153 | Mastering AI Tools: How ChatGPT Projects and Canvas Unlock Transformative Business Value with Isar Meitis

0:00Hello and welcome to Leveraging AI, the podcast that shares practical ethical ways to leverage AI, to improve efficiency, grow your business, and advance your career. This is Isar Maitis, your host, and I've got a special episode for you today. At the end of 2024, OpenAI treated us with an amazing holiday gift that they called the 12 Days of OpenAI. Now, they've had multiple great releases. I'm actually recording this episode before I know what the last three are, But putting that aside, everybody was expecting to get Sora, and we did. Everybody was expecting to get the full ChatGPT 01, and we did.

0:40So a lot of excitement around there, multiple developer advanced capabilities and stuff like that. But from my perspective, the three most impactful tools for day-to-day work are actually the ability to share your screen with ChatGPT, as well as share real-time video with ChatGPT and interact with it. Projects, which I think is a huge, incredible capability that was really missing and that provides a lot of value. And Canvas in custom GPTs. Now, those of you who don't know what I'm talking about, don't worry about this. We're going to dive into two out of the three, and I'm going to show you specific use cases and explain to you how you can make the most out of these amazing capabilities that we never had before, so you can start benefiting from them in your business.

1:31And that is going to be our focus of today's show. In the next few years, AI technology will change our world dramatically. Whether you are a business executive trying to catapult your business forward, or just somebody who refuses to be left behind and want to advance your career, this is the show for you. I'm your host Isar Maitis, a serial entrepreneur and an AI enthusiast. You'll hear invaluable practical tips from innovative business leaders, AI practitioners, and some of the brightest AI minds in our world today on how you can leverage AI in ethical ways to advance your career and grow your business.

2:18Now, before we dive into the two tools out of the three that I want to talk about, I want to give you a little bit of background. There's been a lot of communication recently claiming or arguing whether the development of AI large language models is slowing down. Sundar Pichai specifically said that there is a slowdown, that the hill is steeper, that the ROI is declining. On the other hand, you have Sem Altman, the CEO of OpenAI, and Dario Amadei, the CEO of Anthropic, saying it is no slowing down, there is no wall. So it's up for debate, But what is very clear is that there is a change in direction in all the large companies.

2:55And that direction is actually very helpful for us. And I will explain. If previously the entire focus was on let's train a bigger model with more compute and more data, the focus right now is providing smarter solution that will provide more business value. And that comes in two shapes in the past couple of months. One of it is reasoning models. So models that invest time in thinking of every step of the process before they execute, rather than being 100 % based on the training data that they have. So ChatGPT01 was the first of these models, and we're going to see more and more of them because all the different companies are working on the same thing.

3:32But the other big one that has been maybe neglected a little bit in the news, but I find it to be maybe the biggest thing that happened to AI so far is a big focus on tools, meaning how to take the existing capabilities, stuff you already have, and just packaging it in a way that will provide a lot more value to us, the users, especially if you are in business and you're using it for business use cases. And so, as I mentioned in the intro, the three biggest things that we just got that are new tools that are building on top of existing capabilities, one is screen sharing and the ability to also open your camera and show ChatGPT what you're looking at and get their feedback on that.

4:12I will probably do a complete separate episode about those capabilities because there's about a gazillion use cases we can dive into. So what we're going to focus on today is how to use projects in ChatGPT, what they are and how to apply them for multiple business use cases. And the other is going to be Canvas in general. Many people don't even know what Canvas is in ChatGPT. And those of you who do know, maybe don't know how to push it to its limit on how to use it, or maybe you don't know that now it's available within custom GPTs. And if you don't know what custom GPTs are, don't worry, we're going to talk about this in the episode as well.

4:48So let's get started specifically with projects. And I will start with describing what projects are, and then I'm going to share my screen and explain exactly how to use them and give you a specific business example on how you can apply this. But projects really provide us two things that were missing from ChatGPT and to be fair, missing from all the other models as well. Number one is the ability to organize our chats into folders. So projects are these containers that you can open on the left side of ChatGPT. Again, I will show you how it looks like in a minute. And for those of you who are listening, don't worry about this.

5:23I will explain exactly how it looks like on the screen and how to find it. But you can literally create these folders where you can now aggregate or organize all your historical chats because you can drag and drop chats in and out of these folders, which will allow us to have a much better structure to all of our conversations, which is definitely a great step forward. If you remember, just recently OpenAI added search, which allows you to find older conversations very easily, but now being able to look into specific folders will make it even easier. But that's not the biggest benefit of projects.

