Everyday AI Podcast – An AI and ChatGPT Podcast - EP 332: Claude Projects - What they are and how your company can save time using them

Episode Date: August 8, 2024

Win a free year of ChatGPT or other prizes! Find out how.There's a time-saving, eye-boggling AI gem under your nose. Similar to custom GPTs from ChatGPT, Claude Projects can save you a ton of time. But what are they? And how can your company save time using them? Newsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion: Ask Jordan questions on ClaudeRelated Episodes: Ep 301: Anthropic Claude 3.5 Sonnet – How it compares to ChatGPT’s GPT-4oUpcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:1. Overview of Claude Projects2. Practical Use Cases of Projects3. Claude Projects vs GPT-4Timestamps:01:55 Daily AI news06:30 About Claude Projects11:17 Anthropic Claude lacks essential features and connectivity.19:14 Emphasizing importance of integrating language models in work.21:43 Large companies need generative AI for efficiency.36:51 Claude Projects example42:07 Making Claude Projects an analyst46:50 GPT converses like a human, remembers context.51:37 Create various applications easily and transparently.53:37 Specific vs. broad insights on SEO trends.58:57 Claude's team plan requires five seats minimum.01:07:28 Powerful Claude Projects show data in real-time.01:10:18 Leverage large language models for business success.Keywords:Anthropomorphic Claude, Claude projects, large language model, AI capabilities, Palantir, Microsoft, French startup Mistral, OpenAI, Project Strawberry, Amazon, Anthropic, Chat GPT, Prime Prompt Polish Chat GPT course, business use, Apple's AI developments, Claude Projects features, data visualization, business dashboards, email campaign performance, SEO performance analyzer, everyday AI, Claude artifacts, GPT use, AI News Assistant project, AI in business intelligence, Buzzsprout, podcast distribution, legal compliance, Claude 3.5 SONNET, market research analysisSend Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Start Here ▶️Not sure where to start when it comes to AI? Start with our Start Here Series. You can listen to the first drop -- Episode 691 -- or get free access to our Inner Cricle community and all episodes: StartHereSeries.com Also, here's a link to the entire series on a Spotify playlist. 

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live in Adobe Firefly, the All In One Creative AI Studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome. The assistant accelerates execution. What are clawed projects?

Maybe you've heard of Anthropic Claude, and maybe you haven't, but they actually have a great feature that rolled out a couple of weeks ago that I think people aren't paying enough attention to. It's kind of like creating your own version of a large language model on your data and with some custom instructions. So we're going to be going over clawed projects today, what they are, and even building one live right here on Everyday AI. What's going on, y'all?

My name is Jordan Wilson, and I'm the host of Everyday AI, and this thing is for you. It's a daily live stream podcast and free daily newsletter, helping us all better learn generative AI so we can actually use it and actually leverage it to grow our companies and grow our careers. So today's episode is, I think, especially relevant for that because we always hear a lot like, hey, Jordan, seems like there's so much going on in AI and all these large language models. What are some actual use cases? How can I actually use them? Are they powerful?

Can they work with my data? Well, we're going to be answering all of those questions and giving you, I think, some very practical in real use cases live on today's show. All right. So before we get into that, we're going to start as we do every day by going over the AI News. And if you are new here joining on the podcast or the live stream, thank you so much for tuning in. Make sure to check out the show notes on the podcast for a link to our website. You need to go your everyday AI.com. Sign up for the free daily newsletter where we recap everything that's going on in the world of AI as well as more depth on today's episode. All right, let's take a look at what's

going on in the AI news. All right. So Palantir and Microsoft are partnering to boost AI for US defense. So Palantir Technologies, which works in the defense space, and Microsoft have announced a partnership to enhance AI capabilities for the U.S. defense system and intelligence communities. So this collaboration will integrate Microsoft's Azure Open AI service with Palantir's AI platforms in classified cloud environments. The aim is to operationalize AI for national security tasks like logistics and action planning using advanced. models such as GPT4. The rollout of these services will depend on government authorization and accreditation. Palantar suite of products, including Foundry and Gotham, will be deployed in Microsoft's secure government clouds. Both companies stress their commitment to responsible privacy standards and ethical AI practices. Yeah, this one's interesting. More AI and even from

open AI making its way to the defense industries. All right. Next, French startup and a large language model maker Mistral has announced major updates hours ago, highlighting the introduction of customizable agents and advancements in La Platform. All right. I don't know if that's how you say it in French, right? But this new alpha release of agents offers a new way to wrap models with additional context and instructions.

These agents can be used on led chat or via the API to create custom behaviors and workflows. With the advanced capabilities of Mistral Large 2, their newest and very capable model, developers can build increasingly complex workflows using multiple agents, which are designed to be easily shared within organizations. Future plans for agents include connecting them to various tools and data sources, enhancing their functionality and versatility. A lot platform now supports the customization of flagship models like Mistral Large 2 and Codestrel, also from Mistral, obviously.

And developers can tailor these models using base prompts, few shot prompting, or fine-tuning, even with personal data sets. So yeah, very similar to what we're going to be going over today from Claude in their projects was just released literally hours ago. So make sure to check on the newsletter for more on that. Last but not least, Open AI CEO Sam Altman has been teasing Open AI's newest model. And a lot of people are thinking it's going to be coming out any day, like maybe today. I don't know personally, but here's what's actually happening.

So Open AI CEO, Sam Altman has sparked intrigue with a cryptic tweet about Project Strawberry, hinting at advanced AI developments. So Altman shared a photo of strawberries in a garden with the comments. I love summer in the garden, suggesting a deeper meaning related to Open AI's secret project. So yeah, it was a photo of strawberries, y'all. But according to a recent Reuters report, Project Strawberry, which was previously called Q Star, aims to develop a model with advanced reasoning capabilities moving closer to autonomous AI agents. And also reportedly with this new project strawberry, there is an emphasis on these agents being able to autonomously crawl

the web as well. So internal documents reveal that strawberry from open AI could enable AI to not only generate answers, but also plan and navigate the internet autonomously for deep research. All right. So keep an eye out. If it does break today, we'll obviously be covering it here on Open AI, or sorry, here on everyday AI. But, you know, it's hard to tell with these things. However, Open AI did actually tease and release its last model GPT40 in a similar way with CEO Sam Altman, essentially sending out a similar cryptic tweet. So there could be some new developments from OpenAI after they've kind of been going through a lot recently with reports that they're, you know, losing billions of dollars with other reports of some of their top

leadership leaving for other companies. So it's been kind of a bad week for OpenAI when it comes to press. So, you know, we'll see if this is actually, you know, amounts to anything. All right. That's it, y'all. Let's go ahead and, you know, hey, but as a reminder, make sure to go sign up for that newsletter. There's a lot more today, Your EverydayAI.com. Let's get straight into it. Let's talk about Claude projects, what they are, and how your company can save time using them. And we're also going to be looking at how these new Claude projects compare to GPTs from OpenAI. All right. So here's what we're going to be going over. We're going to be going over Anthropic Claude Overview. So if you are brand new to

Anthropic Claude. We're going to be giving you the high level. Don't worry. Because, yeah, we talk about Chad GBT a lot on the show. But if you don't know Claude, don't worry. We're going to go over an overview of that. We're going to go over a project's overview. And then we're going to say why you or your company might want to use

Anthropic Claude's new projects feature. We're going to build a project live here on the show. Yeah, if you're listening on the podcast, we actually do this as a live stream. This is unedited, unscripted. realist thing at artificial intelligence. And then last but not least, we're going to compare the Claude projects to GPTs from OpenAI.

