Create GPTs to Automate Go-To-Market Research

Hello everyone and welcome to today's session. My name is Rhys and I'll be your moderator and host for today's session. Uh, no Richie today, but you are in my seemingly capable hands. We're a little bit late starting today. We had a few technical issues as we were getting set up. However, we've got it sorted and we are ready to put on a very good session for you. So yeah, thank you for bearing with us. We're gonna get started very shortly. We're just waiting so everyone has a chance to join. However, um, yeah, it's going to be a good one. If you haven't seen already, there was a few setup steps if you did want to join along live today. Those setup steps are listed in the video description, and we also have a session resources, uh, link, a link to the document which is pinned in the chat. So whichever works for you, check the pinned link in the chat or check the video description. That's going to have the link to the session resources and the setup steps. Essentially what you'll need to do if you want to join in with us today is have access to ChatGPT. You're gonna need a link to the shared Google Drive folder. That's also in the video description and the resources. You're gonna need to download the workshop files. Again, those are in the Google Drive link and, uh, have ChatGPT open in a separate browser. Obviously you're gonna be watch this session, have a separate, uh, browser tab open so you can join in with us. The rest of the setup steps we're gonna cover in brief during the start of the session. So yeah, that's pretty much it. It's pretty straightforward in terms of setup for today. Essentially have access to ChatGPT, get access to the shared Google Drive folder, and download the workshop files. The rest of it is going to be incredibly straightforward. So yeah, those are your setup steps if you haven't done so already. However, you will have seen the them in the reminder emails that you got sent as well. That being said, if you have any questions throughout the session, let us know in the chat. I'll be there to answer your questions as we go through, and we'll be dedicating the final 10 minutes of the session to your questions as well. So make sure you answer— you ask those in the chat and we'll answer those for you. If you haven't done so already, please do register for this session. The link to do so is at the top of the chat. I can repost it for anyone that's missed it. And if you are just joining this stream and it's past the start time, make sure you click the, uh, live button on the stream and you'll catch up to real time as well.

Brilliant.

Uh, I'll just run through these housekeeping steps once more for anyone that's just joined. Uh, for, uh, Hannah in the chat who asked, will this video be available on YouTube after? Yes, it will be. However, make sure you register for the session and we'll send you the resources as well.

Cool.

So for anyone that's just joined, welcome to the session. We are just about to get started. Uh, please do check out the session resources. They are pinned in the chat and in the video description as well. That's gonna include the setup instructions as well as the relevant links to what would be co— what we're gonna be covering today. We have setup info, we have a link to a shared Google Drive folder that's gonna be useful for today's session, and then we've also got the slides in there as well. So everything you need is either in the video description or in the pinned link in the chat. Any other questions, let me know throughout the session. But I think we're about ready to get started. So give me 2 seconds. I'll appear on screen and we'll get started. Thanks for waiting, everyone. Okay. Hello everyone and welcome to, yeah, one of the, one of the late sessions in our AI Agents in the Enterprise. Series. This session was also meant to happen on Monday. Thank you for everyone who's turned up today. We had a few issues, we had to move the date, but we're back and we're ready to put on a fantastic session for you. So yeah, welcome to the session. Marketing teams are running more channels and more campaigns with the same headcount. CustomGPTs take repetitive research and content work off your plate without dropping the quality bar. The trick is knowing what to hand to a GPT and what to keep human. Your presenter for today is Len Ward. I'm gonna bring him on stage now. Len, welcome on stage. How you doing?

Great. Thanks for having me.

Perfect. Uh, Len is a managing partner and head of AI at Comexus, where he helps companies, uh, automate their marketing workflows. He's a serial entrepreneur and a, uh, with a previous digital marketing firm and an e-ticket event ticketing platform behind him, plus earlier stints at Credit Suisse First Boston and Morgan Stanley. Today is a prompt along, so Len's gonna show you, show you how to build custom GPTs for marketing tasks. However, this kind of applies to almost any task. Essentially, we're gonna show you how to prompt them for, for a high quality output, how to spot workflows worth automating, And yeah, that's pretty much it. So open ChatGPT and follow along. Everything you need is, is pinned in the session resources in the chat. It's also in the video description. So without further ado, Len, over to you.