5:59And I think the other benefit, which is more important, is why they call it projects versus just folders. So for that, I will open ChatGPT and let's look how it looks like. Again, if you're just listening to the podcast, I will explain everything that's on the screen. Or if you have the time to switch to something in front of the screen, if you're listening to this in front of a computer or where you're not driving or walking your dog, you can go in the show notes and find the link to our YouTube channel. And then you can watch what's on the screen as well. Okay. So the way projects work is, as I mentioned on the left bar, where you have your custom GPTs and you have the history of all your conversations, there's now a new section called projects.

6:40And as I mentioned, it looks like folders and you can drag and drop conversations from your history into each and every one of the folders, just like I'm doing right now. and then it's going to appear over there. But that, as I mentioned, is not the big deal. The big deal is if I click on one of these projects, you can add instructions and you can add files into each and every one of those folders, similar to what you can do with custom GPTs. But it's very different from custom GPTs and I'll explain what are different use cases that you can use it for. So as an example, you can see that I have several different projects that I've already created.

7:17One is called strategy. The other is called speaking gigs. The third is called courses. The next one is called sales. The other one is marketing. And then I have clients. I have client one, client two, client three. I just did it this way. So you don't know the names of my clients, but afterwards I will change it back to the actual names of the clients. And I have leveraging AI, which is the podcast. Now, the reason this is very powerful is because I can provide background information that is relevant for each and every one of these topics or for each and every one of these clients. And then every conversation I have will have all that information to use in order to give me better output.

7:57So let's take a step back before we dive into why this is important. And I'll give you an example of how I'm actually using it. The biggest problem all these tools have is they lack context. They don't know anything about you. They don't know anything about your company. They don't know anything about your industry, about your niche, about your clients, what are the things you're working for, your services, your products, your benefits over your competitors, all the stuff that you and other people in your company take for granted. The AI does not know. And that's why if you don't provide it that information, you're getting answers that are not impressive.

8:34But if you provided all that information, if you provided all that context, you can get significantly better results that would be customized to all that universe that you take for granted and so does everybody else in your company. Now, there were two ways to do it so far. One is to create a custom GPT where you can upload all that information and then give it instructions to do something very specific for that client. As an example, I have these GPTs to write drafts for blog posts on my behalf to write drafts of posts on LinkedIn on my behalf and so on. I then review the drafts and I make my own changes, but the drafts are being created based on historical information that I provided it about myself, about my background as an entrepreneur.

9:21I give it examples about the way I write. So it writes in my tone, et cetera, et cetera. So then it can write just like me and provide me great graphs. The disadvantage of that is that it's obviously it's limited to that particular task, meaning the instructions are very geared towards writing a blog post versus I want to talk to you about everything about this client so I can get better ideas on what to do or write better proposals for them or whatever the case may be. So I can continue creating a thousand of these GPTs, which is not necessarily efficient, especially if it's just a day-to-day ongoing stuff.

9:54I don't want to create a hundred GPTs, but that was the solution we had so far. The other solution that OpenAI provided us is long-term memory. So OpenAI has a memory function where ChatGPT on its own decides what it thinks is important about you, about your company, about your industry, and it's going to remember that. Now, in general, that's not a bad idea because now it's going to know more and more about you and will provide you better and better results. The disadvantage of that is you have very little control on what it's actually going to remember and what kind of information it decides to put in its memory.

10:33The bigger problem is when you're touching on multiple aspects of your business, you will still remember all of them. And if you're like me and you're in consulting and you have multiple clients, then it gets completely confused because it doesn't know what information relates to which client and it's going to remember all of that. And if you want to make it even more extreme, if you have several people sharing the same account, in my particular example, my kids are using it for schoolwork, for research, for gaming that they're building, and for other stuff, that totally throws the AI off as far as what is the goal and what am I trying to do.

11:09Projects solve that problem. So let's look at client one as an example. If I open the client one project, you will see that I can run a chat show, just like in any other model. I have all the tools that I had before. So video tools and so on, I can attach files, et cetera, et cetera. But I can also add custom instructions to explain to it about this particular client, and I can add files. And in this particular case, you can see I added six files. So what are these files? The files are the following. The files are called culture, force proposal, consulting agreement, people, workshop proposal, and about.

11:47Each and every one of these files is providing information to the chat, but to every chat I'm going to have within this project about this client. So it has the course that I've done with them, the AI course that I provided to them. It has the workshop that I provided to them and all the details. And if I wanted the communication about it from Slack or from emails and so on, I have the consulting agreement that I have with them. So the AI knows everything of what I'm doing for them and what's in scope and how do we work together and everything that's defined in the agreement. But it also has an about PDF, that about PDF I actually took from their website.

12:26I actually took from their website and then just uploaded it here. It has a lot of information about what they do, what products they sell, what are the benefits, why are they leading in the market, and so on. I also added an about thing that I created. And the way I created it is using voice mode in ChatGPT. I literally just opened voice mode and explain to it what I'm trying to do, which is to tell it everything it needs to know about this client so it can help me better in the future. And it asked me about 15 different questions that I just answered with voice. It's just like having a conversation with somebody, but then I took that and saved that as a file and upload it into the memory of the project.