We're going to briefly go over the pros and the cons, as well as look at some of the outputs. All right, live stream audience. Thanks for tuning in. Appreciate y'all, as always. So Kathleen, joining us. Thank you. Jennifer, good to see you back.

Zane and Jason and Gordon, everyone else. Big bogey face joining on YouTube, Cecilia, Tara. Let me all. Let me note, have you guys used Claude projects before? What are your biggest questions? Get them in now. I'll try to answer them either as we go or at the end of the show. All right. So let's first start with a Claude overview. So if you are brand new to Anthropic Claude, it's pretty simple. It's a large language model, right? A lot of people, you know, maybe they just know of AI as chat GPT. AI is not new. It's been around for decades. Generative AI, you know, the definition of it is always.

changing, but essentially, you could say the birth of generative AI took place sometime from 2020 to 2022 with chat GPT and the GPT technology. However, Anthropic was founded by former open AI executives. We talked about that on the show on Tuesday, but it is a very capable model, a very capable, large language model. And their newest version of this is Claude 3.5 sonnet. So it scores very high. A lot of people say, that Claude 3.5s on it is the most capable model out there right now, more so than OpenAIs GPT40. I would say not so much. I still personally think GPT4O is better, but that's up for personal taste, right?

When you look at benchmarks, for the most part, OpenAIs GPT4O is still leading at least for products that are available for consumers on the front end. All right. So that's a super high-level overview of Anthropic Claude. It is backed by Amazon in a huge way, I believe, a $4 billion investment. And I do think that they are recruiting Anthropic is some of the top talent in the world, right? So we talked about this earlier. Some of the top executives, even in the last week or last couple of months, have been leaving OpenAI to go to Anthropics. So very capable model.

I think from a features and functionality standpoint, they are behind, at least when you compare them to OpenAI and chat, GBT. We'll talk about that a little bit more. But if you are brand spanking new to Anthropic Clawed, that is it. There is a free plan, which is very limited. I think the chat GBT free plan is much better. So there is a free plan. There's not a lot of outside features and functionalities. And let me just put this one out there first before we even get into any comparisons.

A downside right now of Anthropic Clawed in why I tell like, right, so a lot of people hire us to, you know, consult them. help us, you know, to help them implement these kind of systems for their companies, right? You see all these, you know, studies from, you know, McKinsey Digital that say, you know, 60 to 70 percent of manual knowledge work is going to be automated by generative AI. And companies are like, yo, I want that. So we help companies, you know, you can hire us. We can walk you through this.

But one thing I always talk about with companies is, hey, even if infropic Claude looks great right now, I don't recommend it for a lot of entry-level use. cases, mainly because it is the only, I think, major large language model maker, at least if you kind of look at the quote unquote big four, Open AI, Microsoft co-pilot, which technically uses Open AI's technology, Google Gemini, and then Anthropic Claude. Claude is the only one that doesn't have some sort of internet connectivity out of the box. And that is extremely important in a huge downside for Anthropic because essentially, large language models to oversimplify this, they have a knowledge cutoff date, right?

So these models are trained on data up to a certain point, right? And that's where you can run into trouble because think of your business. Think of how fluid things are, how much trends change in your industry. If you are working with a large language model, you want that model to have the ability to connect to the internet in real time, and you want some sort of ability to navigate it, right? if you do want to start bringing your business operations inside of a large language model, which you need to, you need to know what that model is capable of and what it cannot do. So right now, it is worth in, you know, as we're going over a Claude Anthropic Overview,

or sorry, an Anthropic Cloud Overview, it's important to point out it doesn't have access, right, to the internet in real time. Obviously, if you're using third-party softwares, if you're using Anthropics API and building something on your own, you can access the web, via other ways. But out of the box, if you are just logging on to, you know, Claude.a.i, right? That's the address.

If you want to go there and sign up for a free account or use a paid account, that's where you would access it. But out of the box, it does not have real-time internet access, which is a huge downside. All right. Yeah, Michael, I agree with what Michael said here in the comments, said Claude's free plan is so limited. So I upgraded.

Even there, if I'm being honest, Even their paid plan is extremely upgraded in terms of limits. I hate saying this, right, because I think Claude is a great product. For me, it's almost unusable, right? Again, I am a power, large language model user. I believe it's 30 or so messages every five hours. I could be wrong on that. It might be 50.

But the limits are so low. Obviously, on chat GPT, we have a team's plan, which actually gives you double the limits. but even the paid plan of chat chvety is much, much better in terms of limits than the Claude Anthropic plan. So that's just something to throw out there. All right. Yeah, a lot of people, a lot of people on the live stream here, loving Claude. You know, Darcy says she loves it. Zane says loving it.

Gordon, same thing. Yeah, but the limits. The limits are bad. So, hey, if someone from Anthropic is watching, listening, you need to address that. I do not literally, when companies ask me, hey, what are the pros and the cons and how should we start implementing generative AI? I say Claude's a great model, but two of the main reasons you shouldn't be using it, at least if you're trying to use it as a front end user, right?

If you're using it via the API, those limits are gone and you're just essentially paying for usage, right? But a lot of companies want to get started, especially medium and small companies. They just want to get started by logging onto the front end of, you know, a claw. or a chatGBT.com or a Gemini. And right now, Claude, if you're just logging in via their, even their paid platform or their team's platform, the limits are so low. It is, I think, not super usable, which is worth noting. All right. So now that we have our Claude overview out of the way, let's talk basics on what these new cloud projects are. Okay. So if you've used custom GPTs from OpenAA

I, they are essentially that. If you don't know, let me explain this to you. So think of Claude projects, right? Think of all the different ways that you can use a large language model for your work, for your business, for whatever manual knowledge work tasks that you do. I always recommend start moving all of these things that you do on a day-to-day basis, these knowledge, domain expertise tasks that you do, whatever it is, and start moving them into a large language model like Chat, GPT or Cloud.

So we're going to go through some examples and be doing this live, but essentially Claude Projects is a way to create many different customized versions of their model, right? So essentially, there's pros and cons with big, these huge, large language models. Essentially, they are capable of everything, but sometimes, you know, it's the saying jack of all trades master of none, right? these models by default out of the box, they can do anything. But if you're not good at prompt engineering, if you're not good at the basics of working with a large language model, you might be kind of disappointed when you want a model to do a very specific task for you. It takes a lot of work, right?

