Custom AI: How to Build a Tailored AI Marketing Consultant

 Custom AI: How to build a Tailored AI Marketing Consultant Transcript:

Date of Live Training: October 28th, 2024

It’s called Custom AI: How to build a Tailored AI Marketing Consultant. So let’s jump in. So in 2022, when AI launched in November of 2022, these are, this is my reaction. Okay. So first I was like, astonished by it. I was like, holy cow. I can’t believe. That this AI can write a poem in the voice of Jerry Seinfeld or whatever.

And I was freaked out, this is me being somewhat freaked out. I was like, is this going to kill my business? Is this going to eliminate a ton of jobs? What is going to happen with this AI thing? So I was both astonished and frightened. In 2023, I started using AI. Heavily because I was, because I was astonished and afraid.

And when I first saw it, I was like, I got to learn this. I got to learn this thing. It’s going to be extremely it’s going to have a extremely high impact on marketing, which is my thing. I’ve been doing marketing for almost 20 years. And so I was like, I’m getting into this thing and I’m going to start using it for AI.

And my reaction was I don’t know, I don’t see it. Like it’s not giving me very good output. It was actually quite frustrating. I was like, I don’t see it. Like it’s actually not that useful. And I was like both unimpressed and angry at it. Okay. Michael, I’m at smart marketer live in Denver with smart marketer team.

Molly Pittman, Ezra Firestone, those guys then in 2024, so this year. I have figured some stuff out, and not only that, but the AI models have gotten better over that time. And, at least from a marketing perspective, like we could talk about all the ethics and morals and stuff and what AI is going to do here and there and everything else, but from a marketing perspective, my reaction is this.

I love it. Like I, I am seeing how it can be used and I am using it every day with my clients and I’m teaching other people how to use it. As well. And that’s what today’s about is about what happened between 2023 and 2024 this year. What did I learn about how to use a I properly? So in 2023, like I said, I was angry.

We were promised. Superhuman capabilities, like AI was going to give us the superhuman capabilities. It was going to give us increased quality at greater speed and give us that efficiency and, make us basically like supermarketers, right? So what happened in 2023 for me that made me, Sort of rolling my eyes and be getting angry and frustrated and what we can define this as, and maybe you’re having the same reaction to AI today where you’re just like, I don’t see it.

Like it’s not super useful. It gives me really pretty terrible. useless output. And the reason is something that we call the cold start problem. So the cold start problem with a lot of what today’s about. So let me illustrate the cold start problem with something that’s near and dear to my heart, which is cooking good food.

So I love to make good food and I love to eat good food. So take a look at this. Prompt to Claude. Okay. So I use Claude a lot. I also use Chachapiti a lot. And I’ll show you examples from both today, but here’s me saying what’s a new recipe you might suggest I cook tonight? And the AI comes back and says, how about a cozy lemon garlic shrimp pasta with kale?

And my reaction to this was, How about not? Like, how about not a lemon garlic shrimp pasta with kale? Not interested in making that dish. It totally missed the mark with what I might want to cook. Okay. And, remember all I asked it was, What’s a new recipe you might suggest I cook tonight?

My reaction was, okay, not useful frustrated, wasted my time, rolling my eyes, right? Then, I created something called Chef Russ. Okay, and I’m going to show you what Chef Russ is in a second. And I asked it the same question. What’s a new recipe you would suggest at Cook Tonight? Now, this is the difference.

It says to me, let me suggest something based on reviewing the available recipes. Now, I’m going to tell you what the available recipes are in just a second. I’m going to suggest a fun twist on what I see. I noticed You have ingredients and methods for various cuisines. So I’d suggest making a Thai Mex Confusion Bowl inspired by combining elements from the Thai Red Curry and Refried Beans recipe.

Here’s my suggestion. It goes on to give me a recipe that’s wow, okay, I really want to try to make that. That sounds really delicious. So what made the difference? Difference between the first prompt and the second prompt and what it is it’s a custom knowledge base that I have added to the same tool, Claude 20 bucks a month.

And I added all of my recipes, all the recipes that I like to make. I added them to what is called a custom knowledge or project knowledge in Claude. They call it project knowledge right now. And all I did was upload a bunch of. recipes into its brain. And then I asked it the question, what’s a new recipe idea?