Okay, great. Thanks for having me and everybody. I appreciate you jumping on this today. Just a couple like high-level thoughts about AI. I don't really know where everybody's level is with AI right now. We may have experts in here who may look at this and kind of say this is really not relevant to me. You may have people doing AI who really just kind of maybe just learning about AI, just trying to understand it. But the takeaway with AI that I want everyone to understand, whether you're an expert or not an expert, is what the media is kind of telling you what's going on and what you're hearing is not really connecting with what's happening on the front lines. So if you ever feel like, you know, AI is getting past me, I haven't done an agent or I don't know what custom GPTs are and I haven't done Claude, just know that you're in about 95% of the population. You'd be surprised how many people really don't understand AI the way you need to understand AI. And AI does go a little bit further than just opening up ChatGPT or Claude and asking it a question and trying to get an answer. So what we're gonna do today is we're gonna basically teach you how to create a custom GPT. A custom GPT is a great way to leverage data so you can make really informed decisions. We have a mantra at our company, and it's actually two mantras, and the two things we constantly tell clients and any prospective clients or anybody we're talking to about AI is that speed is your new moat and the fastest company will win. So when you are looking at data and everything every business does, the entire foundation is data. When you have data and you're looking at data, you know, typically before AI, it took a long time to process this data. Now with AI, you can actually process the data and then execute decisions based upon that data in rapid fashion. And one of the best ways to do that is to actually build a custom GPT within ChatGPT where it kind of starts remembering the conversations you had. It understands who you are, what the company is, what you're trying to accomplish, and it all lives in an ecosystem opposed to what we normally see at companies when we walk in is on the left-hand side, people just having tons and tons of chats with really no direction whatsoever. So today what we're gonna do is we are gonna go through and I'm actually gonna show you how to build it. Now I do have a marketing and sales background, so we'll kind of be leaning on the marketing sales background, but everything we're doing here can be applied for anything. It doesn't matter what you're doing in business. It doesn't matter what you're doing personally. You can build custom GPTs for anything. It doesn't have to be sales and marketing. The one caveat is it's funny that I'm actually even teaching this because 18 months ago, this was like the new hot thing. Right now, I'm beginning to believe the custom GPTs are starting to get phased out a little bit, and I'll tell you why. I actually think that the memory in Claude, the memory in ChatGPT, and the memory, you know, in Gemini is getting so good that almost every single conversation that you are having with them, it's starting to pull that data in and remember where it's really getting to know you really well, which, by the way, that is the ultimate goal of AI, to truly understand everything about your life and to help you with that, whether it's work, personal, whatever it may be. So I still think that there is value in building custom GPTs, and I absolutely think you should do it, and it will give you a leg up if you've never done it before. I will say that if they were to bring me back in 18 months, I chat, you know, custom GPTs and Gems and so forth, which is on Gemini, may be a little may be a little antiquated, but we will build this. So, you know, and forgive me on this, I'm not a real big presentation guy. It's— so this slide deck is in there. You can kind of follow along with the slide deck. I'm more of a hands-on teacher, so I'm going to jump into it. But, you know, you know, just kind of— you're kind of taking a look at this stuff. I'm really not going to touch too much on this, but it's something for you to kind of really kind of go in and take a look at. And all the data that's in the Google Drive room we gave you, you're going to be able to walk away from this and really be able to build one of these. So this was my fancy presentation before I kind of said, you know what, I'm just going to kind of jump in and show everybody what to do. So one of the most important documents that I created for you guys right here, which is in the Google Drive, which is basically— this is everything I'm going to talk about, and it's literally laid out. So you could literally save this to your documents, and anytime you want to build a custom GPT, you literally could just use this going forward. So the very first thing you have to do with building a custom GPT is you have to give it a foundation from which to pull its information from. And we like to call them golden reports. So for marketing and sales, we like to kind of go in and create these golden docs. So for marketing and sales, you know, one of the things that would take months would be to pull together the entire company overview of who you are. And then kind of go in and identify your ideal customer profile, which is ICP, then harvest all your competitor intelligence, right? So you need— this is marketing and sales teams are constantly trying to pull this stuff together. They're pulling consultants together. They're just trying to get all these reports, you know, and they're, you know, then they're coming in and they're looking at all the marketing trends and all the market conditions. And now you even have to start looking at AI trends. Before AI, this entire thing, I can tell you personally from a marketing perspective, could take a quarter, maybe a little less than a quarter. It would take that long to do something like this. Now with AI, you don't really have to do this anymore. So something away from ChatGPT that I'm going to kind of show you, I'm going to show you a great tool, which is how we build these reports, which is basically, this is Google Gemini. All right, so in Google Gemini, we are going to create those reports. I'm actually going to go back to my doc and I'm going to cut and paste. So I'm going to create a deep research report for you. Let me go back to my doc here. Okay. And so what I'm going to do is I'm going to create— so I'm not getting ahead of myself. By all means, if you have questions, stop me. I'll answer any questions you have. I'm going to create 3 documents that we're going to use as the foundation for the custom GPT. These documents is what we call, like I said, golden reports, golden docs. And it will— whenever we prompt it, it's always going to pull from this information. This reduces hallucinations. This puts guardrails up and this makes sure the GPT truly understands what you're trying to get out of it. So the very first thing when I'm gonna build a custom GPT, if I'm gonna bring a market to product or I'm thinking about strategy