And that's where projects comes in. So essentially think of projects as this. You can create an unlimited amount of smaller versions of Claude, right? You can essentially, it's not as far as saying, you know, you're fine-tuning it. but you are essentially kind of training it to just work on a specific task. So you're saying, hey, huge model that's capable of everything on the world. When we work on this project, here's what you should be focusing on. So you can put in your own instructions and also your own knowledge base, essentially.

You can drag and drop different files into this Claude project. So you're saying, oh, let's say you have a data analyst project and you dump in, you know, all of your company's data. Again, I need to put out the disclaimer out there. Never upload private, confidential sensitive PII, PHA, private, you know, private or personally identifiable information, private health information. Don't put any of that into a large language model unless, you know, someone in charge of your company says, yes, we can do this. We've checked with, you know, legal compliance, whatever, you know, all laws, right, laws in the EU, et cetera. Right. But essentially, you can dump all of your data or whatever data that you want and then create a project,

And then when you chat with that project, you are still using the base model of Claude. You would probably want to be using the most powerful version, which is Claude 3.5 Sonnet, until they release 3.5 opus or something else. So essentially, you are creating a customized version of Claude on your documents, and you essentially give it instructions. All right. And hey, let me know live through audience. We're going to get building here.

But I want to know from you all, are you using Claude projects? Are you using GPTs? Are you using neither and why? What questions do you have? Brian said, I use Claude in a few packages like Merlin and A4E, but not in their interface. Yeah. Same. I think I use Claude more via other products. So like as an example, if you're using perplexity, you can use Claude 3.5s on it. That is my default when I use perplexity if you have a paid plan. Yeah, Douglas, we're doing this live. Douglas, we're doing this live. All right. So that is the overview of Claude projects. Before we start doing these live, let's talk about why your company might even want to use these. Well, I kind of already referenced it, but I think the most successful companies in 2025 are going to be the companies that in 2024 started to move their business operations inside of a large language model or that have already started to build on top of APIs, as an example, using Anthropic Claw, using OpenAIs, GPD technology, Microsoft co-pilot, Google, Gemini, et cetera. The most successful companies in 2025 are going to be the ones that brought the majority,

or at least the majority of their time-consuming, like manual knowledge work tasks inside of large language models or building. on top of APIs, right? So you need to start rethinking how you do work. I talk about that on the everyday AI show all the time, right? Just because your company has been successful over the last couple of years or a couple of decades doesn't mean you will be successful in an AI first or an AI native world. You have to, which I know sounds weird, right? You have to start unlearning some of your businesses or some of your own personal good habits, which is weird, right? It could be the things that led to you getting a promotion. It can be the things that led, that caused your company to be a leader

in this space, right? Maybe you hang your hat on doing things the old fashion way and, you know, rolling up your sleeves and printing out hundreds of pages and marking them up. That is not a recipe for success in 2025. That is a recipe to get leaped by your competitors. You have to start unlearning what your company and what you have done in years past and you have to start relearning how to work with generative AI. That is the future of work, y'all. If you don't already, if your company is not already using AI and large language models, you are in for a rude awakening, right? Especially if you are here in the U.S. where the rules and regulations are lax, i.e., non-existent, there are really no real rules, right? If you're in the EU, you know, there's many more hoops they have to jump through

and many more boxes you have to check depending on, you know, where you are, what country you're from. But hey, here in the U.S., if you are a small, even a small company, 100 employees, if you're a medium-sized enterprise with a couple thousand employees, if you have not already started to move some of your more time-consuming operations inside of a large language model or building on top of a large language model, let me be honest here. You're going to be screwed. Yeah, let me say that again. If you haven't already, if you have hundreds or thousands of employees and you are based

here in the U.S., and if you haven't already started. to integrate generative AI and large language models into your most time-consuming manual knowledge workflows, you are going to get passed. You are asking to go out of business. So that's a reason why your company should probably be using something like Claude projects. But let's talk about some simple use cases, right? So market research analysis, right? Think of all the time you might spend.

That might be your job. You might be on a team of 10 people in market research. research or analyzing your competitors, right? You can do this as an example inside of Claude projects. Maybe sales strategy development. Maybe you're spending so much time, you know, gobbling up data from past sales cycles, from past sales and marketing campaigns. You can do that inside of Claude projects. You could do product development roadmap, right, collecting user feedback and putting, you know, files of all kinds into cloud projects. It's another thing.

We're going to be doing that live here a minute. It's not just PDFs, right? You can upload documents of all kinds. Employee training programs, right? Maybe this is, maybe you have a large HR team and you're spending a third of your time, just working on employee training programs, right? You can do that all in Claude projects. Essentially, if you have domain expertise, if you have internal company IP, right, your secret sauce, chances are it's recorded somewhere.

It's in documents. It's in conversations. It's in transcripts. It's in data. It's in sales numbers. It's in website analytics. Think almost everything that makes your company successful or your company's secret sauce. Almost every single thing can be turned into data, can be turned into structure.

data or unstructured data. Like I said, transcripts, documents, et cetera. And then you can use all of those and build projects out of them, right? An unlimited number of projects inside of Claude. The same way you can build an unlimited number of GPs. So think, where are you or your team spending the most time in these areas? All right. Should we do one live, y'all?

Hey, Michelle, don't worry if you jumped in late. We're going to do this live. Are you guys ready? Yeah, Jason, Jason says, quote, of the day, screwed. All right. And hey, if you are on the podcast, I'm going to do my best, live stream audience. You guys ready to do something live here? We do it every once in a while. Things can go, things can go awry when you do things live, right? Like if something goes down, there goes the rest of the show. But podcast audience, I'm going to do my best to describe exactly

what I'm doing here so you can follow along. But that's why just check out your show notes. This might be one of those. You just come back and, you know, watch the video on YouTube or LinkedIn. All right. So let's go ahead and build one of these things. shall we? Adobe just introduced an entirely new way to create, bringing the power and precision of its creative suite into one conversational experience. Meet Firefly AI Assistant, now live in the Adobe Firefly app, the All In One Creative AI Studio.

Powered by Adobe's Creative Agent, Firefly AI Assistant lets you start with your vision, just describe what you want, and shape the outcome as it takes form with the Assistant. The Assistant orchestrates multi-step workflows, drawing on 60 plus pro grade tools across Adobe Creative Cloud apps, including Photoshop, Illustrator, Premiere, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre-built workflows for common creative tasks, like batch editing photos, creating mood boards, portrait retouching, and creating social variations. Every step the assistant takes is visible so you can refine, redirect, or take over at any time. You stay in

driver's seat as the creative director. Adobe Firefly AI assistant now in public beta. See it today at firefly.adopi.com. All right. Let's do this thing. So I'm going to try to zoom in. Let me know live stream audience. Can you see my screen?

All right. So now I am in, I'm logged into my Claude plan. I'm on the professional plan. So that's another thing to know. If you are on the free plan, you cannot build projects. You cannot use projects. It's another big difference with GPTs. With GPTs in OpenAI, even if you are on a free plan, you can use custom GPDs.

