The AI Tool Whiparound

The AI Tool Whiparound Transcript:

Date of Live Training: October 22nd, 2024

Welcome to the AI Tool Whiparound. This class is  a whirlwind tour of the major AI tools. It’s going to save you hours of playing around with different tools. I’m going to give you ways  to decide which tool to use and when this is me.

I’m the founder of the click. by attending this session, you’ve become a leader in the A. I. Marketing world. there’s just not a lot of marketers out there that are learning the proper way to use A. I. Tools and congratulations on being here. I would advise you because I’m biased and because I have kids.

And one of them is in college. I would advise you to join my paid program, which is called theCLICK PRO, to help me pay my bills. but this is for marketers that aren’t dabbling with AI. They are convinced that AI is something that they need to learn and they’re willing to invest in it. It’s 348 a year.

And I just pile tons of training like this one. into the ClickPro. We’d love to have you in there. A lot of the people on this call are already members of theCLICK PRO. And if so, I hope you’re getting a ton out of it. And I hope you get a ton out of this training today, because here’s the thing I’m working with no net.

I’ve got a bunch of AI tools open on my in browser tabs right now. These tools do wacky things all the time. Sometimes they don’t do what you thought they would do. you’re going to watch me whip around and do a whole bunch of stuff with AI today.  And learn a lot, in the process in terms of how to choose which tool and when to use it.

So which tool is best at this and which one’s best at that. the first thing I want you to understand is that there are eight faces of how marketing is evolving.  there’s eight ways as marketers. I want you to remember that everything that we do at the click is about marketing. We’re not worried about AI in any other capacity except how it impacts marketers and marketing.

AI is evolving quickly. It’s going from incompetent to partially competent to competent in all these different aspects. And there’s some where it’s still, it’s still like the The, the cave man back here, and there’s other ways that AI is already extremely competent. It’s more competent than you or I could ever hope to be.

It can accomplish a task with better, accuracy, quality, and with lightning fast speed. And there’s other things that it’s just. somewhere in the middle. It’s kind of okay at doing this and doing that. And as marketers, we need to be paying very close attention to what the AI is capable of broadly, but also what each tool that’s out there and available is capable of and what’s evolving today and going to evolve tomorrow.

So there’s  eight faces of. AI evolution, the way that AI is evolving in terms of how marketers need to be thinking about it. And only one of those ways is in the area of intelligence. One of them is that the models are getting more intelligent.  even if the AI models don’t get any more intelligent than they are today, which they will, there’s nobody seriously working on AI that is saying we’ve hit any kind of ceiling when it comes to its, intelligence.

even if it did stop at the present level of intelligence, there are still seven other ways that we are going to see AI continue to evolve and become extremely disruptive to the marketing industry. as a result, our jobs, right? the way we make money. So I think it’s extremely important for us to learn these things.

So what are these eight  ways that we need to be watching AI? Because each one of the tools that are out there have a different sort of. score on these different  areas of evolution. the first one’s intelligence, right? we know that AI is at some level of intelligence and the companies that produce AI are working to make it.

smarter and smarter. This is the newest version of Anthropic’s tool called Claude, if you haven’t played with Claude,  it looks like this, this is the newest,  benchmark data from Claude. To be honest I don’t even know what half this, means 

Math problem solving, high school math competition. I have a sense of what it means, but there are very specific benchmarks measuring these AI models you can see, That according to this benchmark data, Cloud 3. 5 Sonnet’s newest model has seen jumps in its, level of intelligence.

each time a new model comes out, we see new benchmark data like this. And if you want to go and dig into all of this, you can and figure out, you know, what does that mean? High school math competition? Like, what does that mean? Why? What is 16%? Zero shot. chain of thought.

What does that mean? So you can go and dig into that. But as a marketer, what you should know, and what we should all know, is that these models are going to continue to get more intelligent. amongst all people seriously working on AI that I listen to and watch, they’re saying that we aren’t seeing a ceiling as of yet in terms of how intelligent it could get.

The second way though, forget the intelligence. Let’s just say it never gets any smarter. One of the other ways that we are seeing AI.  evolve is in its usability. how usable is it? Like right now when we use AI,  we go into something like chat GPT and we start having a, chat conversation with it.

this interface is not AI, This is simply the interface that you are using to access. the underlying intelligence that is chat GPT, So usability, what we can expect to see and it’s slowly happening. We are going to expect to see Increased usability, increased ways of accessing this intelligence.

So for example,   in terms of usability,  we can expect to see increases in AI’s vision. today, it’s partially competent. at seeing things I’m going to demonstrate. Some of that today, very soon the AI will, be able to see through your camera.

It already kind of can, but, it’s early days,  It’ll be able to see through glasses. We’re already seeing that with, Meta or Facebook’s, Oakley glasses that have built in AI and vision. vision is one way that we are going to see.

Evolution. We are going to see evolution in voice. So today, the only model that’s doing a really good job of both taking in and putting out voice AI, in my opinion, is ChatGPT or OpenAI’s ChatGPT. If you’ve played around with the mobile app, it’s actually pretty good. I would still give it a, C minus in terms of all the growth that is available to evolve in terms of voice capability, then we have web connectivity.

So  the only, the best model in my opinion, and the best overall model still today is chat GPT.  and chat to beauty four. Oh, even though I’m going to show you why Claude is a really good solution as well. if you’re just getting started with AI, for particularly I would start inside of chat to BT.

here’s a demonstration quickly inside of. chat GPT of two things. One is increased usability. if I hit the slash button there’s the ability to create a picture,  search the web, or use its reasoning model. If I hit that slash, it sort of activates a, short code to pop up this picture search or reason.