with sales and marketing, is I want the custom GPT to know who my business is. So I am going to create this prompt. I'm just gonna cut and paste here. This is a prompt you literally can take. You can add your company, whatever your company is. You could literally take this exact prompt. And just change to your company's name. And you're gonna go into Google Gemini. By the way, you will never do reports again once you understand this. If you've never done this before, this is the most powerful tool at our hands right now. I know there's agents and a lot of other things right now, but this is something that is not used. I'm doing what's called a deep research report. What this does is Google is out and and it crawls the entire web and every single piece of data it has to create a report that will blow you away. So I'm just creating a fictitious company, uh, right now that's basically gonna come in and be the foundation of what we're talking about. It's a cybersecurity company. So it's coming in and it's like, okay, I'm gonna create a fake company and every single thing about them. So this is the foundation. Well, let it run the report for you. Uh, and then I'm not gonna run all the other 3 reports cuz it does take time and I, I won't even let this one fully run, but I just want you to get an idea of what it's doing. It shows you the websites it's crawling. It shows you how it's harvesting the information. So anytime you need a report on anything, any type of data in any capacity, you always wanna start with Google Gemini and then click the toggle like I showed you and use deep research. It is a tremendous, tremendous AI tool. And never again are you gonna have to go out and try to create a report. You shouldn't do that. A note on these is even if you have a bunch of documents that you're trying to create a report on, you can upload all those documents and tell Deep Research to create you an excellent report. So it has a foundation. So this is running a little slow, but we'll see if it pops up in a second. If not, I already have the documents and I'll kind of show you as we upload. Um, but you create these 3 reports which we have, which I just kind of mentioned to you before. So we're going to do a company overview, an ICP/competitor, and we're also going to look at market trends and AI trends. So those are the 3 reports we're creating. Those are the 3 foundation reports we're going to do. Obviously, this is running a little slow, but you get the idea. So we will go back here. And what I'd like you to do if you're working on a computer right now is I'd like you to go to ChatGPT. If you go to ChatGPT, you have all your stuff on the left-hand side here. This is just one of our multiple chats that we run, you know, and you are going to go to GPT, you're going to click Explore GPTs. On the upper right-hand side, you're going to hit Create. OK, so if you just want to create something on the fly, right, like you just have like a quick project you're working on and you just want to kind of get some data going or whatever it may be, you could just literally talk to this and say, I want to create a custom GPT-4, and it'll help you do that. They're not as powerful and they're not as strong, but it's a good little start on a custom GPT. So what I want you to do is click configure. And the first thing you're going to do is you're going to name it. So let's go back to my document. 'Cause I wanna follow along on this document so you guys know exactly how to follow along as you're building it. Remember, those are the 3 reports we created. Okay, download the reports, which we already have. You save them to your, you know, save them to your computer. All right, so we just went through this, explore GPTs, create. Now we're gonna name it. The company we're creating is Secure North, is the company we are working with. That's the one who we're gonna call it. So we're gonna go there. All right, so you're going to give it a name. So it's going to be our consultant. By the way, when you build these, this is the greatest consultant of all time that you'll ever have. It truly is amazing. All right. And then it's going to ask for a description. You don't really have to give a description, but it's fine. It's literally just— I don't know why they even have it, but all right, well, that's not working. So little description, sales and marketing GPT, right? Okay. Now here is where it's really important. This is the instructions. For GPTs to work really well, you really have to tell them what the outcome is you're looking for, you know, what its purpose is. You're putting in guardrails. You're basically saying, I need you to do this and do that. Think of it like training a brand new employee or training a brand new teammate, you know, someone that you're bringing on. Think of like how you would train them over the course of, say, 30 or 60 days, except now you're gonna train them in 30 seconds. So if I go back to my document here, by the way, I just got notified because I do want you to see this. It's notified the report's done, but it doesn't seem to be done. We will pop back up because that is a really good tool and I'm hoping I could get back and show you that. All right. So we'll go back over here. All right. So the instructions. All right. So again, you have all this information. So we are going in and we are going to basically give you the entire instructions. And I'm trying to remember, I know they're in here somewhere. Give me a second here. Yeah, great part is my instructions are not in here. So there's your very first mistake. GPT instructions are in here. So we're going to pull this up. All right. This is a perfect document that you want to save and use going forward. Okay. So you literally, no matter what the project is, no matter what your company is, you could take this document you can upload it to ChatGPT and say, I'm creating a custom GPT for my company. Okay? My goal is, is I want, I want this to, I want the custom GPT to produce this, or I want the outcome to be this. Alright? You can actually upload this and it will give you ChatGPT the instructions that you're gonna bring back over to your custom GPT. We have found this layout typically creates the best GPTs. You know, it really kind of goes in. It's like, this is my outcome. You know, this is what I wanna use as a reference. This is who the company is. Here's our typical customers. This is, you know, what, you know, what the job is. This is what your job is not, by the way, which is really important. Again, stopping hallucinations. So the way you do this is you upload this to a custom GPT when you're not with me, and you say what your project is and what you're looking to do. It will take this information and rework it to the point of perfect instruction. So What you see here will look different than the ones that you're creating yourself. I know I'm throwing a lot at you. Stop me if you need me to. We are going to select all. Okay? And we are going to go back to our GPT and we are going to enter in our instructions. And that didn't work at all. So let's go back. Copy. Boom.