And it said, let me suggest something based on reviewing the recipes that you’ve already given me. I noticed you have ingredients for making Thai red curry and refried beans. So you can see my refried beans in here. And my red curry is here and there is a several more recipes in here taking up what is 5 percent of the total amount of knowledge that today I could upload into the AI’s brain.

Now this is the difference between a cold start. Where I give it no idea and avoiding the cold start. So we eliminated the cold start problem by creating a custom knowledge base for the AI. The cold start problem is it’s this set of challenges that occurs. When you’re using an AI tool cold, all right, so you don’t give it any context, no training for your particular business or use case, it creates generic output.

Of course, it’s going to create generic output. It has no context. It has no clue what you want. It’s going to create useless output. You’re not going to be able to use it. In my case, I wasn’t able to make that dish. I just didn’t, I didn’t like the idea of that recipe. It wastes your time. You would have been better off just figuring out what to make for dinner that night on your own, rather than using the AI with a cold start.

It creates frustration and, now this might be over the top, but I’m a marketer and I like to point out the pain points. It creates the slow, painful death of your soul, right? Is this cold start problem. So let’s get back to marketing and how this applies to market. So AI, we know this, right? AI has absorbed more information than any one person could in a thousand lifetimes.

It is insanely knowledgeable. It has a lot of knowledge. It has taken the internet into its. brain and it knows the breadth of all the things that we have created over the last 20 years as a group of people it has ingested that so it’s extremely knowledgeable now that’s an important word it’s knowledgeable but the problem is The AI knows nothing.

It has no clue about the business you are marketing. When you use a generic, off the shelf AI model, just right out of the box, you are going to get generic, boiled chicken, useless data. frustrating, soul crushing marketing output, and you might as well have gone and just done the work yourself completely, 100%.

Now, until after today, when you are going to learn to become an AI savvy marketer, okay, you’re going to learn to bolt tailored business data onto the AI that you are using. So what you do here is you take the broadly knowledgeable AI that’s incredibly capable And you bolt business data into its brain so that it can do its job.

Okay, so let’s get into what this means. Let’s jump back to cooking real quick. Take this Chef Russ AI. Okay, that has all of my recipes attached to it, including my famous Texas beans recipe. This is a famous recipe, at least amongst my family and friends. how to make the Texas beans is located inside of that document right there.

There’s a recipe for Texas beans. Now, it is inside of what we would call a project in Claude called Chef Russ. And if I ask Chef Russ. That AI, who has all my recipes, what are some variations I could try with my Texas beans recipe? It is, of course, going to know instantly what I’m talking about. It’s already in the AI’s brain.

I’ve attached this Texas beans recipe to this broadly knowledgeable and capable system called Claude. And you can see that it comes back and says let me analyze the Texas bean recipe and suggest some variations. And I’ll tell you what, I like these variations. Like I’m going to try some of these variations.

It’s giving me useful, relevant output because I avoided the cold start. So I bolted Russ’s recipes in here. Now I can have a much more productive and useful conversation with the AI. Is this making sense? Is this making sense? So give me a yes or no. Is it making sense? All right. So let’s look at a marketing example here.

Let’s take this company here, PetPlan, okay, so PetPlan, I picked them as an example because it’s pretty easy to understand what they do. They sell pet insurance for health insurance for your dogs, your cats and I think rabbits, too, they sell. Okay, so we are going to look at this company, and let’s say our goal, We’re setting out as a marketer to create a webinar series to sell more dog insurance.

Now, this is a pretty common thing that we might do as a marketer. We’re going to, let’s create a webinar series. Now, here’s the output from an off the shelf AI. So if I go into the AI and I say, let’s set up a webinar series to sell more dog insurance. Now this AI has a cold start. It doesn’t have any background.

At all about what I’m really trying to do. It has what I’ve given it here. I want a webinar series, sell more dog insurance. So what I’m going to get is useless, wasted time, frustrating, soul crushing output from this AI. And as you can see, of course this is a subjective thing here, but I think we can all agree that the name for this webinar is very uninspiring, right?

Why every dog owner needs pet insurance. It’s very generic. It’s very boiled chicken. It’s quite useless for us. We might as well have gone and just done the whole thing all by ourselves rather than move over to an AI because this is useless. Now, if so this is me looking at this output, right?

I’m just like, ah, like wasted time, frustration.