So The first thing I wanna demonstrate is that this is an increase in usability. This didn’t use to be here. You didn’t use to have the ability to just hit the slash and activate its search capability of the internet. that’s a pretty new usability function, and it’s, it’s one little tiny example of how AI is gonna continue to advance in terms of its usability. 

But I also wanted to demonstrate that chat, GPT has the ability to access the web. So if I hit search here, I can  outline the contents.  Of this sales page and I can give it my sales page for theCLICK PRO. 

and it’s going to go to that URL and outline the offer  you see that it’s coming back with,  exactly what is offered on that page it can also do things like, today I can hit slash and search and say what teams are in the world series.

So it’s accessing real time data. That’s important for a lot of things that we do in research projects and things like that. But we want it to have, let’s see if it gets that. So it does know that the Yankees and the Dodgers are going to be in the world series. That’s a relatively new bit of information 

Open A. I. Is is actively indexing the web the same way that Google indexes the web,  you know, basically constantly. So it’s important to recognize this evolution and usability of, web connectivity,  it’s a change in how and what use cases, because now I might use chat GBT instead of using a different tool, which I will tell you later what that tool would be still.

 

but I might be more likely to hit the search button and say, you know, what are the,  the latest. SEO trends and and be confident it’s going to find the latest stuff and not have a cutoff like it used to have where it’s training data cuts off. And it doesn’t know anything before some date, right?

 

that’s how it used to work. But when we activate this search function, we can ask it questions like what are the latest SEO trends and feel confident that it’s actually finding. These latest SEO trends, even a month ago, I would not have used chat GPT to ask the latest SEO trends.

 

I would have used a different tool. because it did not have access to a real time update of what’s going on.  This is an evolution of this tool. So I want you to be paying attention to intelligence.

 

How intelligent are these people?  Can it see, can it talk? Can it hear, can it access the web?  Can it build applications? this is a good time to jump into Claude and introduce Claude. if you want to take one thing out of this, training, it’s going to be that the two models that I’m going to recommend to you are always going to be right now, as of October 22nd at 1 14 PM central time, I would tell you that you’re going to be using chat GPT, or you’re going to be using Claude.

 

it’s sort of me, my workflows to bounce back and forth. And then there’s a third tool. Really a third and fourth tool that I’ll show you that I’ve, I’ve been using more and more and more so  you, but with Claude,  Claude actually has the ability to create real actual applications. like for example, if I was a marketer that wanted to create a lead generation quiz, so I might say create a lead gen quiz with  five questions.

 

That determines someone’s,  content marketing,  skill.  So maybe I want to create a quiz like this. let’s see if Claude does this for me, with that simple prompt. it pops up this window an, a Claude artifact, and it is starting to create this quiz.

 

It’s writing the code. And it’s going to actually create this interactive quiz. if you’re a member of the click pro, we did an entire training on how to use these artifacts to create applications like we’re doing here. what are the pros and cons? What are the ins and outs? And then we have exercises that go along with that inside the AI marketing sandbox training in the click pro, This is an application. it’s early days here, but this is where the AI is going, where we are going to be able to use the English language to tell the AI to build applications, removing the barrier between your creativity. And whatever you want to create, because you’re not going to need a coder to do that anymore.

 

You’re going to be able to just say, Hey, because now I could go and I could say, make it eight questions  and I could give it the questions that I want it to use. And I could, do anything I want to this app. I could just direct it using the English language. And that is an example of.  The A.

 

Usability and features evolving so that it can build apps. And then the last big thing is fine tuning, and I’ll get to that in a little while. So I’m going to get to that last today.  So  the next way that I want to discuss really important way that we are seeing a I evolve.  Is through its multi modality.

 

So this is a term we use to describe the types of data that an AI can take in as input and the types of data that it can put out as output. can your AI tool  take in text? And put out text. Well, yeah, we know it can do that, right? We know any AI model can take in text and put out text.

 

That is one mode of AI. But can your AI take in text and give you back images? Well,  if you look at Chachabit, the answer is yes. Like  I can have it create me a picture with Dolly and I could say, 

 

What was I, what was I talking to it about? SEO trends?  Let’s ask it, you know, create me a picture of a  woman doing SEO work at a laptop and this would be an example of the mode of text to image, right? So text in, image out, right? So you can see that Chachapiti or DALI, it’s DALI model is creating an image here.

 

And that is a mode, right? So I text in image back out and how capable is it? Like, can I go and make changes to this? Can I highlight areas of it? all of those things are evolving. every day we are seeing evolution in modality.  There is no question in my mind that all modes are coming to an AI near you very soon.

 

So you’re going to be able to go text to video, image to video, image to audio, text to code. we’re already capable of doing text to code. We’re already, already capable of doing text to image.  We’re very early days on being able to go text to video or text to audio. but we can generate data or, give back data.

 

multimodality is a really important part of AI’s evolution. is it able to take in and put out all six of these modes in various combinations? text to text is easy. Text to image is.  easy ish, but it’s still, probably a D plus in terms of where it’s going to be in the next year or two years  So, another big area of evolution that marketers are really watching is text to video. when will we be able to take a prompt and create a video of any real length and have any control over it. It’s similar to images in that way where we are able to create video, but you can’t create very long videos and you don’t have much control over it yet.

 

And that’s a big yet. that is watching The face of evolution that is multi modality watching for that. the next one is  customizability and I’m going to skip this one. It’s red in this image for a reason, because there are two  types of evolution on this chart that are red. the reason they’re red is because they are the.

