JOHN CLENDENNING
1:21
Welcome back to another exciting episode of the AI Marketing Experts podcast, or we like to call ourselves Amy. Today we’re going to kind of banter through a bunch of things around how agentic systems are changing. Chris is going to do a demonstration show. We got invited to speak at an event. You’ll probably see something in the podcast world, video world, snippets of that coming out on our channel, stuff like that. But we’re speaking our basically almost had like a podcast in front of a crowd for Healthcare Group. And just the other day, and this was something, part of that presentation that we wanted to kind of, kind of show where things are going. There’s sort of a concept around level one being you’re using AI to get ideas and you’re going back and forth, you know, a little bit better than a glorified, you know, googling something or, or answer engine.
JOHN CLENDENNING
2:12
Part two is where, you know, you’re actually having the agent go do the stuff for you, the skills and stuff like that. So that’s what we’re going to be talking about today. And in that vein, everything’s changing in the world. Like I mean, we’re all Claude, Claude, Claude. So much better, so much better, so much better. ChatGPT just released a 5.5 and a Codex that’s got people going, I’m switching back to ChatGPT. So this never changes or never never stays the same. It’s always changing like in, in the minute.
JOHN CLENDENNING
2:38
So, and Jen was just telling us before we jumped on all about going into Claude Design, which is now a token hungry beta tool, but trying to take the whole design world out of app developments and other developments, taking that whole middle prototyping piece and making it code based and changeable and all kinds of cool stuff. So anyways, with that intro, who wants to have a quick little chat about what’s going on in the world of AI before we kind of let Chris Wilde to show us what the crazy stuff he can build?
BRAD KILGORE
3:11
Well, I think in the vein of latest and greatest, it’s an arms race. It really is. I mean, Google, Gemini and Notebook LM rolled out several new features this week. It’s just insane. If you’re watching these types of things on YouTube and on X and what’s coming, what’s changing the latest features, it’s pretty powerful. I mean, some of the things with Gemini this week were I’m a big fan of NotebookLM and how it creates slides and infographics and content and all these different things. And so now they’re of course baking that into some of their other tools like Google Slides. And so it’s giving you a much better ability to edit things.
BRAD KILGORE
3:57
It could create some really cool things. You’re like, well, it’s not my brand or it’s not my logo or it’s saying the wrong thing. Well, now you can fix those things pretty easily and pretty quickly. So yeah, I mean, it just keeps changing, evolving and improving. And I think that’s the cool part is as you look at these things, you know, whether you’re looking at it at a basic level or you’re looking at it at an advanced level, you’re going, okay, well you know, a few months ago was MCP and Klee or cli, and now it’s not. You just talk it. And I’m just rambling because some of that stuff’s even over my head to where, okay, what does these things do? Okay, before it was custom GPTs and then gems and now it’s what do you call markdown files and what did we do yesterday?
JOHN CLENDENNING
4:51
Skills and yeah, project.
BRAD KILGORE
4:54
I mean, it just keeps changing. And the cool thing is, I think for a lot of People, especially people that are more at beginner level, it’s going to become simpler and it’s going to become easier to use, easier to produce things. And at the end of the day, it’s just going to continue to integrate into our daily lives more and more where kind of like we don’t think much about how much the Internet or our cell phones are integrated into our lives. Like, we couldn’t have imagined that 10 or 15 years ago or 20 years ago. It’s going to be the same, I feel with this.
JOHN CLENDENNING
5:28
Yeah. And I think, Laura, you’re talking about sort of, because what happens, unfortunately, in a world where we life is normally like, hey, let me think about that, that new idea, that thing, that thing that came out. Let me put it on the back burner. Maybe, you know, if you’re old school, like, like me, you still got a printer over there. You print it off, you set it on the desk so you can remind yourself like two weeks later, a month later. Yeah, I really wanted to look at that new tool. Or maybe we’re going to switch over to Microsoft Teams from Zoom because, you know, and in today’s day and age, that’s like, well, that wasn’t a thing yesterday. It’s kind of a concept today.