Paste.

Alright, there we go. There's everything right here. Okay. So back to the documents. We're going to see one more time that this thing ever— wow. How about that for technical difficulties? See if it pops up here. There we go. So I asked it to run a report and it came up with this entire report. I'm going to show you something really, really cool to the bottom here. It's just amazing. I mean, that report took how long to do this? 5 minutes? There's people on this call. I 100% know a lot of you have probably seen this before, but the ones that haven't, these things are so accurate, so good. If you— it's a great equalizer. And I mean, if you ever are at a point where you don't understand something, whether in your business, whether in life, whatever it may be, if you go to Deep Research and ask it to create a deep research report, you can quickly level up to become an expert in what you're asking. So this is the full report. What I really like at the bottom, What Google does is they cite every single source that they went through. They literally went through all of this and created this report for you, this fictional report. Food for thought, or like a, a, a comment. I, when I'm using AI, I have all models open. So I will use— I love ChatGPT for everyday daily use. I love the custom GPTs. I will use Google Gemini for various different things. I actually personally think Google Gemini is one of the more user-friendly. It's just really, really strong. And then I would use Claude for helping me out with coding. And if I'm going to vibe code something, I love Claude doing that. So I pretty much have all three open up because that's a lot of times that's a question, which one do you prefer? All of them. So let's go back to the GPT. Okay, so those are the reports. Remember, we created the reports. So we are going to upload our reports. These are the ones we already have. So I have company overview number one. Okay, we are gonna go second one, which is ideal customer profile, competitor intelligence. Number 3, market and AI trends. Okay, so right now, ChatGPT at this stage knows its name. Okay, it knows what its instructions are. And it now knows it has the golden reports in which it can pull information from. The beauty of this is that as you get more reports, let's just say Q2 sales results, Facebook, you know, or Meta campaign results, Google Analytics traffic, you can upload all of that information in here and now you're giving it more information to pull from. There's always the question of would you put sensitive company information on here?

I wouldn't.

But there is a way in which you can go into ChatGPT and turn off the memory and turn off where it's using your data to train. I would highly recommend doing something like that. But, you know, uploading financials or anything that may reveal customers' information, I probably wouldn't do that. What you can do is you can actually go in and say, you know, I need you to come up with a document that kind of like says, you know, maybe fictitious names and so forth. So that's a call for another day. But just to answer a question, I probably wouldn't give any sensitive data in there, but don't be afraid of giving it some data. Some people are so over the top, they don't want to give anything, and don't do that. With the world of AI, take away sensitive data, you could pretty much find out anything. So don't be afraid to put up your sales, you know, sales data in here and so forth. Some of that stuff you could put in, I think it's going to be fine. All right, so we have all that. Down below, you wanna make sure that all the capabilities are clicked off. So you actually wanna make sure that your GPT can do a full web search, that it could build an app, which I'll touch in a little bit. It can use something like Canvas if you need to design things within there. Okay? Code interpreter. You may be working on a project where you have to go through code and look at stuff. Put that in there. Okay? So you have that. You can also add a bunch of apps there. So touch on apps real quick. Apps are very interesting because we firmly believe that's the future of websites. You're starting to see a lot of different companies come out there right now, and they're starting like 1-800-Flowers, really, really large companies. And what's gonna happen is you're gonna start noticing people are interacting in ChatGPT or Gemini, and they're gonna start pulling your app up. And that app is what's gonna be used like your website. That's also gonna be the future of marketing, which means your company is going to be— have an app on there, and there's gonna be a prompt to say, would you like to, you know, interact with Secure Norris app, but you would click it and the individual will go in there and start working that app, and then you'll be charged money, much like pay-per-click. We believe that is the future, but apps are important. That's not for this call, but just pay attention to apps because that's going to grow. Uh, you building an app for your company, which in theory may replace your websites, although they'll always be there, is going to be something you're probably going to be working on by the end of the year. All right, so we have all that. Now let's do the fancy logo. We're going to update a logo. And we just created a fake logo on Google Gemini. So now we have all that. Once this is all done, verify you have your instructions, you feel good about your instructions, look at everything. Okay. You have all your documents. You're not, we're not gonna really get into the apps right now. Conversation starters are interesting. So if you want to just go in and just like, you have like maybe like there's repetitive questions you consistently ask, like, um, um, top competitors in my, in my sector. One, notice on the right-hand side here, it's popping up. Um, sales forecast, forecasting Q3. Okay, so these are things where you guys, you can go to 4 and you could click it and we'll immediately start running through data and we'll bring it down. So if you have 2 to 3, 4 questions that you're constantly asking your data, those things are excellent to have. So you have that there. Documents. Documents right now, it takes upwards of about 20 documents before you tap out and it won't take anymore. One of the things that we found that works really well is if you start stacking the documents. So you in theory could put these 3 together. I'm not saying start uploading tens of thousands of you know, pages in a document, but we have found that stacking it gives us a little bit more room. And I do know that this is something that OpenAI is working on to try to help build out.

All right.