You cannot build them, but you could as an example, pay for a month, build 50 GPs, cancel your subscription, and then use all those GPs. So I am on the professional plan here because we're going to be building a project, and you do even need to be on a professional plan to use any project. All right. So I am logged into Claw now. You can go to the side panel here. You click projects, and then we're going to click Create Project.

Okay. So what we're going to do here is we need to give the project a name in a short description. Don't worry, this is just for, you know, internal purposes. All right. So what we're going to be doing for this one live, and then as long as I don't go too far over on time, I have another one pre-built that maybe is a little bit more applicable. All right. So this is just going to be called AI News Assistant.

Okay. So all we're going to do with this, I have more than 100 pages of essentially AI news in our newsletter every single day. And I'm going to go ahead and just bring this up on screen so you can see this. And if you didn't know, you can go back and read every single newsletter that we've ever written. However, every day we have in our newsletter, I'm scrolling down here on screen, we usually have five to seven big news stories here, right? So it's called bite-sized news. So here we go. yesterday there were six stories. So essentially, I've gone through the last about four months, and I've grabbed all of these different news stories. So there's hundreds of news stories,

more than 100 pages inside multiple documents. I broke the documents down by month. All right. So now let's go back into Claude. And I have this pre-typed out. This is the wrong one. There we go. All right. So now I'm going to click Create Project. I gave it a description, a name. Now I'm going to click Create Project. Okay, so there's a couple of things that you can do inside of this project builder. Like I said, it is much more limited than GPTs. However, the Claude 3.5 Sonet model and artifacts, yeah, we're going to be getting to that. This is where I think the secret comes and why it's still worth using Claude projects,

even though I personally think custom GPs are much more robust. All right, so let's start with this. So you'll see here we have a simple interface. And right now, I haven't done any custom instructions. So if I just start talking with this project, it's the exact same thing as if I'm just talking with the base model. So over here, you'll see two different sections. I can add content.

This is the project knowledge. And then I can set custom instructions. Okay. So first I'm going to set custom instructions. And I'm going to read these for you. I don't know if I'm going to read it all. I'm going to read a high level overview because it's going to take a while. So essentially I'm saying for this chat, assume the role.

of a helpful data analyst and business strategist for the company everyday AI. In short, oh, wait, let me see. Nope, this is the wrong one. This is why we do it. We do it live, y'all. Sometimes I grab the wrong thing. All right. Here we go.

All right, for this chat, assume the role of an AI news research assistant. You are tasked with helping answer the user's queries about different AI news items. In your knowledge files, you will have more than 100 pages of news recaps. For each user query, you will take your time. time, go step by step, research deeply, and analyze all of your knowledge files in order to give the user a factual and accurate and well-research answer. And then I essentially say, here's the documents. I label them. So as an example, AI news docs May 24, 34, 33 pages of AI news recap. So we have May, June, July, and August, and we have all the different pages of

these new recaps. And then I essentially say, I'm giving Claude some information about, hey, these are in chronological order as well. All right. So now I'm just going to click Save Instructions. Okay. So now anytime I chat with this project, it is first going to go through those custom instructions and remind itself, oh, this is actually what's happening here. And I also, in the custom instructions, obviously call out to the project files here. So I'm going to click add content, okay, and upload from device.

All right. So you can't see my prompt or sorry, my window here. But all I'm doing is I'm selecting the files. I can't actually see that in the screen share. But all I'm doing is selecting these files. I had all these files in Google Docs. I exported them all as PDFs, and now I am uploading them. All right.

So now you'll see it's processing the content here. All right. And it is done. One thing I like about Claude Projects is it gives you the percentage of the knowledge-based size that you can use. So you'll see it's about, and this is not. 100% accurate, but I think you can get just under about 500 pages of just documents. So actually pretty decent size of a knowledge base that you can work with.

All right. And then from there, I am now done. Okay. So now when I want to go use this project, all I have to do is just go inside of my projects. There it is, AI News Assistant. I click it. And now I can just chat with it. I don't have to do literally anything else.

Okay. So now let's go ahead. And I did build the exact same thing inside of chat chvety. Literally, it is verbatim the same. All right. So we're probably going to be toggling between the two. The only difference is I just changed the name when I said, you know, project knowledge. I just said your knowledge files in chat chbt.

Otherwise, it is the exact same. So I just wanted to do some comparisons here. so everyone can see, okay, are these cloud projects? Good. All right. So let's go ahead and get started. So I have some simple, let's go ahead and bring the, there we go. I have some simple questions that we're going to ask.

Okay? And we're going to start super simple. So I'm starting and I'm just saying, please tell me the latest news on large language models. All right. Sorry. had a little internet hiccup there. All right.

So I am entering this into Claude. So now what's happening is it is going through those custom instructions and presumably it is reading through the document. So it says thinking deeply, stand by. All right. So the thing I like by default, again, this is with zero prompt engineering, right? This is just a very quick, quick and dirty example, so to speak. So I like that Claude tells me exactly what it's doing.

It says first, I'll check the most recent document, AI news docs August 24, as it's likely to have the most up-to-date information. All right, I'm going to run this exact same prompt inside my GPT that does the exact same thing. All right. So first, high-level overview. I mean, let's see. So inside Claude, it's saying Gemini 1.5, pro tops chat botarina. That's good. meta-Lama 3.1 release, good.

Mistral's new model, large 2, good. All right, so did a good job. Let's look at chat, GPT. Let's see, same thing. So it's talking about meta's llama 3, mistral large 2, open AI and copyright issues, Microsoft's new AI model, MAI 1. That's actually a little bit older.

So not super recent, but again, I didn't give it context. Recent could mean anything, right? When I'm just saying, please tell me the latest news. All right. So I think both models there did a pretty good job. All right. Now, let's kind of test its reasoning. So this is where project and project files really can, I think, help you.

Because what I'm asking now, it's not tricky necessarily, right? Let me go ahead and get rid of this little banner here at the bottom so you can see. So I'm saying, what happened with Apple in their AI? Again, I'm not being super specific. I'm not giving it prompt engineering. Right here, I am actually testing Claude Project's ability to not reason, but to think rationally. Because if you know anything about Apple in their AI from May to August, it's been wild. There was all these rumors in May about what they're doing, what they're not doing.

Then they announced it in June. And then in July and August, there's all these setbacks. it's been delayed. They might not be able to use Apple intelligence, which is what it's called in the EU. So there's been a lot of development. So now I am just openly asking Claude projects, what happened with Apple in their AI? Give me a comprehensive overview, but keep it short and factual. Please do not waste words, right? I don't want to spend an hour going through a huge project here. All right. So now I'm sending this to Claude projects. And I'm seeing because in theory, I would want it to go through every single month, right? May, June, July, August. We'd have

four months of AI news. And I wanted to hopefully go through and like a human would, go through and understand what's actually going on. Full disclosure, if a human had no idea about this and to give me a good answer, it would take them probably a couple hours, right? A hundred plus pages of AI news is a lot to take in. Claude obviously got the job done in five seconds, right? It's done. So let's see how it did here. So it's starting by talking about Apple intelligence announcement, June 2024 unveiled. So it's given me some bullet point headings. So the first heading is Apple Intelligence announcement was in June 2024.