JOHN CLENDENNING
6:05
And the sooner you adapt, the better because there is a everything everywhere, all at once. Brilliant movie, by the way. But it’s the way I’m kind of thinking of AI and consumer use of AI. Google’s use of AI and now on the business’s use of AI is there is becoming an everything everywhere all at once sort of. Can you tie it all together so that what you’re currently doing is creates everything else you’re doing? And that seems to be what your little concept about sort of what Google’s just added with the teams and some things like that that I thought was interesting as well.
LAURA SUTHERLY
6:39
Yes. And as we’re speaking, obviously the Google event was just wrapping up. So unfortunately there’s a lot of things that was introduced that we don’t currently see in our workspace. But Workspace Intelligence blew my mind, I have to admit. I mean, we are in that Google workspace and obviously see a lot of value, but they were showcasing AI Inbox. In other words, it will actually, once that button appears, and unfortunately it’s not there at mine at the moment, I’m waiting for it, but it’ll actually look at your inbox, go through it and prioritize stuff. So again, very, very powerful. They were showcasing Google sheets where if you are wanting to do some research, competitive research, you could actually list the companies, put your headers, highlight that area and it’ll go out and pull that information in.
LAURA SUTHERLY
7:25
So if you were saying, hey, here’s some prospects we want to, you know, reach out to just by pulling in some stuff, it will go out, research, bring in who you need to talk to, their contact information without you doing extra steps. Again, I thought that was extremely valuable. Again, that’s another thing I do not quite see in my Google sheets yet. And then I know that those, as Brad was saying about the connections and stuff, some of the workflows, some of those micro connections, it now will allow us to just connect. So we’re not needing to maybe reach out to a third party such as N8N or some of the others, that you don’t have to worry about those anymore. So all of that, I guess I would just remind our viewers and our listeners to take a peek at your. And by the way, those connections are in the studio because they were saying some of it again is, again, it’s going to be another button that appears, but it’s in the studio. But take a look, if you’re in the Google workspace, look for these new intelligent items because it could really streamline what you’re doing because it looks at the big picture.
LAURA SUTHERLY
8:38
It doesn’t. They were saying that some of the workflows, when we do those connections, it just sees exactly how we’ve connected the dots A to Z. It doesn’t know the big picture.
JOHN CLENDENNING
8:47
This whole world of, of AI intelligence doing stuff that we never have to actually work with or deal with another spreadsheet. Again, unless you’re a spreadsheet geek and love that thing. Spreadsheets are just there to get the data out. So, you know, if AI can go, hey, let me just compile it all in the background and just give you the answer that you were looking for. The pivot table, the chart or whatever, we don’t even need to worry about the D. Like the spreadsheet itself, it was a means to an end anyways, unless you really enjoy spreadsheets and numbers and formulas and all of those kinds of things. I’ve done way too much Excel and learning advanced formulas and stuff like that. Didn’t enjoy it though.
JOHN CLENDENNING
9:26
Didn’t want to do it. I just needed it to get the end result. So AI is a wonderful, you know, a wonderful change there where we’re more about getting the end work product again, what Chris is going to show, I’m going to paraphrase that or Preface that again just as a teaser but Jen, you were talking about the new Claude design and just starting to playing around in that world because it’s not what, not even a week old as of this recording. So again another moment where what used to be hard or part of other other things and Figmas and all this kind of stuff is, is now, you know, planning to become a full design studio is where they’re taking it. So what is your early impressions of, of that next advancement of, of, of AI pulling everything together.
JENNIFER CREGO
10:15
So I think that, so just looking at Claude design like my first impressions are AI certainly didn’t take it from like step one to 60, you know, in seconds. Like, you know, it took a lot of back and forth to get, to get the design I was looking for. And actually I worked on it was a brand board as far as consideration. So these were like, you know, not established brands. I mean it is an established brand but we were looking for, it’s a rebrand and so between that and then a homepage mock up that’s like based on, on the, these two brand, you know, directions looking at competitors, looking at their, their current site, like looking at all of these using design to you know, Claude design to give us those outputs. Again a lot of back and forth, not very different than what you would expect with a brand new product like this. I did end up actually pulling all of the, all of what it took and completing it with Claude code which made it a lot easier to make changes than Claude design itself. It was a little bit clunky with Claude design, you know, itself to make it exactly what I wanted it to.