So we feel good right now. Okay. We're like, okay, we have a name, we have our logo, we have our description. We clearly have our instructions of what we want this GPT to do. Okay. I have a couple conversation starters. I have my foundation documents. I feel pretty good. This is a critical thing to do that I can't tell you how many people forget to do. Hit create. A lot of people forget that. So who you want to create this with, it all depends. You know, our company, this is one of our company accounts, so you know, this is only shared with anyone with the Connexus link. You could do anyone with a link. If you think your GPT is really interesting and it's a really cool thing and maybe you want to build one for a customer, you can add it to the GPT store, which is kind of like an app store. Um, you know, for us right now, we're just going to say invite only. Especially if it's company, make sure you really, if it's, you know, private company information or whatever, make sure you're sharing it just with your team. We're gonna hit update and then we're gonna go to view GPT. Alright, Reece, I, I don't know if anybody's any questions or if I can just continue to go because I just threw like a lot. I just wanna make sure that, you know, everybody's still okay with it.

Yes. So we've got a few questions, but I was gonna save them for the end, but Um, there's actually a few that we can go through now. Um, we've got some time. So one of the questions that we got early on, uh, was from someone called Digital Health Off the Record, and they said, is there a reason that you're doing the research part in Gemini rather than any other LLM?

I think that's a great question. I think their Gemini, I think that's where Google truly excels. They have access to much more data than anybody else does. Um, just think about it. It's the World Wide Web, and, and Google is the, absolute de facto search engine. So I have found out of using everyone's deep research that Gemini has worked the best. Now you may use ChatGPT and think it's great, which is actually a good product. Claude does really good reports, but I like using Gemini. In addition, if you have not used it or not, it's actually called Notebook LLM. You can pair Notebook LLM, which is a Google product, with Gemini Deep Research, and it is mind-blowing what that can do. Notebook LLM is a custom GPT, like, on steroids. And I didn't— we're not doing that on this one, but I would. So that's why we use it, because when you pair those two, this is, like, really good to have. That's, like, big league type stuff when you move into there. And anybody can do it. It's not that hard. But that's why I choose Google.

Yeah, 100% makes sense. And I— yeah, I can definitely attest to Notebook LLM. and the quality that you get out of there. Being able to— the, the— it's almost like it's using like RAG systems to directly pull out stuff from your own documentation. If anyone's not, um, familiar with RAG, it's Retrieval Augmented Generation. That essentially just means it's pulling out facts rather than what it thinks is the right thing. Uh, so if you need something that needs to be factual, then anything that is RAG is usually really useful, especially in a business context. But yeah, whenever I've used NotebookLM, it is fantastic. And the extra abilities that you have in there for it to be multimodal and generate diagrams and everything else that it does in there is fantastic. So yeah, we've got a couple more questions, but I think we can save it for the end. I think it's best to carry on. But that being said, for anyone, if you do have any questions at any point, let us know in the chat. We're going to be running through them in about 20 minutes or so. So yeah, back over to you, Len.

Okay. All right. So we're just going to kind of go right in. Now, remember, this document is great because everything we're doing again, marketing, operations, sales, legal, all of this, all this is in here for you to kind of go through. So I am going to ask a couple of questions. All right. So we are going to basically say, and, you know, it's funny because you'd be surprised and I'm sure people in here have actually— maybe you're laughing a little bit. How many times have you sat in a boardroom? How many times have you sat, you know, with, with your team and people literally would say, well, who is our ideal customers? You'd be surprised how many companies 50 years in business literally don't know who their ideal customer is. Like, they still bat it around. You know, what is our biggest pain points for the chief investment officer? Like, how should we position against our competitors? These are all things that have typically taken how many hours and how many days of meetings and retreats and all that. And I'm not saying that there's not a place for that because there is a place for it. But again, speed is the new moat. The fastest company's going to win. We are moving beyond the stage of let's sit around and powwow and go through all the data. And now it's how can we interpret the data as fast as humanly possible, verify the data after we're looking at it, and then make immediate decisions on that data. That is where we are. You will clearly see the companies that are excelling at that in the next 18 months to 2 years because they're able to do exactly that. Build custom GPTs, work with a notebook LLM, string together and link all of these things together to work with each other and really get data back. Because we all know people on here, obviously you guys are all experts at data. Whoever understands the data the most and then can implement that data or, you know, you know, actions based upon that data are always the ones that are ahead. So these things right here, if you're a data junkie, this is like This is a tremendous tool for you to have. So we'll just kind of go and let's just ask questions. Who are ideal customers? Well, that's great there. I'll start interacting. I am just outside of Philadelphia, Pennsylvania. So New America. So I am going to put that in there. So Security Norris Strategy, remember, they are a cybersecurity compliance company. And if I am sitting here, if I am the CMO or if I'm the director of sales, you know, maybe I just got hired, you know, maybe I just came in and I just started. Maybe they're sitting me down at this custom GPT to kind of help me really understand what's going on with this company. I'm not going to read a bunch of past reports. I'm not going to go through all historical data. I'm going to start talking to the data, which is exactly what we're doing now. So if I'm a brand new person just starting here, I'm now learning everything right now.