Now let’s take one step, even one step towards building what I call an AI marketing hub. Okay. So I’m going to introduce this concept that I’m teaching inside of theCLICK PRO called the AI marketing hub. So let’s take one step towards the hub and to define what an AI marketing hub is, It’s your central command center from which you do all your AI marketing for any one business.

Okay, so you take a business and you build a hub. If you’ve got a second business that you’re working on, you build another hub. You got a third business you work on, you build a third hub. Okay, so you build one for each client or each business that you are marketing. And what do you think you will do inside of this?

AI marketing hub. You’re going to bring the relevant documents into it and you’re going to teach it and train it. So the first step towards this is to upload into This AI customer personas offers, and what I call a business profile. So customer personas and offers. Let’s just talk about that for a second.

There is almost no time when we are doing marketing that we don’t want to be considering who are we selling to and what are we selling? That is the foundation of marketing is what are we selling? And who are we selling it to? If you take a marketer. That’s got a little bit of training and you tell them who you’re selling to and what their problems are.

And then what are you selling? They can do a reasonable job of marketing. If you don’t tell them the answer to this question, what are we selling? Who are we selling it to? You’re going to get really bad output. I don’t care how good that marketer is. They don’t have enough information. They don’t have enough context.

And then this third document. called the business profile, which is going to tell the AI a lot more about that business. When was it founded? What’s its mission? What are its marketing goals? That’s a big one inside of what we call the business profile. So let’s look at, again, inside of Claude, but you can do this inside of ChatGPT or Gemini.

as well. This is an example inside of Claude of what I call an AI marketing hub. So you’ve got pet plans, AI marketing hub, and over here you’ve got the project knowledge. Now, just like the recipes that I bolted in there, I now have context for this AI. I’ve added the business profile, which is literally just a PDF document.

This takes no coding skills, no development skills are needed. Offers. So pet plans, dog insurance, pet plans, cat insurance, and two personas that pet plan is trying to reach. The first one is Taylor, the expensive dog breed owner. So this is the person with the bougie dog, right? And then you’ve got the persona of Violet, the first time pet owner.

So these documents that I’m talking about, these documents here are no different than uploading some recipes, right? You just need to get them into. A PDF form or a spreadsheet form or however you like to create documents. The AI is getting better and better at accepting different types of documents and then bringing that into its knowledge base.

Now you can see that this is the business profile for this client pet plan. This is their cat insurance offer, and there’s a whole bunch of information about what the cat insurance offer is so that the AI is informed and understands this business. Here’s the dog insurance offer. These are separate documents, just like recipes, that we plug into the AI Spring.

Then we have the persona document containing demographic information, psychographic information, and pain points. Really important. Those pain points. What is Taylor, the expensive dog brain owners challenges? What is stopping them from getting what they want? It’s really important information to give to the AI.

And then here’s Violet, the first time pet owner. So these are just all the documents that just PDFs attach them. And you can see I’m using only 3 percent of the total knowledge that I could give this AI. And. Now, I’m going to ask it the same question, but this time I’m going to ask it inside of the hub.

I’m going to say, create a webinar series. To sell more dog insurance to Taylor. Now it knows who Taylor is because I’ve given it the Taylor customer persona. It knows very specifically what is our dog insurance offer? How much does it cost? Is there a guarantee all of the different data points that, that I recommend that you put inside of an offer that you put inside of a persona to feed to an AI, and now you can see, and again, this is subjective.

Yeah. But you tell me whether you think that this first webinar idea is better than the other one that they gave us outside the hub. So it knows that we are targeting a particular persona. This is that expensive dog breed owner persona. And the first session is the prestige of pedigree, understanding your luxury dog breed.

It’s guys, you tell me in the chat, which one’s better. The first one that they, that the AI gave us. Which was why every dog owner needs pet insurance. Or the second idea that it’s given us inside of an AI marketing hub, that’s been properly trained, the prestige of pedigree, understanding your luxury dog breed, which one do you want to use?

Which one is more useful to you? So I think it’s obvious, right? That the second one is great. Like it’s actually really good. Now I may not be able to use it exactly as they’ve suggested it, but I’m going to be able to use. It’s going to get me like halfway there. 75 percent there. This is where efficiency gets.