JENNIFER CREGO
11:47
And it also, even though it had a lot of the kind of brand details to begin with, it kind of gave like, you know, the, the typical output to where it’s like every website that you want to create is, is definitely a separate SaaS website that needs you know, like that very typical kind of, you know, output. So, so yeah, I think that it could be amazing once it matures a bit. But yeah, it just.
JOHN CLENDENNING
12:16
What I’ve heard is this, yeah, that’s a good input because the stack is obviously the, the whole idea is to bounce between the tools. According to the way Claude released this is like if you’ve got that, if you can not have a prototype like the old days of web design, get somebody to go do it in Photoshop and then hand it to the web designer and you know, so a couple of weeks later you got a Photoshop version that you show the client and then that has to now be carved up and turned into web code. And so it goes from a picture to code which is like it’s rebuilt from the ground up. And the whole idea, you know, and Figma is an example of that, to design apps and design websites and design all that kind of stuff. And the whole plan with this is that it’s building it right in a code base so you can take it back into cloud code, you can take it back into cloud cowork and say, okay, now let’s iterate because it’s just code like let’s change this, let’s change that, let’s do this. And you’ve created a design brain that can bring it back into an engine that can play with it more. So that, that apparently is their ultimate goal as it gets dialed into being just the design side, right? So you can grab a website or an app or something you like and say make it look like that, but with this and have it spin up the actual end code that can then be edited back in one of the other tools and then deployed live from clog. So back to cloud cowork fixed up and whatever and then back to cloud code to be deployed live on a server somewhere.
JOHN CLENDENNING
13:46
And then you’ve now completed the full lifecycle. So yeah, I think, I mean that’s again, the. All these platforms are iterating like crazy. We talked a little bit the other day off, off, off this call about how anthropic Claude world is doing this where everything is about connectors and MCPs and, and, and that kind of thing. And OpenAI just went this way and they’re going, you know what, what about we have a tool that sits on top of your computer, not just in your browser, not that can see everything and interact that way. Like without any connectors. Like what if it. Like and speed of human or faster button pushing and, and finding and figuring out.
JOHN CLENDENNING
14:28
So they’re going down the road of, you know, sort of like, like an agent that sits and works on your computer in, in real visual. And CLAUDE is going down the road of agents talking to agents where no humans even need to be in the loop. Like I’m, my, my API is going to talk to your ba. My cli and I get it. And that divergence has just happened. So I think it’s really kind of cool. And I think that leads us into Chris because Chris has been mad scientist building, as we’ve been telling you for the past quite a while, inside Claude code and doing a lot of stuff. And like anything as you get used to it.
JOHN CLENDENNING
15:06
Next stuff, next stuff, next stuff becomes, well, I can do that now because it’s like anything else. Like you learn a little bit of Photoshop and next thing you know, you can do better Photoshop than somebody else or whatever it is. And Claude code is like, it’s this skill set that keeps evolving. And Chris is right on the cutting edge of that. And there was a. We were doing a presentation on a basic level about, hey, how to do level one, how to take skills and really good prompts and then really good skills and get an output. You’re the human in the loop and, and you have to be the, the agent. You are the agent.
JOHN CLENDENNING
15:40
You’re.
JOHN CLENDENNING
15:41
You’re initiating this, doing this, and then getting the AI to go do the work and come back and, and then you grab that and do something else. And, and Chris’s part of it was. Yeah, and would you like that to kind of almost pretty much all run for you and just be an agentic flow level 2? So with that, I’m handing it over to you, Chris, because you’re going to show us your level two on, on how this all fits together and where things are heading in the world of let’s do a whole bunch of things all at once by training something to do it all.
CHRIS HUNTER
16:13
Yeah. So before we jump into all of that, right, Every single bit of this starts with a process, right? You’ve got to have a process before you can just blindly give an agent or anything like that. Just carte blanche, right? So the best way to even start is with even an SOP as a, as a basic. Right. You can do a chat back and forth. I prefer typically in cloud code, like you mentioned, because it can take an idea and it can do all the chatting and all those kinds of things and take that idea and then say, well, do you want me to build that? Right, and you can do it one of two ways inside of Claude code. And originally when we, when we were talking about doing this presentation, I was going to build out some agents inside of cloud code, and I actually went ahead and did that and Claude and I went back and forth on what exactly we were going to use and what agents we were going to build. Right? So we built out together and this took maybe two hours, okay.