Okay.

So it's kind of giving you everything. Now think about this for a minute. How long would this have taken 2 years ago to pull this data? Forever. So really kind of goes in. So here's what I'm going to do. Okay. Regional law firm. So marketing background, taking a look. What is the best way to target law firms who are our ICP, ideal customer profile member, ICP, on LinkedIn? So now I have a brand new CMO. I'm coming in, I unders— I know who the company is, I'm talking to the data. I understand the target market we're going after. I immediately know that lawyers typically do like to, like to be on LinkedIn. So I'm going in right here. Again, this is kind of, you could probably pull some of this information directly from ChatGPT. I know that, but you got to remember though, this is staying within the line of who SecureNorth is. So it's kind of making sure that it's constantly pulling from that information. It knows exactly who you are. All right. So I'm just kind of going there. Now, because it understands the competitors, the market trends, it understands our company and everything we do. What marketing taglines would we use? Sorry about that. Could we use to separate— bad spell right now. Sorry about this. Us from our competitors. I'm normally a good speller. Sorry about that. So here you go. So we're all the way down to the point where— think about this— with a custom GPT, Brand new person walking in, understands the company, understands who our ICP is, who we're exactly targeting. Now is basically saying, okay, I, I just because of my marketing background, I know LinkedIn's a good spot to go to. I have a LinkedIn strategy. Okay, now I'm going in and saying, all right, I need to go, I need a tagline almost immediately that's gonna separate us from our competitors. Here's your tagline. So it's scrolling through all the information we have and it's saying, this is what I would do. And it's giving you all this information. And then I could just basically say, create an email drip campaign using our new tagline that we can run when lawyers sign up for one of our webinars. Because I'm gonna say, I think, I think this is something that they would do. They would, they would join webinars, something we wanna do.

Here you go.

So again, the pushback on this from people on the call may be like, I do that already on ChatGPT. You may, but the problem is that you're doing ChatGPT is you're doing it with unclean data. This is clean data that I— this, this, this right here actually thinks it's working for Secure North strategy. And as you continuously interact with it, this thing will learn and it will start getting smarter and smarter and smarter. This is as dumb as it's ever going to be right here, right now. You keep working with this custom GPT, within 30 days this thing is going to surpass what you're doing. This could also be the beginning of a foundation if you're thinking about building out an agent, okay? And you're like, okay, or if you're thinking about vibe coding a product or something, let's kind of go and let's jump in there right there because I think that's a good idea. I'm also the belief that the future of AI where we are is building. You're going to be vibe coding products that's going to break bottlenecks. That's exactly what you do. So we're going to say create a vibe coding outline or product. I can, I can build, I can build on Replit, we'll say, right? You can use Replit. That helps attorneys out and will Help generate us leads, what we service. So remember, I walked in, within 10 minutes I know who the company is, I know who the identify— who the customer profile is, um, I understand the competitors, I have understood— I've already built out a LinkedIn campaign, okay? I've come up with a tagline That separates us from our competitors. And now I want to, you know, now we also, we have an email campaign. So if we do a webinar and now I want something different where after the webinar, you know, maybe the, the, the pitch is I'm gonna sell them on one of these Vibe Codem products, right? So we're gonna create a product that the attorneys are gonna use. So this is how quick you're going to market right now. All right, so here you go. So, Client Trust Risk Checker, however you want to look at it. So without getting overly in depth, you know, basically what we're doing is we're going in the Replit, and I, I'll show you something in a minute, you know. So you have all those. So basically, it basically— okay, this is, this is the product we're going to build for you. Okay. And I'm just going to finish up. Oh, here you go. Actually, it gave me the prompt. You can actually cut this prompt or copy this prompt. You can go to Replit, you can drop it in Replit, and Replit will build the actual product for you right in front of you. You may or may not have done that. So that is what a custom GPT, how fast it can move. I know you can do this most likely on Claude or whatever it may be. But like I said, the problem is, is that eventually if you stick in one chat on Claude or whatever it may be, it doesn't truly understand SecureNorth and every single thing you're trying to get out of it. You have to constantly re-prompt it and say, this is what I'm looking to do, this is what I'm not looking to do. Where this is different, where every time you jump in, it's going to remember that it's SecureNorth, it's going to remember what it has to do. So go-to-market happened that fast. I'm a brand new CMO. I just sat down. I understand the, the, the company. I understand the ideal customer profile. I understand my competitors. I am now understand the market and the AI trends. I now have said, okay, we're gonna do a LinkedIn campaign. We're now gonna have an email. The LinkedIn campaign will be for the webinar. We're now gonna have an email drip once they sign up. We have that. And then after all that's done, the final drip after the webinar is we're gonna have a product that we're gonna pitch them. Okay? And this is gonna be the product. That's how fast it just happens. And it just happened within the walls of a custom GPT. That's where we come back and we tell people again, speed is the new moat and the fastest company will win. Now, the thing I'll say in wrapping up is all of this stuff is great and it's all good, but this is where human eyes have got to be on it. So you have got to look at every single thing that's just created. You have to look at it and say, okay, I agree with that. I don't agree with that because you want to talk back to them, be like, nah, you know what, that product's not going to work. You know, literally I could say, I don't like that product. Give me another. Just want to kind of throw that at you. Okay, so, but this is where business is headed. This is what we're going to be doing. And you're going to be seeing this disposable software per se emerging from custom GPTs. You can also do what's called Gems in Google Gemini. You can use a Gem there as well. You can do various different things. But that in a nutshell is exactly why you wanna build a custom GPT and how you can build it, what you can do with a custom GPT. And it doesn't have to be sales and marketing. It could be datasets. It could be maybe you're building a new product. It could be, you know, I need a custom GPT for how do I build the right team for my, you know, my AI warehouse. You know, whatever it may be, you can build a custom GPT for any facet of your business, any angle of your business, and it becomes an absolute expert. Now, when you go to 2.0, there's ways in which you can chain this stuff. So, you know, you can actually go and you can start pulling up like different GPTs, like all here. So here's this, you know, so now this GPT can interact with this GPT and you can chain them all the way down where they're all kind of talking to each other. Eventually what's going to happen is you're going to actually be able to click this and it's going to actually be able to hook up an agent mode. And that's one final thing I'll give you. This is something a lot of people are not using yet. Okay? And it is really, really powerful. Search for my ECP I should target. Let's do easier. Um, I'm just trying to come up kind of— this is something I was doing on the fly there. So agent mode. I'll just say, uh, find me the best Firms to target in Philly. Again, this is really, really weak what I'm doing right here, but you get the idea. I want you to see Agent Mode if you have never done it before. Okay, so now me as the CMO, I can now walk away from this computer. I can go now look at my other computer and I can start doing work somewhere else as this is now working for me. So this is all within the custom GPT. So just let this show and then I'll start answering questions. So here you go. And now I'm just walking away. It's going to come back and it's going to give me the reports. Um, so if you've never done agents before, and I know everybody's out there telling you they're building agents and doing this and doing that, um, I think 5% of people are getting agents right. Maybe 10% are really, really doing a good job with it. But this is kind of where agents are right now. Uh, ChatGPT, I think, does a nice job of kind of giving you what an agent looks like, what it's going to be. But it's actually doing the work for you. So think about that. You're not scrolling and searching and looking for that. It's doing it for you and you're off to your next project. So, you know, Rhys, I'll answer any questions, you know, people may have. Hopefully you kind of get an idea of the power of these things. And unfortunately, if you were to bring me back in 12 months, I think these things are going to be antiquated. But you have that at least now.