That promise of speed and quality and efficiency, better marketing and done faster. This is the promise of AI and this is how it gets done. So here you can see me just asking my AI marketing hub hey, What are our goals? I forgot, like I forgot what the goals are, our marketing goals. And I just wanted to illustrate that I can talk to this business profile document.

See, it’s going to say, Hey, certainly based on your business profile, here are your business goals, increased customer lifetime value, that information lives inside of here. So rather than using a generic off the shelf. cold start AI. I always move inside of my hub, whether it’s in Claude or Gemini or chat GPT.

I move inside of a hub, a container inside of the AI where I can have a informed, intelligent conversation with the AI about what I’m trying to do. It knows my business goals. It knows my offers. It knows my personas. We also set And all the AI models have this capability to set custom instructions. This is what we would call system instructions.

And you can see here that I’ve given it custom instructions. I’ve said, Hey, act as an AI powered marketing assistant, supporting the business and achieving its goals. And there is a set of custom instructions in here that basically just say, Hey, you’re an AI marketing consultant. And I want you to familiarize yourself with all the, with our business model, our business goals, our products, our services, the people we serve, everything.

And there’s some more instructions in there, but you get the point. Like we want to give it standing orders, right? Standing instructions. Here. So a technical term for what we’re talking about here is something called fine tuning. Okay, so fine tuning an AI. So in 2025, you’re going to see fine tuned AI everywhere.

Like it’s going to be everywhere. It’s going to be AI that has been customized for a certain use case, right? So you take. This generic underlying intelligence, like whatever open AI has at chat, GPT or whatever Google has through Gemini and you build on top of it, a set of custom instructions and documentation to take care of a certain use case.

Like for example, this company, these are companies that already exist. I’m going to, I’m going to show you three companies that already exist. That are just fine tuned ai, they don’t, they’re not any different from generic chat, GPT or Claw or Gemini. They are built on top of it, and in this case, REPLICA has built a bi a, an AI that is, that cares, right?

It’s got a lot of empathy. It’s very kind and it’s designed to try to be your companion and all it is customized ai. It’s fine tuned ai, it’s been trained. To be more empathetic, kind, and act like a companion. This is the same thing, except it’s a custom AI, right? So it’s built on top of. Maybe ChatGPT, or Claude, or Gemini, or even Facebook’s llama models that, that are really good.

This is called HealthAssist, and all it is it’s an AI that has been fine tuned for a particular use case. In this case, delivering support to patients, providers in the healthcare space. And then last example is a company called CoCounsel that Thomson Reuters just bought for six, it was 650 million acquisition of this custom AI tool called CoCounsel, which gives people attorneys as it creates a partner for attorneys, an AI partner for lawyers, and all it Is it’s this is literally built on top of chat GPT and it is fine tuned to be a legal assistant for lawyers and Thomson Reuters saw this tool and bought it for 650 million.

You’re going to see a lot of customized AI that has been fine tuned and tailored. For a specific use case, fine tuning the definition of this as far as I’m concerned is to take a pre trained AI model. So something like chat GPT or Claude and further train it on some specific data set to adapt it for some particular task.

So when you fine tune the AI as a market, as a marketer, remember, all I care about is marketing in this case, right? Like I’m a marketer for 20 years. So what I’m talking about as it relates to marketing is when you do this, you take off the shelf AI, like cloud or chat GPT that costs 20 bucks a month, and you transform it into an AI marketing hub.

And that transforms you into a marketer who possesses. A super intelligent assistant, right? So this is how we create assistant, that consultant that we can work with during the day. So this assistant of course, never takes the day off. It works with unfathomable speed. And most importantly, because you’ve used this AI marketing hub process, it truly understands the business that you’re helping to grow.

Fine tuning is the difference between frustration, soul crushing, useless, wasted time use of AI, and actually unlocking the capability of AI in business. Let’s talk about what capabilities are. I’m referring to here. So in the AI marketing hub workshop, which is part of theCLICK PRO, we talk about four capabilities.

And the first one is capability zero. So capability zero, I call it this capability zero because. If you don’t have this level, this basic level of fine tuning, you, you really have no chance of getting anything useful from AI. So you have to fine tune at this basic level. And it’s using something that I call the optimal primes, which are again, personas offers and a business profile.

You want to start every marketing conversation with an AI having uploaded into the system personas offers and a business profile. And you want to save that as a custom AI hub. For that company that you’re working on personas, the offers in the profile, and then you want to start your work. This is capability zero without this level of information, you’re going to have a horrible time using AI.