CHRIS HUNTER
17:30
To build the entire thing. Okay. And when I say the entire thing, I mean that we built actual agents together that would take this idea and say, okay, well, we’re gonna, we’re gonna start here and we’re gonna end over here. So what I told it essentially is that we’re going to build an. In a marketing department essentially that will take an idea, create a plan for it, and then build out a blog, 15 social posts, a press release, video scripts, if we want to do video. Right. To go along with this. And I even eventually had it create image prompts that we could then throw into Canva because like Jen mentioned, it’s not quite there for building out imagery.
CHRIS HUNTER
18:30
What I have found to work as far as our process is to build out prompts that we can then throw into something like Canva. Most of the image things nowadays have a way to agentically create images. With that being said, that’s what I threw into cloud code. It created the agents. We went back and forth with it. It created the agents, it created the skills. And that’s really where we started with was the skills, which are essentially just instructions. Okay.
CHRIS HUNTER
19:09
And then the agent. And the agent is really just another set of instructions saying to use these skills. Okay. So we did all of that. And then I decided, you know what, I kind of want a project manager to manage this whole thing so I can just give it my idea, my vision and it can go and use the different agents with the different skills and create everything. All for me. It did a great job. Loved it inside of cloud code.
CHRIS HUNTER
19:43
So I decided later I was like, well, maybe I should take this idea and check out Cowork, because I’ve heard that Cowork is kind of a dumbed down version of cloud code. I’m already in cloud code. I’ve been in cloud code for quite a while. And so I decided, you know, let’s, let’s see if I can do this and show people how to create an entire month’s worth of content in whatever amount of time. Right. That, that it takes. And so the output that it created was honestly really good. So I decided yesterday, and this is.
CHRIS HUNTER
20:20
I’m. The reason I’m sitting in my chair right now is because I’m having technology issues. And so I’m sitting out in my living room instead of my home office. Right. Anyways, so we had those tech issues yesterday and I wasn’t able to. To show what we’re going to show here today. So maybe we can send this video over to the Kansas area health care.
JENNIFER CREGO
20:44
What’s the C. Cancer Association?
CHRIS HUNTER
20:47
Association, yeah. Of healthcare communicators. Yeah. So the kahcc. So what we did, and just as a base, before I even show my screen, is that we created a fictional hospital called Acme Hospital in Salina, Kansas. Okay. And we, I Worked with Claude code. It built out 100%.
CHRIS HUNTER
21:10
This master prompt, which is just a set of instructions of this is who we are. You know, this is, these are the services, this is who we serve. This is our ideal customer profile. It built all of that out for me. Okay? So just know that that’s the basis of all of this, is that we have a fictitious hospital that we are creating content for. Okay. All right, so let me show my screen after all that, that huge intro there.
JOHN CLENDENNING
21:39
Great foundational work. So now we know where we’re at and let’s see what we built out in. And I do agree with you, Claude Cowork is again, some people say a dumbed down version, but it’s spinning up agents on its own. It’s thinking that through by itself. So instead of having to do that work to then decide it’s doing it and you’re trusting it to do well, and it does pretty well. So it’s sort of like just another moment.
CHRIS HUNTER
22:01
Right?
JOHN CLENDENNING
22:01
So.
CHRIS HUNTER
22:01
Right.
LAURA SUTHERLY
22:02
And also I’m glad that, Chris, you reminded our listeners, our viewers, that a strong start matters. The, you know, the sop, you know, you just can’t just say, do my marketing for the month. It won’t have a clue.
CHRIS HUNTER
22:16
So it all starts with a folder on your computer, okay? And so in this folder, I created Acme Hospital here, okay? And you go into that folder and this could be anywhere on your folder or on your computer, it doesn’t really matter. And it doesn’t matter if you’re Mac or Windows, I’m a Windows guy. But as you can see, we have a folder here. We have this thing called Master Prompt. And over on the left screen you can see this is all the details, right, of this fictional company. It even says in the master prompt prompted, this is a fictional 100 bed community regional hospital in Selena, Kansas, used by cloud code agents and so forth, right? So it’s got all of this, these things, you know, primary care surgery or orthopedics, all of the things that a normal hospital would have, tells us all the audiences, right. Who is our, our ideal customers. Secondary is employees and so forth and donors and all of these different things.