 

Ones to watch. They are the ones to watch. customizability and autonomy are two places to keep a hawk’s eye on in terms of what is the AI capable of and what’s coming down the line. So I’m going to come back to customizability.  The next one is reliability. So reliability is a problem, right?

 

We still see a I that is hallucinating. In other words, it makes things up. It fabricates or tells outright lies.  You can call them lies, even though there’s no intent to lie. I think it just lies, right? Like it just lies. So again, another area that we are seeing improvement is in reliability.

 

Like  when we get to a point where the AIs are starting to make decisions on their own, which is going to come, and that’s when we get to autonomy  We will need to be near or at 100 percent reliability, won’t we? So, in order to automate certain tasks and let the AI make decisions and,  make choices. 

 

Number six is security, and I won’t spend a lot of time here, but security is about, data security. If I upload something into an AI, is my data secure? if I put something into its brain, like a bunch of, meeting notes is there security there?

 

is it able to protect my data? That’s an important area of evolution, particularly for large enterprises that have legal departments and are terrified of,  putting customer data inside of an A. I. For example, We’ve overcome this in every technological way. we started by digitizing people’s contact information and CRM information onto servers that were kept on property.

 

then, a company called salesforce. com, convinced the world that they could handle your Customer data, even in finance or health care up in the cloud, they created  cloud based sass, basically created the sass category  by guaranteeing and demonstrating that they could keep that stuff secure.

 

open AI, the people that make chat GPT, the people, you know, the people over at Microsoft, everybody that’s making AI is working on.  Convincing the world that they can keep our data secure.  The next important area of evolution is memory. the amount of recent information that a model can remember at any one time is what we would call the context window of that model.

 

So every AI model has a context window. It is the amount of information that it can remember. In any  one, you know, instance of it. So  for example, chat GPT fours context window was 8, 192 tokens. And a token can pretty much be,  equal to a word. If you want it to be, it’s not quite that simple. some words are multiple tokens, 

 

Longer words are going to be two or three tokens, but basically one token is one word. if you think about ChachiBG4, which launched in March of last year, 8, 000 tokens, so 8, 000 words. that’s not very big. It can’t remember that much. Now, look at  the current Gemini model.

 

This is me asking Gemini earlier, what is your context window? it says, I have a 1 million token context window. This means I can process and understand up to 1500  pages of information at a time. this is an example  of AI evolving. not because of its intelligence. It’s not its intelligence that evolved.

 

It’s its memory capability, and it opens up new use cases. For example, now I can put large documents in here, or I can put high definition video. I can put lots of audio. I can put lots of images in here. into a AI system and it can remember all of that, right? And so one million tokens in one year, this will look like a tiny context window.

 

We will move towards basically an unlimited context window where the amount of context that an AI can remember about any one thing is essentially not an issue, It can remember everything, you can see that as being an area of evolution. Now, the last area of evolution  is autonomy. 

 

So autonomy  Another one that I put as red in this graphic because between autonomy and customizability as a marketer, as someone in marketing, and really as a human being, we need to be watching what’s happening with AI autonomy and where is it going to go? Because even at its current intelligence level, we are able to create autonomous  Agentic AI, what we call agents.

 

an agent is an AI that’s able to make decisions on its own based on information. It might go out and reference different systems or talk to other AI or ask you, the human, for input or feedback, and then go off and take care of tasks. Now, I want to show you something. This literally came out today as well.

 

I mentioned that, Claude has  a new model and here’s its intelligence report, or its report card. But what I want to show you Is a quick demonstration of what Claude is calling computer use.

 

So computer use is.  A very early version of what we are going to see over the next year. 2025 will be in my opinion, prediction time, the year of AI automation of autonomy of agentic AI.  That will have big impact on businesses and people’s jobs. You’re going to need to learn how AI works so that you’re not scared of it, but instead learning how to, work these systems.

 

if you take anything away from this call, take that you need to invest in this. This is something that’s important, computer use, Is the first time that I have seen  a demonstration of just a regular off the shelf AI model that is able to  do agent has agentic behavior.

 

So it’s able to control your mouse and your keyboard. open and close tabs search the internet and take actions on your behalf, they’re calling it. Computer use. this is an evolution of AI in the area of autonomy.

 

So let’s take a look at this. 

 

Going through a quick demo today a customer the equipment company has asked us to fill out a vendor request form. The data I need to fill out this form is scattered  in various places on my computer. we’re going to ask Claude to Look at the spreadsheets, check if an equipment is in there, and if not, move over to the CRM and try and find some more information there.

 

Once it has this data, Claude’s gonna then fill out the form for us and hopefully transfer the information to the vendor form.  First thing that’s gonna happen is Claude’s gonna start taking screenshots of my screen and quickly realizes that the, an equipment company isn’t actually in the spreadsheet.

 

the first thing it does is it swaps over to A CRM.  The AI goes in  and looks for this ACME or Ant Equipment Company, doesn’t find it, and then switches over into the company’s CRM, which is where it is now, and now it’s going to search the CRM. That’s the company we’re interested in. 

 

Watch right here.  That’s the AI. 

 

Luckily, we get a search match, and Claude then starts scrolling through the page, looking for all the information it needs to fill out this form. 

 

Alright, now it’s going to take the data from this CRM and put it into the vendor request Form, which is something that, a human would do today, might be 30 minutes of their job every day to do stuff like this instead the soon like, an AI like this, we’ll be able to just handle this kind of  workflow. 