Integration with OpenAI, which it says is July 2024. Technically, that was in June. So is it right? Yes. Is it wrong? Yes, as well. So, okay, now we're talking about, you know, Apple gaining seat on OpenAI's board. That's true.

Then in it's talking, the third headline is talking about partnerships. from July to August, potential plans to integrate other third party models, correct, EU restrictions, there we go in August 2024, did a good job. So overall, overall, it did a really good job. The only thing that I don't think Claude projects did a really good job on is, is really that May to June, which was pretty important because there's a lot of uncertainty rumors going all over the place with what was going on with Apple intelligence. So overall, Claude, I think, did a really good job. Again, this would take something a huge that's uninitiated to what Apple is doing, this would take a long time.

Even if you follow, you know, AI and Apple, you probably wouldn't know this. Even if you read our newsletter every day or listen to the podcast every day, that might be difficult. All right. Let's run the exact same thing inside of chat GPT. Again, this is literally verbatim, the exact same knowledge base, the exact same instructions. So let's see how chat GPT and GPTs do. Again, custom instructions, custom knowledge base, and it's going through.

All right. So let's see. So first one, it says Apple intelligence. Okay. So that's good. Again, but starting off with the keynote in June 2020. Then it's talking about AI partnerships. Same thing.

You know, about potentially Apple working with meta and Google in the future in addition to Open AI. So nailed that. Privacy and Safety, talking about some things with the Biden administration. new AI features for Apple intelligence. Let's see here. Yep, this is all correct. So it's talking about new iOS 18 features due to the AI.

Then it's also talking about AI chip development. So chat GPT went a slightly different route. I would say if I was just comparing head to head. Again, this is with a terrible prompt, y'all. This is open-ended, right? If I had something a little more pointed in the questions that I was asking, I think I could do a real legitimate like apples to apples. comparison. And hey, live stream audience, let me know if you can still hear and see. I think I'm

having some internet issues here. So hopefully you can. But I think they both did a good enough job, right? I'm not going to say one failed. One's better than the other. If I had to choose, I would say the Claude Project is slightly better. But again, this is an open-ended question, not great prompt engineering. I'm not asking something super specific. All right, now we're just going to do a slight kind of needle in a haystack test. What that means is, let's see, in one of the files here, all right? So this file is June 20, 24, on page, let's see, 20 of page 31 for this month. It's talking about OpenAI's new CFO and CPO.

So essentially this, I'm asking one very specific question that is buried in this project knowledge. So more than 100 pages of documents, I'm asking for an answer that can only be found in one place. I'm not telling it where to look. This is another thing. You should always be doing needle in the haystack test when you're using, whether it's Claude Projects, whether you're using GPTs and chat, GPT, you always need to be testing it because you need to write trust in transparency when working with models is important. You should always be doing ongoing testing to make sure these models are correctly pulling in information. All right. So let's go ahead and test this out in Claude Project.

So a very specific question, and I'm not telling it where to find. So it's saying to answer this question accurately, I'll search through the documents chronologically, focusing on news about Open AIs executive appointments. All right. So it got to it pretty quickly, right? So there's the answer. CFO, Sarah Fryer, CPO, Kevin Whale. All right. So pretty good job. Claude found it pretty quickly.

All right. So now let's ask the same thing. Again, doing a little needle in the haystack test with the custom GPT. So you'll see right away, it's saying searching my knowledge. I asked the exact same question. And very quickly, chat, GBT found it as well, said appointed Sarah Friar as the CFO and Kevin Whale as the CPO. All right.

So there you go. There is a very quick overview, at Lisa, basics. Now, I want to do something a little more in depth, something that is maybe a little more relevant for you or your business, because you might be thinking, all right, why would I want to use this as a research assistant, right? I could just do that on the internet. I could just use perplexity, et cetera. Okay, cool.

Well, that was just an easy way to look at it. All right. So now I also created GPT. Well, actually, let me make sure. I think I created a GPT for this. Or maybe I just, okay, I did. Cool. All right.

So now we're doing the exact same thing, except instead of an AI research assistant, here's what we're doing. I uploaded all of everyday AI's stats into this new project. We're not going to build this one live. It's already built. All right. So essentially what we have going on here,

I'm looking at the custom instruction. And again, is the same for Claude and ChatGBT. But I'm saying for this chat, assume the role of a helpful data analyst and business strategist for the company, Everyday AI. I give a little bit of information about Everyday AI. We're a daily newsletter podcast, live stream, helping people grow their career. Then I'm saying, in your project knowledge, you have a set of CSV files. These files are exported from Everyday AI's different platforms.

And then I'm essentially telling it, here's the data that you have. You have Google Search Council data. So this is essentially how people are finding everyday AI on the web and what pages they're going to. Email data, which is a big part of what we do here at Everyday AI. Yeah, maybe you listen on to the podcast. I think our newsletter, if I'm being honest, that's probably our best platform. It's written by me, a human. It is, you know, there's dozens or hundreds of AI newsletters out there.

We're the only one that brings you everyday, exclusive insights that you can't find anywhere else because we recap our podcast every day. And our podcast is unique. and we bring on experts from across the world. So right here, I'm giving the chat to, or sorry, I'm giving Claude projects access to all of our email data. Not your email addresses, don't worry, we didn't put that in there. This is open rates, click-through rates, trends, et cetera, right?

We have a lot of data. So this is our data from Beehive. That is our email service provider. And then podcast data. So I'm putting in there. So I'm saying these are stats from our podcast. We use BuzzSpout to distribute our, our podcasts on all major, major platforms, you know, Apple, Spotify, all these other places.

So then essentially I'm saying for this task, you will answer the user's queries by always first analyzing the correct file. So this is another thing, right? I'm telling chat, or sorry, I'm telling Claude projects, hey, depending on what the user asks you, you need to find the correct file. You know the information you have. You have some website and Google data. You have some email newsletter data. And you have have some podcast newsletter data. So if the user is asking about one of those platforms in particular, you need to go and make sure you look at the right data.

All right. And then I'm also saying when the user asks you to use artifacts, always use the artifacts feature to try and best visualize the data that the user asked for. More on that in a minute. And I did the exact same thing inside of chat, GPT. But instead of artifacts, I said advanced data analysis.

All right. So that's it. So again, think of, your company, right? Think of your company and how you could use something like this. Again, I'm fine giving all of this data to a large language model. I have paid plans of Claude and chat GPT. I turn off model training. All right. So I'm not too worried about it. Don't, like I said, don't dump in a bunch of PII, PHA confidential things unless the person in charge at your company is like, oh, yeah, we're on an enterprise plan. We can do that. All right. But I've,

uploaded all of this data. There's a good amount of data. I would say we have hundreds of thousands of data points, all right? And we've used 73% of our knowledge file here in Claw. So actually a ton of data. All right. So think, think. Maybe you work in data or you work as a business analyst.

