two seconds, is converting data, raw data, like really large Excel files. So if I open this, you will see that it has multiple columns, like dozens of columns, and then thousands of rows in this particular file. And this could be a sales report, this could be financial analysis, this could be a scraping of customer pricing, like whatever the source data is. And it turns it into a very detailed report in the end that shows, in this particular case, sales, and it has the table of content, and it has an executive summary, and it has a business overview, and it has a lot of other stuff that shows graphs and charts and different information.

19:26Isar Meitis:By the way, all the information, this one is completely fake. It's based on a random generated data, but the concept is perfect, right? You can see a very detailed, again, those of you are seeing, those of you who don't have to believe me. It is a well-branded, well-structured, well-organized report with graphs and charts and analysis of all the data from the source file. And this can be built either as a custom GPT or as a project or as a skill. So that doesn't really matter. So now between custom GPTs and projects, let's look at a quick comparison table. The purpose of both of them is roughly the same, right?

20:00Isar Meitis:Is to be able to have a conversation, but the project has another benefit. The project is more of a workspace or like I said, a context bubble where you can have free conversation. So while the custom GPT is built to do a recurring task again and again and again, the project can do the same thing, but can also allow you to just have any open conversation inside the project while taking into account the memory of the project and the information that was attached to it, while a custom GPT doesn't do it, or at least doesn't do it as good. You can attach files to both of them. You can give instructions to both of them.

20:35Isar Meitis:The amount of data you can give into each and every one is roughly the same, depending on the specific plan you have. There are three big differences. One is memory. As I mentioned, custom GPTs, every single chat is a new chat. It doesn't know anything about what happened in the previous chats with that custom GPT. Inside the project, there is a project memory where it's going to learn more and more information as you have more conversations inside the project. The other benefit of running the conversations inside a project is the conversations live inside the project, where you can see each conversation that happened inside the project, one under the other, while in a custom GPT, the conversations appear in the regular conversations of chat GPT, which means they're going to be hidden between the other 10 ,000 conversations that you're having.

21:20Isar Meitis:So if you want to be able to see the same report that you created last week, it will be extremely easy to do in the projects because it's just going to show there as the previous line item. And in the chat GPT, you'll have to do search and filter and find the right one and check it and so on. So that's another benefit. The only real benefit of custom GPTs is that you can add code at the bottom of the custom GPT, which means you can connect it to an API of external tools. This is something you cannot do in projects right now. That being said, with skills, you can now do similar things. And that being said, I probably have over 50 custom GPTs that I created and used, and I created code in two of them.

21:57Isar Meitis:So it's not something that I've done very commonly, and I doubt that a lot of people did. While it is a benefit, it is not a huge benefit. Once I understood the benefits of projects, I stopped using custom GPTs completely. I converted the main ones into projects, not all them. And then I said, again, I converted more or less everything into skills and agents. How do I create all of them? Whether I'm creating a custom GPT, a project, and or a skill, I'm always creating them starting with a regular chat. I'm going into a regular chat, either in Claude or in ChatGPT. When I do this in Claude, I do this in Claude Cowork.

22:31Isar Meitis:When I do this in ChatGPT, it doesn't matter. You can do this in the regular ChatGPT and or in Codex. And you can explain what you want to do. You can give it the files, and you just work through the process. You give it the data. I said, okay, let's clean the data. I want you to understand what's in the data. I'm going to give you this kind of data every single time. Now you iterate through the process. You explain exactly how to get to the outcome you want to get to, whether the final report or the proposal or the analysis or whatever it is that you're trying to do. Once you get to that outcome, you can ask the AI, again, whether Chachupiti or Claude or Gemini or any other, and say, I want you to turn everything we did right now, just the stuff that worked, not the stuff that didn't work into a X.

23:09Isar Meitis:This could be instructions for a custom GPT, instructions for a project or a skill. And you will know how to do that. If it builds a skill for you, it will package it and will give you an install button as soon as it's done. If it is a custom GPT or a project, you will have to then take the instructions and paste them into a custom GPT you create or a project that you create. What you should tell the AI is which files you're going to give the custom GPT or project as references. So it can use it in the instructions that it's creating. So let's look at a quick example. In this example, I was building a custom GPT that writes questions for sessions of the courses that I teach.