 

So you can see it, fill it out the information over here. Well then autonomously starts transferring the information across without me having to do anything 

 

and goes through the steps and fills out all the information needed and 

 

This example is representative of a lot of drudge work that people have to do. This is available in the API. We’re excited for people to try it and we should expect things to get a lot better over the coming months.  this is an example of an evolution in  autonomy,  a step towards agentic AI.

 

this is not something that is easily accessible to you or I, inside Claude even today.  you would need to hire a developer to create that workflow for you.  But again, take into account all of the Ways things are evolving. So right now, to create  that agentic behavior inside a cloud, I need to have a developer, which is a bad usability  situation.

 

And they know that  within months,  a year tops, you’re going to see better usability on this autonomous type feature, and you’re going to be able to build these kinds of flows yourself without needing to code.  the first part of this training today is designed to give you a framework around which to see how AI is evolving through the areas of intelligence,  better usability,  more modes,  higher customizability, better reliability, greater security, longer context windows, bigger context windows, and more.

 

And then in the areas of agentic behavior or autonomy. this is why I keep telling people this is the time to learn this stuff because all of these ways are evolving day after day, we are seeing small steps towards.  a greater and greater  usability, multimodality, all of these things that are making AI impossible to ignore if you’re in marketing,  let’s get into 

 

the three ways  to choose a tool. The first is to ask yourself questions like,  capability. Which AI tool is best in class for this task? The second way to choose a tool is to think about its multimodality. Sometimes you just cannot. Use a certain tool because it doesn’t have that mode.

 

It’s not able to accept that mode or create. It can’t take in or put out something that you’re using because as marketers, we’re constantly working with documents or we’re working with, you know, we’re working with video, we’re working with audio, we’re working with content, right? Like we might be dealing with something that’s already online.

 

We might be dealing with something that’s offline and needs to get digitized and put online. Like we’re, we’re constantly working with this kind of stuff as marketers, working with spreadsheets and data. We’re working inside systems. so you’ve got to ask yourself, Which one of these tools has the modality capability that I need?

 

And then the last one is, which one has the feature I need to get this task done best, right? So let’s think about this idea of AI capability. Because,  AI  is able to interact with data and content. just like we do, It’s very good at interacting with data and content, which is one of the reasons why marketing is  such a place where there’s opportunity for you to learn this AI stuff and grow as a professional, 

 

The AI works with video, it works with audio, it works with text, images, code, data, numbers, spreadsheets. So, I want you to think about AI in terms of these core capabilities, because most people think of AI and they think, Oh, it’s a creation engine, right? It creates things, it creates new ideas, it creates new content, like Blog posts LinkedIn updates, or a white paper 

 

But in reality, AI is quite  powerful in its organized function,  to structure, manage, and present information efficiently. It’s really good at researching.  gathering information, and evaluation.

 

In other words, taking in something that you give it and having an asking the AI to evaluate that grade it, improve it, compare and contrast it. That kind of stuff is super powerful in those four ways. in fact, I would say that still to this day, the create function, its ability to write like us or create like us is its weakest of the four capabilities.

 

It’s actually stronger at any of these other three, even though it’s impressive what it can do in terms of creation, we will see continued improvement in its ability to create and generate new ideas and drafts of things.  But don’t forget that it does all of these other things as well.

 

So let me demonstrate what I mean by this. I want you to use this core capabilities framework as you go through your day and you start to go into a create mode or an organized mode or a research or an evaluate mode, ask yourself  if it makes sense for you to use an AI to assist with this particular task.

 

I think we all know the create one, right? Like you probably used it to generate an article or something like that. So let’s skip along to organize. And I want to go and look at what I mean by organized. So have a look at this data set right here.

 

This is from a wine shop. So this, this company sells wine and they also sell.  They sell wine, they sell beer, they sell spirits, they sell cheese. Okay, so they sell these four things. They also have a wine club. So this would be like some CRM data here coming out of their, their point of sale system or their CRM system.

 

That’s showing that, you know, here’s customer one named Zoe, is a female at age 43 with a PhD education and single. With 139, 000 income. And I won’t get into what this is. This is from another training that I did, inside of the click pro, the recency and frequency information. Just going to hide that 

 

for this training, but here’s how much customer one Zoe spent on wine craft beer, spirits, and cheese. And then we have that information for. All of our customers in this big, long spreadsheet. So if I take that spreadsheet and I attach it to AI, let’s start a new thread here.

 

And I want to get into an organized type function. In other words, I want to take this data and transform it into something. I’m just gonna, attach this as an Excel file. So there’s my.  data set right there. And I want to use AI in this case, in an organized fashion. So I might start asking it questions, like 

 

what is the revenue breakdown  by education level?   We’re asking the AI, and again, I’m playing with no net here. let’s see, let’s make sure that it does what I ask it to do. I might have some questions and I can actually talk to my data. I can have a conversation  with my data and  you can see that, here’s my revenue breakdown over that period of time by,  education level.

 

So it went into this, Data set and it went ahead and evaluated that for me. It organized this data in a way transformed it, for me, like that, right? Like super, super fast. Now I might say, I might want to further organize this and say, can you turn this  into a bar graph?  I want to look at it in a different way.

 

we’ll see if it does that for me. Again, we’re playing with no net here, guys.  this is an example of how marketers, business people are using AI, to organize things. any time you are working with large amounts of information or data, for example, you might have a document that’s 500 pages long and you need to pull certain information out of that document.

 

you can upload that thing and start at, and you can talk to that document. You can ask it to outline that document, summarize that document. And do all those things. So this is an example of what I mean when I say, a lot of people think of an AI as a creation engine, when in fact it’s organized function. 