CHRIS HUNTER
23:15
And, and it’s, as you can see, it’s a very detailed document, right? This is the absolute foundation for getting really good outputs. Okay? And to me, if you, if you don’t have a master prompt, I highly suggest that you create this master prompt, at least a beginning of it and you can add to it over time. My master prompt for my companies are very detailed. We have our goals. We even have our rocks. And who works in the organizations. And if, if you followed us for any period of time, you understand that this is a very important document to have that you can feed into any LLM, Right? And this is something that you would keep on your computer or in Google Drive or something like that and periodically update. I update mine quarterly.
CHRIS HUNTER
24:09
Okay. All right. So that being said, what we did is, is, as you can see, I’ve got several folders here. I was playing around, quote, unquote. That’s what I do. I’m a geek, so I like playing around with this stuff. And essentially what we did, and we started here with this May, I think it’s supposed to be June. I said May because I couldn’t remember exactly which month.
CHRIS HUNTER
24:30
But we decided that May we were going to make it Alzheimer’s month. I think it truly is supposed to be June. But these are the outputs. Right. So we’ve got the blog right here, and that’s this here. Okay. Actually, let me go back a second. It wasn’t the blog that it created first.
CHRIS HUNTER
24:48
It created the editorial plan for all of these things. Okay. So this looks like gobbledygook, I’m sure, to you, because it’s in markdown language. This is the language that computers understand better than, let’s say, a Word document. But there’s all sorts of things in here. Like if you see two pound signs, that just means it’s a heading two. Okay. If you have one pound sign, that means it’s a H1 or a heading one.
CHRIS HUNTER
25:17
Okay.
JOHN CLENDENNING
25:17
And if you wanted to see it in pretty, you just open it up in Word or Google Doc and it will convert that and it looks all nice and lovely.
CHRIS HUNTER
25:24
So. Right, exactly. And right now we’re looking at it inside of my notepad, because that’s typically what I have open while I’m working with these things. I understand this stuff. It’s just like looking at HTML to me. So I understand how to read these things. Okay. But just know that this is just like a regular Word document.
CHRIS HUNTER
25:43
Right. That it created. Okay. Next it. It created a blog, and after that it created a press release and then social media and then a video script for all of this. And the image prompts, which I’ll go ahead and open here. And these are just simply the prompts that you would throw into canva. Right.
CHRIS HUNTER
26:03
Canva now has a way. And really I say now, but it’s been about a year, I think that they’ve had this ability to where you can just throw A prompt into Canva, and it’ll create the image based off of this prompt. As you can see, it’s very detailed here. It’s 1200 by 630 pixels, right. For web 16 by 9, it’s, you know, an adult daughter is very, very detailed here.
JOHN CLENDENNING
26:26
Especially if you had your. Your actual branding, like in Canva and all those tools, you can create a brand board. And now you got brand board plus prompt go, and you’re, like, dialed in.
CHRIS HUNTER
26:36
Right. And the whole idea with this entire process from the very beginning is how fast, how efficiently can we create an entire month’s worth of content? Okay. All right, so going back to this editorial plan, you’ll see actually in here, like, published dates, you know, April 28, May 1, and and so forth. This was all created inside of Cowork, all right. Which I’m about to show you here. All right. Cowork is very easy to use once you get used to it. It took a little bit for even me to get used to used to it, just because I typically work inside of CLAUDE code.
CHRIS HUNTER
27:19
And honestly, this is my normal setup right here. Okay. It’s command line interface for command. Doing it via the command prompts. Okay. But most people, when they’re talking Claude code, including my business partner, he works. I don’t know why I can’t get that to stay up. Maybe I have to click it.
CHRIS HUNTER
27:39
There you go. Is that he’s typically working in this interface right here. Okay. Most people do that. All right, so going back to this, all you have to do is create a project. Okay. You just go to projects. Create.
CHRIS HUNTER
27:55
Create new. As you can see, I’ve got one right here. That’s the one that I had open. You give it some instructions. And I gave it very basic instructions because I’m going to tell it to use all of the things that I’ve already created. Okay. Which are the. The agents and the skills.