Well, that's, that's part of the game, isn't it?

Yeah.

Constantly evolving. Cool. So we got quite a lot of questions. I'll hide the screen share for the moment just while we wait. For that to go through and then give us a little bit more room on screen. So yeah, if anyone does have any more questions, let us know in the chat, but we've got quite a few to go through now. One good one that we had from Mark that other people were interested in. So Mark says, can you update a GPT with new data such as new sales data for a new quarter? So yeah, what would, what would the case be there?

Yeah, you just simply upload that back on the file document. So if I go back in, So right here you go to edit GPT and you're right back in and then you can upload your files right there. And if you want to rework the instructions here, you can do that as well. But again, remember, make sure you hit update. So many people forget to hit update.

Yep, that does actually seem like a bit of a UI flaw there. So yeah, it is.

I agree.

Good reminder. Gina asked in the chat, can you show us again how you accessed creating the GPT webpage at the beginning? We did actually answer this one in the chat, which was essentially go to the GPTs, explore GPTs area on the sidebar, and then just click create where the update button is now.

And you mean I make it do it real quick if they want me to. So you just go here, explore GPTs, create on the right-hand side right there, and you're back, back to it again.

Easy. Before we go to another question, we did actually have a good comment from someone called Trampy Granger on YouTube. So they said, make sure you check your company's data privacy and IT policies and what can be shared with LLMs. Not everyone will have the same data access rights even within the same team.

100% right. Yeah, agree. Great comment. 100% agree.

Yeah. We had a really good one from Eduardo who said, as organizations scale these golden docs, how do you prevent institutional bias and outdated assumptions from becoming embedded into the GPT's strategic recommendations. So yeah, what's the kind of, I guess, governance process for keeping these updated and, um, yeah, making sure that you're all singing from the same hymn sheet and that it's accurate, basically?

So making sure that, that— so is the question making sure that the company is all relying on the same data, or is it making sure, making sure that you're updating the data? I'm trying to figure out what would that question be again.

So the, the core of it was how do you prevent institutional bias and outdated assumptions from becoming embedded into the GPT's recommendations?