 

And again, that is just one little tiny use case of what I’m referring to is, think about all the different times in your day that you are having to organize information or transform it or change it in any way, and you’re And think about whether or not the AI might be able to assist in that. Now let’s move on to research.

 

And for that, I’m going to recommend another tool. And that tool is called perplexity. So you can see earlier today, I was actually talking to my son. we were talking about how many total territories does the United States own? Now, this question, I could totally ask this question.

 

to chat GPT or I could ask it to Claude, but research type questions when I get into this mode and I know I’m going to be in this mode for a little while  and I want the best engine for research right now that tool is called perplexity and it is only good at this one Use case, which is research they may try to become a creation engine or an organized engine or an evaluate engine, but it’s not best in class, in anything else, but it is the best engine right now in terms of research.

 

And so I’ll demonstrate a little bit.  We start a new thread inside of perplexity.  I’m just going to start with a very basic question. I’m going to say, I’m working with a client  in the  skin care and makeup.  e commerce  industry  and I  need to do some market research. 

 

How do you suggest we get started?  when I go into research mode and I want access to the latest information, Perplexity has a very, good user interface for research, And you can see that it’s going to say, to get started with market research for your client and skincare makeup industry, let’s start here.

 

Industry trends.  And it’s going to start laying this stuff out for me and help me do this research. some things that I love that Perplexity has is it has image search, it has video search, it, it has the ability to generate images, which is new, very cool. But the other thing that it does really well from a usability standpoint, is it helps me think what I might want to do next.

 

it’s got these related questions. And a lot of times it’s very well done and helps me move through. you just simply state to it, what you’re trying to do. What are you researching? It’s going to get you started. You can read through this and then you can, access the source material.

 

This is another thing that. Perplexity is doing really, really well is they are sourcing. Everything that they’re saying doesn’t mean that it’s not going to hallucinate, it might hallucinate. It might, you might go check this source  and find out that that’s not what they said on that page at all. but again, that’s the area of reliability in terms of its evolution. you can see that we have these,  extra questions. I might be like, okay, this is something that I want to make sure that I, that I learn about is most effective social media platforms. And so I select that and it’s just going to continue the conversation. 

 

Now, the last thing that I’ll tell you about perplexity is that perplexity is not a frontier AI model. They did not create an AI model. Instead, they are built on top of other models. So you can go in and change the models you’re using. you could say. I want to use Sonnet or GPT 4.

 

or any of these other ones. And this is, this is the perplexity pro plan. It’s 20 bucks a month. and you can check, you could change your, image generation model. Do you want to use playground, dolly, stable diffusion, or flux? I’ll get into flux here in a little bit.

 

even though we only have 12 minutes left.  Lux  is. Another image generation model that you can use So from a research standpoint, my vote is perplexity. let’s look at  what I recommend for each of the capabilities. If I currently am in the creation mode, I’m going to be inside a Claude 3.

 

5 Sonnet. I still believe that it is the best.  When I’m working, when I’m going from text to text, when I want pros or when I want something  created, right? New ideas, new content, new outlines, new drafts. I go to Cloud 3. 5 Sonnet. If I want to organize information, I can use Cloud for that. I wouldn’t say that ChatGPT4, is Is that much better at organized, but right now I just give the nose to, or, to chat GPT four here from a research standpoint, I’ve told you perplexity, it’s the best research model.

 

And a big thing is because perplexity has access to the real time internet. Now it’s up to date in terms of what’s been going on we’ve seen an evolution in chat GPT, where they are making it really easy to search the web and they are showing that they have, access to real time data.

 

So I could say, you know, what’s the latest marketing news today.  And it should be able to pull, from current headlines that it’s indexing, across  the internet, perplexity might be a little bit, in trouble with how chat GPT has evolved in this way, but you know, perplexity is, is got its own, mojo, right?

 

You can see it’s got this really slick interface and it’s a really good product. so if you do a lot of research, I do recommend that you use perplexity, the free tool for a little while, and if you like it, go ahead and buy the 20 bucks a month. It’s worth it. from an evaluate standpoint, I like GPT 4, let’s play it really quickly with an evaluate type, use case where we might want the AI to evaluate something.

 

So I might go into, GPT 4 and say, what are the best practices  for an opt in page where I want to generate leads. Let’s say I’m working with a software company.  So I’m going to ask it for the best practice. That’s the first question I’m going to ask it. And it’s going to say, okay, you want a good headline.

 

You want a good subhead. You want a strong value proposition.  So these are, this is what they’re telling me is what I want. it’s being quite thorough it’s giving me a whole bunch of things that it thinks should be The best practices of an opt in page now I’m gonna open up an opt in page that I took a screenshot.

 

It’s got a screenshot of this opt in page This is from a software company called benchmark Which is an email software. And this is their quick start to email marketing. This is a landing page. This is an opt in page, and I’m going to go back into chat GPT, and I’m going to attach this screenshot to the AI, and I’m going to say, okay, well, you just told me that these are the 12 things that should be on an opt in page.

 

I’m going to grade this opt in page  based on those criteria.  it’s going to use its AI vision,  and it’s going to look at the page, it’s going to try to understand everything that’s going on from the headline, on, and it’s going to evaluate. Claude can do this as well.

 

It can accept screenshots you can see that, it’s giving strengths and opportunities for improvement.  And then here at the end, I might use more of an organized type  prompt and say, please create a bulleted  list of suggested.