CHRIS HUNTER
28:12
Okay. And I’ll show you what those actually look like here in a minute. But I want to get this thing going. All right, so how do we get this thing going? All right, What I want to do is let me go back up here. All right, We’ve got May, June, July. I’m missing August. Okay. So I’m just going to simply use something like Whisper Flow to talk to my computer and give it a prompt.
CHRIS HUNTER
28:35
And you just want to be as detailed as you possibly can. Hey, I think that I’d like to work on August’s content now. Let’s go ahead and work on that. Not very detailed, honestly. Okay, let’s see if it’ll do what I want it to do. Okay. Now, as it’s working here, it’s probably going to come up and ask me a few questions, which it typically does. It’s been trained in order to do that.
CHRIS HUNTER
29:00
Just like an employee of like. Well, I don’t quite understand what you want. What exactly do we want here? Now, prior to this, I did build out an entire year’s worth of what we were going to work on, on each month. So it does have that information for it to go through. Now, the other thing that I didn’t really go over. And by the way, you can click up into the main folder and you can see that this is actually working in the background. I can just let this go. Okay.
CHRIS HUNTER
29:30
So what I’ve done is that I’ve given it on the context side of things, this folder, as you can see right here, okay, it’s got access to the same exact folder that we were in right over here. It’s got all of this detailed information. Okay. The master prompt. And additionally, as I come up here, I’ll show you, they’ve got these folders here called dot claude and inside of dot Claud. This is where all of your settings for this project reside. This is where anything that you have trained it is, is going to be. So all of the agents are all right here.
CHRIS HUNTER
30:11
Okay. We’ve got a blog writer, campaign producer. This is the kind of that project manager, right? And the project itself has been instructed to use this to orchestrate pretty much everything else. Okay. Skills. These are just kind of like those sets of instructions, okay. For creating a blog or creating a design brief or planning out this entire thing. Okay.
CHRIS HUNTER
30:41
All right. So it looks like it’s done. Let me go back to that. And I didn’t show you this, but it had a little hand that was. That was up over there, right. So that’s how I knew that I needed to go back to it. By the way, Claude code inside of this interface. The Claude desktop interface does the same exact thing.
CHRIS HUNTER
31:01
Okay. Okay. So before I dispatch the campaign producer, year end plan locks as August is back to school health with back to school physics Physical Saturday. I think that’s great. Saturday, October, August 1st is perfect. Campaign spine, emphasis. Where should the content weight sit? I’m going to say balanced here. And as you can see, I’m just clicking on things.
CHRIS HUNTER
31:30
Claude code is a little bit different inside of the command line interface. It’s going to ask you some questions. You press 1 or 2 or 3 or whatever. Give it an A or a B? All right, so as you can see, it’s. It’s actually going here and it’s working away. So I’m going to pause for just a second. Are y’ all with me here or did I lose you on my Internet?
JOHN CLENDENNING
31:52
You’re good. We’re.
CHRIS HUNTER
31:55
I can’t see you. I’m just looking at my screen. Okay, perfect.
JOHN CLENDENNING
31:58
Yeah. And I think the point, so a couple of good points is like, if you think like, holy crap, this seems like a lot of work. You build things skills you build agents that can run the skills you build, you know, a director of the whole entire thing. All this guy. Yeah, once.
CHRIS HUNTER
32:11
Right.
JOHN CLENDENNING
32:11
Or you hire somebody to build it for you once. Right from that point forward. Like that initial prompt you gave was the weakest prompt I’ve ever heard you ever give in my life, Chris. And look what you got. I ready. Like, I mean, I’m just saying, like, it’s already going, oh, what about this? And like, oh, back to school and phys. Like, it is going, yeah, that’s good enough, dude. I don’t need much anymore because you built me so well in the background that we’re.
JOHN CLENDENNING
32:31
We’re ready to go. We’re cooking with gasoline, so.
CHRIS HUNTER
32:34
Absolutely right. And yes, you’re right. That was a pretty weak prompt. Hey, I think that I’d like to work on August content now. Let’s go right. Usually our prompts are very, very detailed and we tell it to ask questions, but you don’t have to tell coworkers
JOHN CLENDENNING
32:51
the more because we. We taught year or months ago. Months ago. Years ago. Year, years ago. In the AI world a few months ago about the, like the crit framework or how you prompt. We even talked about it on the, on the training. The.