13:05So now it knows everything I said about this company and everything that it needs to know about it. But then there's two separate sections out of that. One of them is culture. This company has a very unique culture that really makes everything they do different. And it's very important that the AI knows that because it can then refer to that when it provides me every output that I need, whether it's ideas or resolutions for problems that I'm having or brainstorming or content that I want to help them create. All of that is relating to their culture. And then the last thing is people. I described one by one, each and every one of the people that I'm working with in that company, what they do, what do I think about them, what's their personal approach to things have been, and everything I know about them.

13:48So then it can refer to that as well. So all of that prep work I had to do once, but now every conversation that I have about this client knows all these things and will be significantly more specific without me having to prompt it every single time and waste half the context window on explaining about the client. This is already built in. I don't have to waste time. I don't have to waste memory. And it's just magical. So two examples that I have here, one is a consulting services proposal. So they asked me to provide them another proposal with additional consulting capabilities that were not included in the original scope.

14:29And the AI was able to create for me the draft for this proposal in seconds because it understands everything about this client. It has the previous proposal, so it knows the exact format that I'm using, and it created a very good first draft for me in a few seconds. All I had to do is explain with my voice using the voice-activated solution on what is it that I want to provide in the new scope of consulting. The other one is client relationship strategy. I have a situation with that company where they brought in another firm, another consulting firm, which is perfectly fine. They're not really competing with me.

15:09They're doing other stuff, but there is some overlap. And I want to prevent issues that can arise in the future. And I just needed somebody to consult with and brainstorm with on how to reduce the friction with the other company and potentially even how to create great collaboration with them in this client and with other clients as well, because we do complementary things. So I had a brainstorming session with the AI, but again, it knows everything about the company, about its history, about what they do, about what they're doing with me. And then it was able to ask me about what is the other company doing?

15:41And then it really helped me think through how can I approach this in the most effective way? This is pure magic. Now, as I mentioned, I have other folders for strategy, my strategy, my business strategy, where I explain everything about my business, about the different clients, about who do I serve, how do I serve them, about courses. When I come to plan courses, so it has all my courses, content, all the feedback, all the testimonials, all of that is built into now a universe that I can operate in and get significantly better tailored and a lot more value from every single conversation without any fancy prompting and prompt libraries that explain everything again and again.

16:22I really think this might be the most valuable thing that ChatGPT gave us in the past year. It's a game changer when it comes to its ability to understand your specific needs on specific topics. Like I said, this could be a marketing thing. Or if you want to go deeper in your company, you can do social media marketing project versus blogging versus video marketing versus paid marketing. And each and every one of them will have a universe of data and different instructions. And then yet still people can have conversations with it that will have that as part of context. And as I said earlier, context is the most important factor when it comes to getting good results from these large language models.

17:10So this is what I wanted to share with you about projects. Try it out. Start with one thing. It doesn't matter what it is. It could be an aspect of your business, could be an aspect of your personal life, could be a specific client. Give it as much information as possible. And as I mentioned, you can use voice to create some of that information, but many of these files you already have. You can download communications from multiple channels and upload them here and so on and so forth. It's amazing. Don't miss out on this really important capability. Other amazing tool that I want to share with you that ChatGPT has given us is called Canvas.

17:47Now, Canvas is not new. They released Canvas a couple of months ago, but Canvas in custom GPTs are new. And we're going to explore both at the same time because both capabilities, custom GPTs and Canvas and the combination of them is absolutely magical. So this can help you create graphs, documents, very detailed in a collaborative way together with AI, with your ability to really be involved back and forth in the process while doing this in an automated way. So let's take an example. I'm planning charity events. And in this particular case, and I created a custom GPT to help me create a script or videos that I'm going to use to promote those charity events.

18:35So what is a GPT? A GPT is basically a set of instructions that allows you to repeat a process. And we have a gazillion repeatable recurring things that we do in our business in every single department. You can build these mini automations that can do that. Now, this is a very simplistic example for us to use in this episode, but I have probably 20 of them that I use regularly in my business to do multiple aspects of my business. They appear on the top left corner of ChatGPT. And if you want to see how to use it, there's a button called Explore GPT. In that button, when you click on it, you have all the GPTs that people shared in the GPT store.

19:19So there's millions of them that do multiple things and you can search through them. but on the top right corner, there's a create button where you can create your own GPT. Instead of doing that, we're going to go to one of the GPTs that I already created. That's called fundraising script. And again, this is just a simplistic example. What the GPT has inside of it is a few things. It has a name. In this case, it's called fundraising script writer. It has a description, helps write scripts for fundraising. And then it has instructions. Again, in this particular case, the instructions are really short.