 

 You can see I don’t even care about these misspellings because the AI is really good at  understanding what I mean. So I don’t when I’m talking with it, I don’t tend to spend a lot of time  correcting my spelling. and don’t judge me on that.  so, you know, I just ask it to reorganize what it sent me here and just say, Hey, okay, well, how about just send me a bulleted list?

 

imagine when we reach that level of agentic AI, where you could just click something and say, push all of these. to dues into, a sauna or monday dot com or some to do list or project management software, that might take and assign that to different people don’t feel wigged out when I say this, but it might assign your, your, your project management software might assign.

 

Certain ones of these to other ais, They might assign that autonomy, that agentic behavior so that when you log into something like Asana or Trello or whatever, you would find that the AI’s done have to work and you just need to sign off let’s burn through  some other concepts the big one being, the idea of creating custom customizability inside of an AI.  one of the major areas of evolution to keep an eye on is customizability.

 

So  what we’ve seen with AI, and we will continue to see with AI, is the ability to bring custom knowledge into an ai. I’m gonna show you this in both Claude and inside of chat, GPT In Claude, we call this.  Projects. So Claude Projects, let’s click on the AI Marketing Hub. This is my business’s AI hub, inside of this project, think of it like a container inside of my ai  where this is the only place that I talk about my business with this ai, when I’m doing any marketing work about my business. 

 

I work inside of this project. And the reason I’m doing that is because I have customized this project to understand my business, my challenges, my offers, my customers, my editorial, guidelines, my brand voice, everything has been customized so that the AI has.  A custom set of knowledge.  we’re almost at the top of the hour, but I’m gonna end up going about 10 or 15 minutes over.

 

Feel free to jump off. I’ll email out the recording. if you have to jump off I understand that a hundred percent.  demonstrating further how customizability works in how this works, where I create this kind of container inside, which, I’m able to do all of my marketing for this business.

 

And then I create another container for this client and another container for this client. each one of these containers or projects in Claude  contains custom knowledge about that business. It’s customers, it’s offers. And so I’m training this AI.  to work with this business.  And the way we do that is we just customize it.

 

So we add lots of custom knowledge So here’s your project knowledge. if you’re using the paid version of Claude or chat GPT,  you can create. A container. I call it a marketing hub, and it’s part of a training inside of the click pro. 

 

That is extremely powerful. you can see that I upload lots of things into my mind so that it has all this context, and I’m able to have conversations with the AI, and it already knows. Things like, what am I offering? And, who are my personas?

 

Like, who am I trying to reach things like that? What’s my business, you know, when was it founded? What’s its mission statement? Who are its personas? what are its marketing objectives? All of that stuff is uploaded inside of its brain. That’s what we do. In the AI marketing hub workshop, which is part of the click pro.

 

one of the reasons why I think you need to join the click pro is that you want to learn how to customize AI  because working inside of a hub like this, like look at this offer here, look at, let’s just go to this persona. this is, a, persona that I call Reese.

 

this person is a founder of a company.  I’m going to ask  the AI marketing hub,  a question that I could never ask a generic AI that hasn’t been customized, I’m going to say.  I want to create  a new product  for Reese.  What suggestions do you have now? It’s going to reference all of my conversations that I’ve had in here.

 

You can see The strategic and tactical conversations I’ve had with this AI,  it’s going to reference all of its project knowledge to give me a answer to this question. I want to create a new product for Reese. Now imagine asking this to just regular old Claude or Chad, it doesn’t know the answer.

 

It doesn’t even know who Reese is much less have any idea what to suggest. For a new product. But I think what you’ll find is that  Claude is going to give me some pretty damn good ideas here.   this is a pretty good idea.  To create this product, which is team training and AI for founders. It’s a really good idea. Let’s see what, chat GPT comes up with and to do this inside of chat GPT, we’re going to use what we call a custom GPT.

 

Both of these are very easy to build, like extremely easy. Once you know how to do it, you could build one in an hour, like you could have a fully functioning AI marketing hub like this one. So check this out This is the clicks AI marketing hub 

 

Inside of chat GPT, if I go into edit you can see that all of the same knowledge documents, PDFs and documents I’ve created are here I can go back into this  GPT,  a custom version of chat GPT that has it.

 

other knowledge uploaded into its brain. And I’m going to ask it the same question. my guess is it’s going to give me a pretty good answer. And the reason is because it knows who Reese is. It knows everything else I offer because I’ve given it that information. I’ve customized this AI.

 

And in this case, it’s giving me six different ideas to think about and explore further, these are good ideas.  because I took the time to create an AI marketing hub,  this is an extremely powerful concept we want to create customized AI.

 

you can do this for the clients you serve. You can do this for your own business. You can do this for, if you’re an employee, you can do this for the business you work for. You can create a hub.  And get  mind blowing activity going inside as  you use your AI as a strategist.

 

As a consultant, it becomes your assistant, your  right arm where it’s like, you know, here’s something I’m thinking about. Let me bounce this off the A. I. when you do that, the more conversations you have inside your hub the more the A.

 

1. Learns about what you’re trying to do. What are your goals? really powerful. In this AI tool whip around, I am going to pop through everybody else. all the other, tools that are out there that are important, and I’ll tell you when I think you should use them if at all.

 

So everything else out there, you’ve got perplexity. you’ve got chat, GPT and Claude, I think are tied for first place right now  when chat GPT 5 comes out, which should happen sometime this winter.  I expect chat GPT 5 to absolutely blow away chat GPT 4. 0 and Claude 3. 5 We don’t know what’s, what it’s going to look like.