This is a good chunk of what you would be doing at your job, right? Conversely, let's say maybe you don't have a dedicated data analyst. You don't have someone working in business intelligence, but you would love, right? You know you have access to all this data and you would love to do something with it. Here's a great example, dumping all of your analytics, all of your data into a single project or a single GPT, and then just being able to have a conversation, right? So this is literally as if you were to hire, as an example, a data consulting company, a business intelligence company, a digital consulting company, right?

Because you're like, man, we have mountains of data. and we have no clue what it means. This is what you used to have to do. Guess what? Large language models are fantastic at this. They are so, so good, especially with artifacts in advanced data analysis. So I'm excited for this one.

Enough chit chat, y'all. Let's jump straight into it. All right. So let's go ahead and run some general prompts here. Okay. So what I'm saying, and I'm going to zoom in just a little bit, I'm saying, tell me the 10 most impactful trends that I need to know in order to grow everyday AI.

I am intentionally being vague in general about this. I always tell people when you start with large language models, when you're having a conversation, right? The other thing about this is I'm going to be chatting with just this one project. So as I have a conversation with the project, the same thing applies for the GPT. all of the conversation and all of the insights that we on earth, they're going to be in the context window of the conversation, right? Just like a human. When you sit down and ask a human questions,

presumably the human is going to be able to remember what you're talking about. Unlike if you're using like a Siri or in Alexa and you ask one question and then you ask a follow-up and it fails, that's not how at least AI smart assistants, voice assistance work. Large language models work iteratively, right? So you can ask general questions and then ask follow-up questions without the need to clarify, right? So again, think of you are chatting with a team full of statisticians and business analysts, right? So I like to start general. And I like to see how good is the model at finding some of these things on its own without me pointing it in the right direction.

So I'm not even saying go look at the podcast, go look at the newsletter, go look at the website. I'm just saying, find me trends. Okay. So there we go. and then I'm going to run this exact same prompt inside chat tpT. So if you join on the podcast, I'm just doing a bunch of window flipping right now. All right. So Claude Projects right now says pondering standby.

This one might take a little bit of time because it can go in a number of different directions. As an example, Claude just might give me a bullet point list or it actually might start building graphs using the artifacts feature. It will go different way each time. All right. So right now it's saying, yeah. So you'll see now it's using the. artifacts feature. So if you are brand new, I love the artifacts feature, right? If it wasn't for the artifacts feature in Claude, I wouldn't use Claude if I'm being

honest because I think the GPT40 model is better. ChatGPT has internet connectivity. GPTs are more flexible and robust, but Claude has artifacts, right? So essentially, to talk very plainly, right, it can not just run code, but it can execute it. So it can run and execute HTML, CSS, Python, right? So in theory, with the artifacts feature, you can literally build applications and not have to go, you know, interactive applications as well, right? And not have to go into a third party platform or, you know, have your own web hosting. You can literally build business dashboards, interactive things inside the artifacts feature. With GPTs, it has advanced data analysis.

and you can do interactive graphs and charts, but not like dashboards, okay? So all that used artifacts for right here was to essentially just give me a list, right? So it didn't actually, you know, run any code or anything like that. So another thing is I can add this to the current project, right? So I can click that.

And now what this just created is now in that knowledge base, right? but if I keep talking with this particular chat, it will stay there. However, I can start a new chat with this same project. And now that information that Claude just gave me will stay in the project knowledge, which I like that. All right. So I'm just going to go over there very, I'm going to go over these very, very quickly. All right.

So Claude says, number one, mobile optimization is crucial. It's giving me some mobile versus desktop as well as action, right? So maybe I'll just go through the first one. And then I'll just read off some of the rest. So essentially it's saying mobile devices accounted for a significant portion of engagement. 24% of clicks came from mobile devices. It has a higher click-through rate compared to desktop. All right.

So essentially it's saying make sure the action is giving me is ensure all your content is mobile friendly, including emails and website designs because it's telling me the majority of your engagement right now for everyday AI is coming from mobile devices. All right. So then it's also talking about email marketing effectiveness, geographic expansion opportunities, podcast, growth potential. Right. So it's giving me a lot of good information that I'm going to read later.

All right. Let's jump over into chat GPD. So one thing I love about chat, GPT, is the new advanced data analysis V2 is crazy. It is so good. And they actually just two days ago updated even the V2 of advanced data analysis. So again, it can't do everything that Claude artifacts can do. in terms of, like I said, you know, it can render code, right? It can render HTML.

It can build a literal website and show you how the website works. ChadGBTBT can write code, but it can't render the code unless it is, you know, spreadsheets or, you know, essentially interactive tables, data's, data visualization, right? But it can't create like websites or business dashboards or apps, right? You can literally create apps, web apps, actual apps, games inside of, Claude artifacts without even having to take the information anywhere. So one thing I love about GPTs that I think it does a better job than Claude is for the most part, it shows you its work.

Right. So right here, I'm getting a breakdown of exactly what it's doing. It's telling me step by step what is doing. It says, I'm loading, you know, I'm loading and analyzing this file. And then it's telling me what it's looking at. It's giving me the Python code. So I can actually go and see, you know, I can. read basic Python. I'm not great at writing it, but I can literally see what it's doing

step by step under the hood. This is fantastic. All right. So let's go ahead. I'm going to do the same thing. I mean, look at this from chat, GBT. It's, I love, I love the way that GPTs and advanced data analysis show you the work. All right. So I'm going to, I'm going to read the number one and then just kind of go through some of the rest. So number one, this is in the GPTs. Again, we're just doing some comparisons. It says high open and click rates and emails. So it says certain email subjects like, you know, this is actually the one from yesterday. Amazon's Titan V2 image model is here. Show exceptionally high, open, and click rates. And then it gives me action. So interesting. It actually gave me GPT and Claude projects. Give me a very similar format. It essentially gave me an

insight and then an action, even though I didn't ask it to, right? So the action here, continue to create engaging and intriguing subject lines and content similar to these successful emails to maintain high engagement rates. All right. So there you'll see. Chat GBT GBT GPT gave me, you know, number two, top performing URLs and email. Three, effective subscriber acquisition sources for popular search queries, drawing traffic, right? High performing pages, right? Just looking at these eyeball test, I actually like the, I like the insights from chat GPD a little bit more, right? So let's just say as an example, right?

So number four and five, it's a little more specific from the GPD. So it's giving me popular search queries driving traffic. and high-performing pages. So it's giving me very specific and actionable insights around SEO, right? Whereas in Claude projects, it's a little more vague. It's just saying like search engine optimization focus, right? So it's giving me kind of broad, which is okay because I didn't ask for specifics. But I like a model's ability to, because if I'm asking it for impactful trends,

just telling me to focus on SEO, not great, not very impactful. All right. Let's go ahead. We're going to run a couple other prompts here. All right. And we're going to go through these pretty quickly because I know this episode is running a little long, but I wanted to give everyone really a good idea of Claude projects, but also to compare it to chat GPT.