When you add this, it’s freaking game changer telling you. So when you add this, it’s how you go from a generic idea for a webinar name series, and a one that’s maybe. useful, right? That suggestion is actually useful. So the second capability is what I call the scribe. So a lot of us want to use AI to create ideas, outlines, and drafts of content and copy.

So one of the things that I don’t know about you, but as a marketer, I am doing a lot, is coming up with ideas, outlines, and drafts of content and copy. So copy, and I want it to align with the brand that I am working on. So when I want to do this, I want to customize my hub further By giving it writing samples that are good, right?

I can say this is the, an example of what good writing looks like. We want to upload that into its brain. We will, just like I would upload my recipes and it would go, I can see what you like to eat. Or I can, I can see your taste here. Same thing here. We want to show it examples of writing samples that are good.

We want to give it brand guidelines, like brand voice, and we want to maybe give it an editorial calendar. And I’m sure sitting out there listening to this, you can come up with ideas. For other things that you might want to bolt into an AI sprain in order to do this job, the scribe’s job, ideate, outline, and draft compelling marketing content and copy.

What else might you want to add to this AI sprain to help it do its job? For example, inside of PetPlan’s AI Hub, I have this document. This is PetPlan’s brand voice, and it talks about things like Who is the target audience? What’s what is its brand voice? Is it professional? Is it playful?

Does it use a lot of metaphors? Does it use a lot of idioms? And we build this kind of document, and we bolt it into the AI so that it understands our brand voice. And now we’ve got an AI that’s not just broadly capable and understands. All of what we’ve uploaded in the internet, but it also understands because we’ve given it capability zero, the optimal primes, it knows the personas, it knows the offers, it understands the business, and it understands now the brand voice.

Of this business. The second capability is what I call the optimizer. So the optimizer is when we want to use the AI, or we are doing Marketing activities where we want to evaluate things and improve things like landing pages, articles, ads, emails. We want them to perform better. We want them to convert better.

And for this in the AI marketing hub process that I teach in the ClickPro, we are using optimization rubrics. So what does a rubric look like? First of all, here is. the AI marketing hub for my business, the click inside of chat GPT. And you can see, I’ve just bolted a whole bunch of documents into its brain.

And this is a custom version of 20 a month chat GPT. It is called a custom GPT. And all it is I’ve uploaded a whole bunch of document into its brain the same way I might do that with a bunch of recipes. In this case, I’ve given it a whole bunch of information about my business, the offers, the personas the background of that business in my business profile, including my goals, but also some rubrics like this opt in page rubric here.

So this is what a rubric looks like. It’s a set of elements that we would want to improve. To get better output from any marketing asset in this case, an opt in page. So if we have a page where we are asking people to give us their contact information, we want that page to perform the best that it possibly can.

We want it to convert the best it possibly can. And there are elements that we as marketers have always. looked at when we’re looking at an opt in page. And those include the headline, the quality of that offer, the relevance of that offer to that persona and several other elements here. And so when I give the AI inside of the hub, so let’s take a look at this screenshot here.

This is me talking inside of my AI marketing hub. I give it a screenshot of the An opt in page and I say, hey, using the opt in page rubric, grade this opt in page and make suggestions to improve the conversion rate. Now you can see that the first thing that it starts to do when I say that to it. Is it goes and starts to search its knowledge.

It’s searching its knowledge. It’s searching inside here, inside the hub. What information have you given me that could help me do this job is what it’s saying. And so it’s going to find the rubric inside of there. This opt in page rubric, it’s going to find it. It’s going to look at my attached screenshot that I’ve given of the opt in page.

And it’s going to come back with this. Okay. To grade the opt in page based on the rubric, I’ll assess the page’s elements, headline, offer, quality, offer, relevance, offer, congruence, and copy. And you can see that it comes back and it says, okay your headline gets a two out of three. Here’s some suggestions to improve that.

Your offer quality gets a two out of three. Here’s some suggestions to improve that. So this is the optimizer. It is able to look at assets like emails, ads. anything you want to optimize and if you give it a set of criteria on which to grade that asset, it’s going to do a really good job of helping you optimize those assets, like ads and emails and everything else that we work on, articles, anything that we work on inside of Google.