 

I expect that it’s going to be a massive jump forward when we see GPT 5.  Right now, perplexity is an amazing tool for research and is a Google competitor. It is a place that I go when I used to use Google to search for something and research something perplexity. is a thousand times better experience.

 

Then sifting through all the links of Google and trying to find the answer to something, you just go to the AI and you ask the question. use it when you’re doing research and need that up to date information.  Then we have OpenAI’s O1 models, right? So these are fundamentally different AI.

 

They are different from OpenAI’s GPT models. And They function differently. the difference between ChatGPT 4.  0, for example, and  the, a one model is that it is, it is.  So let, let, let me just give you, let me give you an example.  let’s give it the same query. I gave perplexity.

 

I’m doing some market research  for a beauty and skincare  company.  the difference between, one. And regular JGPT is that it has reasoning capabilities. It’s got chain of thought reasoning capabilities. It’s going to stop and think about what I asked you can open this up and see what its chain of thought was. they call this oh one preview because this is just kind of like out there, people are playing with it, but this is an important evolution towards agentic behavior.  right now it looks. very unimpressive, right?

 

But the fact that they have now figured out how to create AI systems that stop, pause and think through a problem  is a major step forward and it’s going to end up  getting us closer to agentic AI. I think in 2025 we’ll see that  O1’s, preview is what is used to create agents  inside of chat GPT.

 

So you can see that there’s all these different decisions it went through, and that’s the difference between  oh one and GPT 4. 0. And so when do you use it? Honestly, I don’t, I don’t, I don’t really use a one preview. but I also haven’t had a situation lately where I need to get in and do something that,  requires that much reasoning.

 

So you can see that, OpenAI has five stages or levels of artificial intelligence. The chatbots.  it’s AI that can have conversational language. That’s what they launched in November 2022. then competitors came out like Bard and then Gemini and now, and then Anthropic dropped out Claude and all these models came out that, that hit level one right off the bat. 

 

OpenAI’s O1 model represents the early days of moving into level two, which is the area of reasoning and human level problem solving.  Its ability to think through a problem and make decisions gets us into level two. level three is agents.

 

So this is what we will see as we move into 2025, we will see level two continue to evolve the reasoning capabilities. And then we will see agents, which we’re already seeing inside of, Claude, we’re already starting to see agents emerging. 

 

then you have Gemini. use it when you need a massive context window. Remember I said, when you ask Gemini, Hey, what’s your context window? It’s like, it’s a million It blows away anything else on the market in terms of the size of documents and stuff that you could bring inside of it.

 

If you ever have a tremendous amount of data, you need to crunch through  use Gemini because it’s got a one million context window, and the rest of the models aren’t even close to that.  it’s important to understand Gemini because it’s Google. Google is already able to, plug into things like your Google drive, Gmail,  Google docs and sheets  especially if your company uses Google workspace, for example, it’s going to have access to all that data.

 

and presumably would be able to have conversations about anything inside of your Google drive. it’s  Like taking Claude projects or custom GPTs and putting it on steroids, right? Where it’s like. you’re kind of cute that you’re able to upload some PDFs and stuff into this cloud project, and customize it.

 

I’m Google. I’m able to access the, the entire Google drive. I’m able to access all of their Gmail since the day that they started using Gmail. So  you talk about being able to plug a ton of context on to a model. That’s going to be Gemini. So don’t sleep on Gemini being. A very important player in the AI space.

 

Then you have llama, which is Facebook or Meta’s model. It’s important because it’s open source. You can build anything you want on top of it. And they have a huge distribution network for their AI through things like Instagram, WhatsApp, and Facebook. In other words, it’s free.

 

Anybody can build on top of it and their tool is extremely powerful. It doesn’t have all the bells and whistles that an open AI chat GPT has. So it doesn’t make sense most of the time for us as marketers to use it, but don’t forget that they are there and that llama is an important model because there’s a ton of other people building things on top of llamas intelligence.

 

Then you have grok. Elon Musk’s, company X AI has grok two out and it’s best known right now for its image generation capability that has no guardrails. the only time it makes sense to use grok is if you want to search through Twitter.

 

it is plugged into Twitter, and the only way to access Grok 2 is through a paid Twitter plan.  And then,  you know, the biggest use case we’re seeing people going to Grok 2 for is complete no guardrails in terms of its image generation, which is why I put this picture of Barack Obama, looking like he was in the 19th century playing a cello  So use it when you want to create limitless images, or if you want access to, Twitter. the last one I want to show you is  NotebookLM, which is a darling, in the AI space. as I heard, Paul Reitzer and, Mike Caput call it a darling, everybody’s talking about NotebookLM.

 

It’s a product from Google. it’s a good example of.  What it looks like when AI becomes more usable, This tool is not more intelligent. It’s more usable. it gives us new bells and whistles and it’s basically just using the Gemini model 

 

If you go into Gemini this way,  It can do all of these cool things because of its interface. So let’s, go into notebook LM. This is that tool here. this is a notebook and I’m going to call this notebook,  the AI powered  marketer course.  this is a course that I created and I’m going to want to do some  improvements to it.

 

I might want to get some summaries of it. I might want to have conversations with the actual course. Right. so I’m going to take all of the transcripts from this course. this is all of the transcripts.  I took the videos for this course, gave them to a transcription service, and they came back and transcribed this entire course, I’m just going to dump these inside of this notebook so I’m going to take all these text files. 