19:49You're an expert script writer specializing in writing scripts or fundraising for nonprofits. Your task is to assist the user in writing great scripts for videos. All scripts must include the following components. Background, the why, what we are doing to solve the problem, and a call to action. Your first task is to ask the user as many questions as necessary to identify all information you need to write a script according to the four topics above, do it in Canvas. So this is where I add my request to add it in Canvas, which we're going to see what it does in addition to that. Now, in addition, I created a conversation starter.

20:28A conversation starter is basically a button that's going to appear that will allow people who do not know what the GPT does to start interacting with it, because otherwise you're just staring at it and you don't really know what to do. So in this particular case, what I wrote is wrote, please help me write a script. Now you can add as many of these as you want, and you can even refer to them in the instructions. So different things will happen depending on which buttons people click. Now I can upload files. In this particular case, I will upload files about my company and about the charity and about what we're donating for.

20:59So it has all that information in its database. And then there's capabilities. These capabilities can be web search, canvas, DALI image generation, and code interpreter and data analysis. In this particular case, All I need is web search, so you can search stuff about the event or about the charity. I need Canvas because I want to use Canvas, which we'll explain in a minute what that is. And I want DAL-E because I want to be able to create images that will give me ideas when I actually do the script or when I want to send it as an email and so on. So let's see how this actually operates and see what Canvas does.

21:31And we'll also combine it with voice capabilities because that makes it even more powerful. So when I open the GPT, which is just clicking on it on the left menu bar, you can see it has the name Fundraising Script Writer by Isar Maitis, helps write scripts for fundraisers. And I have the button that I created that says, please help me write a script. When I click on the button, ChatGPT will follow up explaining what it's going to do and asking me the questions that I ask it to ask me. Now, the reality is I requested it to ask the questions in Canvas. So what I have right now is I have a bunch of questions that I need to answer.

22:07So it says to craft compelling fundraising scripts for your nonprofit video, I'll need some information from you. Here are the key areas I'd like to explore. Background, and then there's three questions. The why, and then there's three questions. What are we doing to solve the problem? And then there's three questions, and then a call to actions, and then there's three questions. Now, what I can do, because Canvas is a collaborative environment, I can literally go to each question, hit enter, and then write the answer. This will allow me to collaborate back and forth with the AI inside the document.

22:41And then based on that, I can go ahead and ask it to create the script. Another thing that I could do that is even cooler is to actually use voice mode. Right now, advanced voice mode is integrated into everything in ChatGPT. And so what I can do right now is literally click the voice mode button, which looks like a sound wave on the bottom right of the prompt section in ChatGPT. And I can literally click on that and just answer the questions. So let's do this very quickly so we can continue and I will show you even more things you can do with Canvas. So I'm clicking on the voice button and starting ChatGPT.

23:15Hi, ChatGPT. I'm going to answer all the questions that you provided me in voice. So background, what is the name of the nonprofit organization? It's Isar's Charity LTD. Got it. Isar's Charity Litted. What's your mission or vision? My mission and vision is to help local teens to do better in their lives while supporting the community. The next question you had is, can you share a brief history of context of your organization? We've been doing this for a few years here in our local area in Central Florida. Then you asked about the why, what specific problems or issues we are addressing is we're helping teens find purpose and find activities that they can do to help the local community.

23:58Why is this issue urgent or significant? It really helps the local communities while raising a new generation of people who are community aware. Your last question was, how does this problem impact individual or communities? It is helping multiple people across various levels of society in our area develop a younger generation that is aware of the issues that surround them and that are willing and driven to help with that. Let's stop with this. I wanted to summarize that in a document and add it to the canvas. I've summarized your input and added it to the canvas. Let me know if there's anything you'd like to adjust or expand on.

24:42So what I have right now is I have a new canvas in the conversation that has all of that information that I just gave it. And what I'm going to do now is I'm going to ask it to write the script for my video based on the information as well as the structure it had initially. So I'm typing in this particular case, please create a new canvas with the video script using the information I provided. Please create the script in Canvas. Now, in many cases, you don't even have to say you want to create it in Canvas. It will do it on its own. But the benefit of asking is you A, guarantee that. By the way, you can also create Canvas with stuff that you have from before.

25:35So you can upload a document and then engage with that document in Canvas. And I'll show you in a minute what I mean by engage with the document. So what's happening right now is it's writing a script for a video for me that will take into account the information that I provided with my own voice so I didn't have to type a lot that follows the question that it has asked me based on the instructions in the GPT. So the GPT told it what questions to ask or not exactly, but what topics to ask about. I use my voice to provide it information. And now it created me the script. So the script has opening scene, narrator with voiceover.