JOHN CLENDENNING
33:03
The other yesterday is like you, you prompt better by giving it a role. You are going to be this. You give it context, you give it tasks and like, you know, and, and, you know, interview me and ask me questions. Well, you think that was the original way of sort of telling AI what, what it was, what it does. An agent is here’s your role. A skill is. By the way, here’s. Here’s your task and how to do it.
JOHN CLENDENNING
33:28
Here’s all your examples. Right? So you’re taking that ideal great prompt and breaking it into repeatable things forever and always. So now you can use a very simple prompt at the beginning because you’ve actually built all the rest of that framework in the actual agents and skills in the background.
CHRIS HUNTER
33:50
Now what’s interesting here and I didn’t test this before we jumped on is that it’s saying that it can’t use the campaign producer. So it built its own thing. It understood what I wanted to do here. Okay. Okay. So it has already built out the editorial plan, which is right here. That’s way better formatting than that looking at it in notepad. So as you can see here.
CHRIS HUNTER
34:18
And I’m going to collapse this down so we can actually look at this. Right? It’s got the campaign snapshot. It’s, you know, it’s got all of the detailed information for each of these. So the press release is going to go out. The blog’s going to go out October 1st, right? LinkedIn, Facebook, Instagram posts. It built all of this for me, right? Meaning it’s, it has actually said, okay, well, on October 2nd, we’re going to do this in October. If you give it more detailed information, like our event is actually going to be October, you know, fifth, whatever date, it’ll work backwards from there. And the cool part about Cowork is that if you actually connect your blog, if you actually connect your, you know, your website and your blog, if you connect to the tools like LinkedIn, like Facebook, like Instagram, it’ll schedule these things out for you.
CHRIS HUNTER
35:18
And that’s, that’s really more of the power of this entire tool. And this is a very easy to use tool. In order to do that, all you have to do is go into your customize section over here, right. And. And connect any of your things. Right? All right. So it’s taking a little while to work. You don’t have to sit here and watch it and babysit it, right.
CHRIS HUNTER
35:46
You can go. Go on with other tasks. So let’s say that I wanted to build out an entire year’s worth of prod. Worth of content right in here. I could totally do this. And it could be working with. While I work on something else. I could have this sitting, working away while I go and have lunch, and then when I come back, boom, it’s done.
CHRIS HUNTER
36:07
That’s the whole idea with agentic things, I think, is one. It’s making me more productive. I don’t have to think about things nearly as much. And it’s just running in the background, doing the things just as if I had a whole marketing department doing all of this stuff.
JOHN CLENDENNING
36:27
No, I think that’s brilliant. And I think they kind of put a pin in it for everybody because I think that, that, that demonstration is awesome. And the, the whole, the, the same, the whole Internet world of garbage in, garbage out. It’s the build at the front end. The master prompt that we talked about. It’s the, if, if it has the right context, then having it run while you’re going to have lunch and doing, doing work and then coming back and getting something else done. And getting something else done is be and being in the right folders and all of that. It all is because you are, you’ve, you’ve given it that, that knowledge base and now it’s running with that and it’s speaking in your voice and doing your thing and it just starts sounding and feeling and looking like what you want.
JOHN CLENDENNING
37:12
And that’s, that’s. So anybody who’s using any of these tools and going, well, I don’t get results like that. Go look at how your, how you started. Like, is there, is there that much context in the background?
CHRIS HUNTER
37:23
Yeah. And, and so before we wrap this all up, I’m, I’m not going to sit here while it’s, it’s doing all of this, but if, if we wanted to. Right. And I was trying to see if I could go find the blog and look at it nicely, but in a nicer version than what I was looking at it in. If you want to, you could always see what it’s actually working on right now. Right.
JOHN CLENDENNING
37:48
Do you have your main one, the one that you already ran, just as an example as we wrap up?
CHRIS HUNTER
37:52
Yeah, I was just trying to do it not in, you know, the agentic or. Sorry, in the notepad thing. Right. But this was a version right in notepad here. So this is the heading two. This is the heading one. You know, this is all done and it doesn’t sound like junk because of the master prompt, because of the editorial planning calendar that we created and because of all sorts of things that go into it. You know, if you’re still having issues with, let’s say, chat GPT creating junk.