 

All, whatever this is, 10 of them,  and I’m going to dump it in there. Now they have changed how this works, but there we go.  So now it’s going to take into its brain, my course, and you can see already, it’s given me a summary  and I can have it quickly generate an FAQ out of that,  and that becomes a new note inside of this notebook,  and this is a really interesting interface for using AI.

 

1. Definitely recommend that you go and play with notebook LM because, it’s an example of that evolution in terms of usability. It’s the same level of intelligence we’ve always had. It’s just adding all of these cool, little capabilities.  one of the things that’s really sort of crappy from a usability standpoint for this tool is that you have to go down here and click here  to get back to that spot where I was.

 

I don’t know why, because it took me a little while to figure out, to click on that thing, to get this back. maybe I’m just an idiot. I don’t know, but you can have it create a study guide. You can have it create a table of contents out of all of these documents here. This is about four hours of training.

 

of a course of mine. I just dump all the transcripts in there and I say, make me a table of contents.  The other thing  that made  notebook LM go viral in the AI community was this capability here, where you’re able to create these  podcast type,  summaries of the documents you created.

 

And I’m not going to play with that today.  I’ll jump over into another notebook real quick before we jump off  to give you an idea of what it sounds like,  since I’ve done this AI tool whip around, they’ve made this even cooler so you can customize this. And you can tell the, the, the hosts, you can give them instructions because before it just said generate here and it would generate a conversation.

 

Let me give you an example of what I mean.  when I say conversation,  I would say I’m a little confused anyways, about the usability of this guy, I think  this one has already a conversation inside of it.  No, it doesn’t.  But you could generate and in a few minutes it’ll generate a very lifelike and interesting conversation between two people talking about the documents you gave it.

 

And it sounds like a podcast, right? I encourage you to go take a look at Notebook LM. I encourage you to play around with Perplexity Claude and ChatGPT based on what you learned today. I’ll stick around for about two minutes and check to see if there’s any questions let me know.

 

Did you like this training? Is this useful to see a whip around of?  These tools,  does this make sense? This concept of the eight faces of AI evolution. let me know in the  chat here before you take off.  And, and let me know whether you finding this kind of training useful.  I’m planning to do this every month, the AI tool whip around and check out what’s changed what’s new and what capabilities are out there, I’m still trying to settle on exactly how to teach this to you guys the best way and, and, and right now, this is how I’m doing it.

 

Any thoughts from you? If you want to join the click code hop in there, there’s a lot of awesome training. In addition to this one in there,  and we’ll see you on the next training. 

 

So I’ll stick around for a couple of questions and people are jumping off, but Carolyn says, what AI would you recommend for social media marketing? if you mean creating social media content,  The first thing I would do is create an AI marketing hub. I’ve got a whole training on that inside of the click pro.

 

I would encourage you to join and learn this workshop. It’s available to you right now.  once you create that hub, it’ll look like this  and create  your social content inside of this hub  for your business or whatever business you’re marketing you could also use ChatGBT, but you would want to build an AI marketing hub for the business  In that AI marketing hub workshop, we generate brand voice guidelines and all kinds of things that help the AI write better in your voice. 

 

And even with all of that work, it’s going to get you 75, 80 percent there. And then you’re going to need to, of course, bring your own human touch to anything before you publish it. 

 

Which AI platform do you use for graphics like charts, images, realistic and cartoonish.  if I need charts based on data, I’m going to use GPT four, because it, it has a capability inside of it called data.  what’s it called? 

 

Code interpreter and data analysis. I think it’s better than, Claude at creating things like charts and graphs.  I really don’t play around a lot with things like mid journey, for images or flux or anything outside of Dolly when it comes to image generation. simply because I think we’re still in the D minus. 

 

area of image generation.  it does okay. but I still find myself moving over into Canva when I need to create graphics and images. either Canva is going to become extremely good at AI image generation. Or they’re going to be in trouble because somebody’s going to create something easier to use than Canva, when, creating graphs, charts, all of that stuff, but I, I suspect canvas all over it.

 

And they’re already got a lot of things in there, with their AI magic studio, Adobe is a player as well, they are creating. tools as well have not played with Adobe’s tool set. but they are, they are very much someone to keep an eye on. 

 

Renee asks, which AIs do you recommend the paid versions of?  So I would choose either ChachiBT or Claude. And then if you do a lot of research, also invest in Perplexity.  Notebook LM, that last one I showed you is free. you don’t need Gemini right now unless you have  something you’re doing where you need a giant context window.

 

I would stay out of Gemini for now even though you might play with the free version of it and just kind of get a sense of what they’re doing over there just to stay up on it. But I would choose either Clod or GPT, Chachapiti,  and then if you do a lot of research also get Perplexity. 

 

I don’t recommend Gemini for anything at this point  except to keep an eye on it.   Google has such distribution, first of all, but they also have your data. Like they have so much access to your data, like your important data is if you’re in business, like they have your, if, if you’re using Google drive, like I use Google drive for everything.

 

I use Gmail for everything. I use Google calendar for everything. Now, imagine an AI that has access to all of that. When that becomes something that’s really very usable. And also I’m sure that I’m going to be using Gemini more and more 

 

cool.  What about go daddy arrow? I have never played with that. Victoria.  Yeah, I would drop Gemini for now.  Very cool guys. I hope you got a lot out of this one Maybe I ought to keep that in mind next time and just make it a 90 minute training so everybody’s. Planning on being here for 90 minutes and we can, as I felt a little rush there at the end, but we’ll be doing this every month 

 

If you’re going to be here Thursday, we’re going to be doing battle prompts inside the AI marketing sandbox. Hope you’ll join us for that.  talk to you then. Bye. See you guys.