All right. So now for this one, I'm saying, please give me 10 areas where I'm trending in the right direction and 10 areas where I'm trending downward, right? Focus on big wins or potential big losses. be creative and strategic in your approach. Be ultra-specific in your result, but do not waste words and be succinct, factual, and creative. All right.

So again, similar. I'm not asking for specific, oh, give me this from the podcast, give me this from Google search. But I'm saying, give me 10 areas where things are going very well. And give me 10 areas where things are trending in the wrong direction. So presumably what this requires of clawed projects and also GPTs is to do some reasoning, right? to do some logic work because like I said, there are hundreds of thousands of rows or

sorry, hundreds of thousands of cells of data in this. And that's a lot of work, right? Again, if I were to hand this off to even a company that specializes in this, number one, it's going to be expensive. Very minimum to play ball with these type of companies, five, six figures easily, right, just to be able to start working and or start engaging. Look at this. This is good stuff.

This is good stuff. And this is first, very first question. I could do 10 follow-ups and really drill down. All right. So Claude did some great work here. So give me 10 positive trends. So trending positively, podcast growth, email open rates, mobile engagement, you know, international expansion, good stuff, areas of concern, tablet engagement.

Okay. Geographic gaps. So, okay. It's like, all right, saying, you're not getting engagement from Japan and South Korea. podcast download consistency. That's normal. Email click rates. All right. So it's giving me some essentially pros and cons there. Let's look at chat GBT, the custom GPT here. So things that are trending in the right direction, high email open rates, top URL, clickthroughs, subscriber growth from specific sources. All right. Some of the downsides here, let's look. Stagnant open rates

on some emails, declining click through rates, underperforming subscriber sources. All right. So Again, I think both of them pass the test here. All right. So I think Claude projects and custom GPTs did a great job, right? And again, I would want to have much more information and much more depth. But I specifically said, yo, smart AI models. Give me high level stuff here, right? Because then presumably I would then be having a conversation based on what they say and really drilling down into those topics.

All right, we're going to do about one and a half more things here. All right. So now here's where we're really going to start separating what's currently capable in Claude Projects versus GPTs. All right. So now what I'm asking is I'm saying, based on the data that I have, please, please give me an idea of five data visualizations or dashboards that you could build for everyday AI. That would be the most helpful. Yeah.

Here's where we're going to start to shine. I save the good stuff for last, right? If you're here after 55 plus minutes, this is where we put the gems, y'all. All right. So what Claude is actually doing right now for our podcast audience? It didn't, which it's like, okay, can I really be mad at Claude for doing this? So I asked specifically for ideas of data visualizations. And Claude just essentially said, yeah, you know what?

I could give you ideas. So it gave me five ideas, but it went ahead and just built one of them, right? It built me. Let me zoom out a little bit here. So it built me literally an interactive dashboard for podcast performance tracker. I love this. So it's giving me episode numbers and then it's giving me the number of downloads for our last couple of episodes for some of our most recent episodes.

Y'all, that's pretty amazing, right? Again, this is an interactive dashboard, simple, right? And all I did was say, hey, give me five ideas, right? And you can really build inside Claude projects with the artifacts feature, some very impressive data visualizations in interactive dashboards. And here's the thing. With Claude projects, you can then share these with your team. I didn't even get to that.

I wish Claude would change their rules a little bit more because you have to have minimum. You have to pay for five seats, essentially for their team plan to then be able to share these with your team. So you're talking about it at $30 a month, or maybe it's $20 a month. It might be more for the team plan. Yeah, I think it's $30 a month. So I think you're thinking it's a minimum $150 a month in order to be able to do this

and share this with your team, where with chat, GBT, the minimum is just two users for a team, right? And then again, you can share GPs, though, with even free users. But this is pretty impressive, all right? this podcast performance tracker that Claw just literally went ahead and built, very impressive. All right. So what I'm going to do, and I'm going to do this for each of them, I'm going to have them essentially try to build me one in more depth and detail. So for Clawed projects, I'm asking, I like this idea that it said SEO performance

analyzer. So I'm just going to say, please build number four in artifacts. and I'm going to say making it more in-depth, detailed, and useful. All right. So very, very broad, you know, instructions there. So let's see actually what it does. All right. And then we're going to jump over real quick to chat chbt.

So again, it gave me five different ideas for things that it could visualize. So advanced data analysis or code interpreter, as it's sometimes called, inside chat dbt is a little different. It can visualize data. It can plot things, create charts. It can't create interactive dashboards, right? It can't create apps. It can't create websites and render them.

It can write the code, but you can't actually see them. So I'm going to go through here and I'm going to say, okay, same thing. So for number one, it's that email campaign performance dashboard. I don't think chat GPT is going to be able to build that the best. So it's giving me a lot of dashboards, right, which is fine because that's technically what I asked for. So first, I'm going to do a follow-up questions, and I'm going to say, I'm going to say which one of these would be best as an interactive chart or data visualization. Okay. And then what I'm going to do is I'm going to choose one of these, just like I chose one

in Claude Projects. And I'm going to have chat, GPT, via advanced data analysis, code interpreter, build one of these. Okay, so it's telling me subscriber growth and acquisition dash, So it's still saying a dashboard, which I didn't necessarily want. So I'm going to say, I don't think this is going to work very well. I'm going to say, please, I'm going to say using advanced data analysis. I hate typing live. Please build this interactive. I'm going to say chart because it cannot technically build a dashboard.

It can build interactive data visualizations. So I'm going to say, please build this interactive chart or data. visualization. I'm going to say if you're not able to, please reformat this idea in a way that you can visualize the data using advanced data analysis. All right. I hate typing live because with my setup here, I got cramped dinosaurs arms. All right. So let's see. And this is this is the last thing we're doing, y'all. I know this has been a longer episode. I hope it's been helpful, though. All right. So now let's go ahead and jump in. Let's see how Claude did. Wow. This is impressive. Okay. So again, as a reminder, I asked it to build. It gave me five

suggestions. I like this one of an SEO performance analyzer. So I said, hey, go build that. And you'll see using Claude projects, it did it. So if you're here on the live stream, y'all, this is pretty impressive here. So again, it's giving me different terms, right? So it's giving me different terms that have brought in a high number of clicks and a high number of impressions. So it's a pretty simple data visualization, right? But the good thing is, oh, wait, it built actually five of them. Interesting. Okay.

So it built me a series of different interactive graphs. Man, Claude Artifax is great. So here we have a simple bar chart for top performing queries. Again, I can hover over them and it's giving me more information. Now it's doing a little scatter plot for click. through rate versus average position. This is super helpful, y'all. This is some stuff I didn't even know.

So I can, let's see if I can hover over. Okay, so it's not naming everything. So the first shot on this, not super helpful, but I would probably go through and iteratively say, hey, I'm not getting labels on these, you know, average positions and click through rates. That's okay because that's an easy fix. So now it's saying device performance comparison. in. So it's showing me, let's see, clicks and click through rate. So, okay, I didn't even know this.