CHRIS HUNTER
38:33
Right. For you and it doesn’t sound like you. And you’ve got to do a whole bunch of editing to make it sound like you start with this master prompt idea. Tell it the voice that you want it to write in. Tell it who you want to be writing towards. Okay. At a very minimum. And as you can tell, this is a very detailed master prompt here, but this is something that you do once.
CHRIS HUNTER
39:00
And yes, I do say that you go back to it periodically because this is where you would adjust things and, and adjust what we say. Like this right here. Do and say what we say and do say and what we don’t say. You know, sample sentences, the more that we train it on how we speak. Like, for instance, I tell my master prompt to never speak informal English because I don’t. I don’t talk that way. I talk with contractions. So I say that anything that you write for me has to have contractions in it.
CHRIS HUNTER
39:35
Okay.
JOHN CLENDENNING
39:36
And if it doesn’t have y’ all in it at least once, it’s not sounding like Chris. Just, just.
BRAD KILGORE
39:41
It’s not realistic.
CHRIS HUNTER
39:42
Big facts. Yep. And the funny thing is, is that because I’ve trained, like, even my actual agents on. On this, you know, so as you know, I’ve got two agents here, I’ve got Hope, and I’ve got Chase. Right. Chase is our sales agent, and she sounds just like me. She says, how are y’ all doing today? Right. Yeah, it’s pretty funny.
JOHN CLENDENNING
40:06
Cool. So from the panel, anybody chiming in? What. What do we think as we wrap up this episode on how we can actually use some of the, like the evolution of just from a. From a better prompt? Again, years and years ago in, in. In the AI world, like six months ago in the real world, to now building out systems that can do the work for you because you’ve stacked your way to that point. Just getting better. Yeah.
LAURA SUTHERLY
40:35
Yeah.
JENNIFER CREGO
40:35
The only thing that I would say about it is that it used to be that we were so focused on the prompt. I mean, it was like prompt engineering was like, that was a very big deal in the beginning when we kind of started really getting serious about AI and 2023, if you can believe it’s been since then. And so, yeah, it was. It was all about the prompt. And now it’s really more like you tell AI the outcome that you want the final, and it figures out the rest. I mean, like, it’s. It’s just entirely different now. And, you know, better prompt engineers were
JOHN CLENDENNING
41:12
375, $375,000 a year job three years ago. A brand new category of job, prompt engineering that doesn’t exist anymore, really. Right. Isn’t that crazy? Yes.
LAURA SUTHERLY
41:24
And I was just going to say, I know that this group, we always, you know, really say foundation. Foundation. Foundation. So if what Chris just showed was. Just blew you away, you can just again, think about that foundation. Start with some questioning of your platform of choice and see where it can take you. So just don’t think, oh, I don’t understand or I can’t quite see where Chris was going from this to that to the other. Just doing something, you’ll start feeling more comfortable with it.
LAURA SUTHERLY
41:55
And that’s what we encourage. Think of the foundation. Think of what you do every day, what task you can ask it to do. I admit something I was working on with Claude Cowork, I had an error. I ignored it. I mean, it was just, it left it. I did not get back to it. It healed itself within two weeks.
LAURA SUTHERLY
42:12
And so because I didn’t get back to it and I think I lost you again, I just encourage you to stay with it.
JOHN CLENDENNING
42:22
Yeah. Cool. Well, any last words from the panel before we wrap up another exciting episode? Cool. Well, thanks everybody for joining another episode of the AI Marketing Experts podcast. Again, like, subscribe, share us out to everybody in the show notes. You’ll you can jump into the AI Marketing Experts free community where there’s behind the scenes show notes, some more examples of stuff, all that kind of thing. So take a look at that. Join that community.
JOHN CLENDENNING
42:50
We’d love to have you over there, ask questions. Let’s go a little bit deeper on some of these topics. And yeah, because we’ve done our first panel speaking gig, if you’ve got a panel that you want us to speak at, we’re available. So reach out to us as well. So with that said, thanks everybody for joining another episode and we’ll see you next time.