A day as a marketer and the last capability that I’m very excited about that you can unlock with this A. I. Marketing hub process is what I call the strategist. So the strategist is the person you talk to when you’re talking bigger picture. You want to identify growth opportunities. You want to create longer term marketing plans.

So for this, you want to you might want to add to its brain. You might want to add an analysis of the market that you’re working in. You might want to add further information about competitors. You might put a SWOT analysis in here and bolt that onto its brain. So it understands that. But look at again, inside of chat GPT, inside of pet plans, AI marketing hub, you can see that.

I have said to it, Hey, let’s create a lead gen strategy for Violet, the first time pet owner to sell more of our dog insurance offer. Now, first of all, it’s because we are inside of this hub. That we are able to even say this in the first place, we want to do a lead gen strategy, and it knows right off the bat who Violet is, and what is our dog insurance offer.

Remember that is bolted into the AI’s brain into its custom knowledge. And so it comes right back and starts to say, okay, It knows who Violet is. It knows the pet plan, dog insurance offer. And it says let’s start to address her unique characteristics, motivations, and challenges. This is exactly what good marketers do, right?

They look at the persona. They look at the offer. They understand the goal, which is lead generation. And they start to build. a strategic plan. You’re going to get amazing results when you build a hub like this, when you’re talking strategy, and you’re going to get very generic, boiled chicken, frustrating output when you don’t use a customized AI like this.

So fine tuning is this missing link. It’s why people are frustrated. It’s what’s separating truly AI powered marketers that are going to be highly sought after already are highly sought after in the market. So it’s how we fine tune 20 a month AI. And the cool thing about this is that you can build these for your own business.

You can build these as a service. To other businesses, every company on the planet needs a hub like this. They need custom AI so you can build it for your business. You can build it for a company that you work for. You can build it for every client you serve so that you have an AI hub for client a and AI hub for client B and AI hub for client C and you talk to it.

You work on marketing. Anytime you’re leveraging AI, you work inside of client A, you work inside their hub, if you’re working on client B, you work inside of that hub. It’s this central command center from which you do all your AI marketing work. Does that make sense? Okay, so I told you I’m doing this presentation from stage tomorrow and so a couple things.

One, if that, if you understand that, the concept that I just gave you, give me a yes, give me a yes or no in here. If you have questions, you can put them in the Q& A, you can put them in the chat, if you have a comment or anything like that. I’m going to invite you right now to go join theClickPro, which is where All of the process around building one of these hubs is located inside of my training called theCLICK PRO.

You can go to join the click. ai. Join it. It’s $348 a year. So every everything that I am building in terms of ai, including this workshop, $348 a year. And if you think that this concept is smart, I really would suggest that you go take this workshop if you’re newer to ai. And that was a little bit, and that felt like a lot to you.

What I just went through. You should still join theCLICK PRO because we’ve got different levels of training. The first level of training is called the AI marketing jumpstart. It is less than an hour. And when you’re done with the AI marketing jumpstart, I know there’s a couple of people on here that are in the AI that have taken the AI marketing jumpstart.

The way to get started with this. By the end of it, you’ll feel like. Okay, I think I understand broadly how to use this and then you can start to graduate from there. You can start to take what we call the AI powered marketer course, which is, I think, around three and a half hours of content here for intermediate AI users.

And then you want to graduate up to the AI marketing hub workshop. And start attending things and watching the recordings of the AI marketing sandbox sessions, which are sessions that show you very tactically how to do certain things in marketing using AI. Questions, comments, thoughts, feelings, and emotions about the idea of building custom AI.

How do you feel about it? You can email me by scanning this if you want to, if you want to try scanning. I would love so because I just created that QR code. If you want these slides. You can scan that, or you can shoot me an email at russetmodernpublisher. com. I’ll send you the, these slides and you can join theCLICK PRO at https://jointheclick.ai

Cool. Sounds like a good approach says Dave. Very good. Okay. That’s the end of the training. I’ll stick around in case there’s any questions, but don’t want to, don’t want to hold you up if you, I’m sure you have lots of things to do today. It’s Monday. Wish me luck at the event tomorrow.

And I hope you got a lot out of this presentation and I hope you join. theCLICK PRO we would love to have you in there again. It’s 348 a year, and we’re piling lots and lots of AI training for marketers into this community.