So as an example, okay, this is interesting. Much, much higher click through rate on mobile than we do on desktop. So we have more, way more overall clicks on desktop for our Google search council, but a way higher click through rate. So that's pretty interesting there. Jeez, I mean, y'all, live stream audience, are you guys impressed with what, again, this is inside of a project so I can share this with my team, but it's using the artifacts feature. This is now, geez, this is good. Okay. An interactive chart here, it's a little line graph, and I can hover over, and it's showing

me impressions and clicks by day. Very, very cool here. So it's giving me the number of clicks or impressions by day. Very cool. Now it's going over to top performing pages here again. So Claude literally just built me five interactive graphs and dashboards, right? And in theory, if I wanted to, I could click this code button. I could copy this and I could put this in a live production website where I didn't

even need someone to be logged into Claude. That's the other thing. It is both giving me the code if I wanted to do something and build something with it, but then also I can just grab it. All right. So let's see. All right, let's see how ChatGBTD did. This is a lot, right?

showing me its work, a lot of Python here is gobbling up data. So let's see what it actually built. So again, I asked it, let me go back up here real quick. So I said, please build this interactive chart. And the one that it gave me was the one for subscriber growth and acquisition dashboard. All right. So scroll down here. Let's see what it did using advanced data analysis and code interpreter.

A lot of data here. lot of data. Okay. So you'll see right here, it gave me way too much data, right? But it's also extremely impressive that it brought all this in. So here's what I'm going to say. I'm going to take a screenshot of this, right? I'm going to drop this screenshot into chat chbtee, and I'm going to say, this is too much data and I can't read the labels. And I'm going to say, please maybe only focus on 10 so I can easily read all of the data. Right. So that's another thing.

So these models can see. So I don't have to go through and be like, huh, let's say you're using this to build a dashboard for your team or an interactive graph that you can share with someone. And you're like, how am I going to troubleshoot this? It looks like hundreds or maybe thousands of different data points on this chart.

How can I work? Well, upload a screenshot, be like, yo, this is all jumbled. It's going to see, it's going to understand, oh, yeah, I put way too much information on that chart and you can't make it, you can't make use of it. Right. So now it's going through and it's recap. And now look, now we have a pretty, pretty beautiful, nice little chart here inside

of chat GBT. And then the thing that I love is a lot of the different graphics within chat GBT and data analysis. you can just chat with certain things. So it's showing me top new subscribers by top 10 acquisition sources for our newsletter. So now I could continue. And it's a nice looking graph. So it's not as interactive as the one from Claude Projects.

But again, y'all, I just showed you live in real time, how powerful Claude projects are. Right. Again, that data, those visualizations, those dashboards, y'all, especially with Claude artifacts. This was the fact that this is capable and available to you, if you pay for a, you know, $20 a month, Claude Pro subscription, that's wild, right? Again, to either to get this type of data, to get these type of insights, to get these types of visualizations, and dashboards a year ago, it would take you a lot of money, a lot of time. And now look, y'all, I did it live on a podcast with very little effort, right? You saw that.

We didn't have to put in super structured prompts and go through all this prompt engineering. Y'all, if this didn't just show you what large language models are capable of, I feel sorry for your company. I feel sorry for your career. If I'm being honest, right, think of all the publicly accessible data that you have, right? Here's the thing, y'all. You don't even need permission to use data that's publicly available. So many companies don't even know how much data they have publicly available, right?

Companies that are public, they have to put out their 10Ks. Think of all these interactive dashboards, all of these insights. As an example, here's a company's, you know, 10K financial reporting for the last five years, you know, give me detailed breakdowns and trends in a dashboard that's easy to use. Bam! Think of all the different applications for this and how easy it is. Y'all, this is both, well, if I'm, let me be honest here. I think there's equal amounts of excitement.

And there's equal amounts of like, okay, doom. Let me explain what I mean. If you are not excited by what I just showed you, you got to check your pulse. from being honest, right? This level of business intelligence, of automation, of AI doing the grunt work, this is truly mind-boggling. I say this because I've spent, right, I've spent, as an example, probably 20 hours going through pre-generative AI and putting together a very rudimentary version of what you just

saw, right? I would go through, I would, you know, read long PDF reports, grab all this data, try to put it into a little graph. It would look very ugly and it would. take me like 20 hours. Do this in 20 seconds. This changes. So from an opportunity,

whether you're an entrepreneur, whether you're leading your department, if this just didn't make you think that I should change the way that I'm working, that our company should really take large language models seriously, on the flip side of that is doom. If you're not, I'm being honest.

If your company is not already building on top of these APIs for large language models, if you're not already using them in some way, shape, or form, if you are not personally going in here, even paying, right? Oh, my company won't pay for it because we don't, okay, well, pay for it yourself. Pay for it yourself and use publicly available data and information. This changes what humans are capable of. It changes how business operates and it completely rewrites the business future. If you are not using large language models, if you're not using Claude Projects, if you're not using GPTs, you're putting your own career, you're putting your own company at risk.

You shouldn't do it. All right, y'all, I hope this was helpful. If you're still listening, share this episode. I got a little something special for those of you that do share this. So if you're on Twitter, I think you can like retweet it or whatever. If you're on LinkedIn, click that repost button. All right. I have, I'm going to record a video on some even more specific examples.

and use cases that I think are great for Claude projects and for GPTs. All right. So go ahead. If this is helpful, if you're still listening, click repost on this. If you are on the podcast and you're still listening to this, go check your show notes. I always leave a link to the LinkedIn kind of live stream for this. Go click that repost. I'm going to be creating a video going over some more specific use cases that I think really

have the potential for explosive growth. So I'm going to be going over some of my favorite use cases for Claude projects and for chat GPTs, GPTs. All right. So repost this if this was helpful. I'll get that over to you probably early next week. I got a little stuff going on the rest of today. So give yourself some time.

Go repost this. But if you're listening now, do it live. Also, if you're listening now, you need to go sign up for our thanks a million giveaway. If everyday AI is helpful for you, all you have to do is sign up. It'll be in the newsletter. It's on our website. I think, I'm pretty sure if you refer like three friends, you're in first place. So, you know, it's essentially a referral giveaway.

You just sign up. You get a link. You send that to your friends. If your friends sign up for the everyday AI newsletter, you earn points. And like I said, I think if you refer three friends, you're in first place. So we're giving away a year of chat chbt or a year of Claude Anthropic, right? So a personal paid subscription for a year to your favorite large language model. And we're going to be talking about a lot of other prizes.

One of them will be a 101 consult with me. So if you find everyday AI helpful, if you want to maybe what you saw today, you are rowed by it, but you don't really want to pay $20 or $30 a month. That's cool. We will pay it for you. So go sign up and refer friends to our thanks a million giveaway. And we'll see you back tomorrow for more everyday AI. Thanks all.

Meet Firefly AI assistant. Now live in Adobe Firefly, the Allman One Creative AI Studio. Just describe what you want to create in your own words and the assistant handles the rest, orchestrating multi-step workflows across Adobe Creative Cloud apps, including Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome while the assistant accelerates execution. Stand control with the ability to step in and refine at any time. See it today at firefly.adop.com.

And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit Your EverydayAI.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.