I think that comes back to making sure you're keeping up to date on the instructions. So you should be testing it out. You should be running tests. If you don't have an AI pilot at your company, you should identify somebody as an AI pilot. And their job is twofold. The first job is to make sure that the data room is up to date and clean. So let's call her Jenny. Jenny, what she does is she monitors the entire data room where all of your, you know, golden docs are living and making sure that the documents are updated and they're fresh and that you agree, whether you got to go through legal, you know, IT, whatever it may be. Make sure that those documents are clean and that they're there and that they're good. The next thing you want to do is go in and test a series of questions. Make sure that it's answering everything properly. Make sure that it's sticking exactly to what you have in those golden docs. If you find that it's veering away from those golden docs and maybe hallucinating a little bit, that's where you need to tighten up your instructions or potentially rework your golden docs or so forth. That's actually a really good question. That's something the role of AI pilot is going to be one of the most in-demand roles, in my opinion, I think in the next 18 months.

Yep. We had another fun one from Mark. Feel free to skip over this one, but he said, how long do you think it will be before we see an AI becomes the de facto CEO of a company? When are we going to get AI CEOs?

I think it's almost there now. Yeah. I think, I think if it's, if it's, I would not be surprised if Elon Musk came out and said that, that there's been a de facto CEO running X for a while. It would not surprise me. I actually learned a great custom GPT a long time ago. Maybe what I can do is I can, in a separate email, I can send it over to Reese. He can send it out to you. There's actually a great podcast out there, a marketing podcast, and they created what's called a co-CEO. And I actually created the co-CEO custom GPT. It's tremendous. It literally acts like a CEO, and it's awesome. So to answer your question, I think one's already probably here. They're just afraid maybe to say it. It would not surprise me.

I've also heard similar things. There is, I think there's something called the Brain Rot Project, which is essentially, I think it was maybe 3 or 4 years ago, there was a group of AI researchers who got a grant paid to them to develop a project where they build a set of AI agents in order to stop people suffering from brain rot on the internet and to go out and let the agents do their own thing on what they thought was the best strategy in order to get people away from looking at the phones and doing something more productive with their time. That's tremendous for a while, and it's come back, um, and it's actually kind of— it's kind of blowing up on TikTok a little bit. So yeah, yeah, you know, that's good. More often—

it's funny you say that. The only thing I'll just— I'll veer off and say that I agree with that. I actually think that the more AI comes and the more it's here, I think more people are going to be going out and touching grass. I think more people are going to kind of less digitized because of what's being run and they're going to reengage with the actual real world. Just a little philosophy for you, but I actually think that's coming.

Yeah. We had a question from Hannah relatively early on. They, and they said, other than GPTs, what are the tools that you would spend more energy and time on for your own kind of self-development and utilization? So yes, we're learning about GPTs today. What are the other things that you've been working on recently? What have I found?

Well, like Reese, like mentioned earlier, I think Notebook LLM is one of the most underutilized, most powerful AI tools out there. It is, it again is almost just like a custom GPT where you literally can stuff all of your data in a room and then have it like go through your data and then it can actually pull different types of information. I think it's a vastly more powerful custom GPT. So I love Notebook LLM. I'll be honest, I was never a big Claude guy up until about 6 months ago. And the more I get into Claude, the more I really, really like it. So from an AI perspective, I think it's tremendous. But I think it's really good at building stuff. I do find its answers pretty good, but I still think Gemini and ChatGPT are just a little stronger when it comes to answers. But Claude for building things and solving problems, coding problems, Great. Little things in business I don't think you should really be doing anymore. Like if you're looking for a couple little small AI tools you should be using, like if you're editing videos for any capacity, you should be using something like Opus Clip, which is really cheap and really easy to do. It edits all your videos in seconds. It's really good. How do you think all the TikTokers and Instagram are getting all these videos up so fast? They're using Opus Clip. And if you ever have to do a presentation like the one I did here on a, or if my screen share, the PDF you were seeing on our presentation that was created by Beautify. Beautify is a really, really easy, very affordable product to create a presentation where you shouldn't have to really work too hard to create a presentation. You go in, you tell them what you want to do. You literally can have your custom GPT say, create a presentation for me that I can put on Beautify. It will give you the prompt and then create that presentation. Notice what I'm saying. It all comes back to speed. Within a year's time, The person you're working for, your client, uh, whoever it may be, they're going to expect that presentation to be done in 5 minutes. They're going to expect the report to be ready in 10 minutes. No longer, hey, I'm going to work the whole weekend, I'm putting a presentation together. We were at that point. So really embrace these tools. So hopefully those tools are— hopefully I'd answer your question on a couple small tools you can use.

Yeah, definitely. Um, we got one more question from the audience that I'll cover. So Um, Take My Star says, um, what is even more useful when I'm uploading new data all the time? So if essentially, just to reword slightly, when you constantly have new data coming in, uh, can you give a GPT or a gem or an agent access to a business layer, uh, say read-only sales data? Can you do that for a GPT or is a GPT more static in terms of the files and data you're able to give it?

No, you can, you can actually connect a bunch of stuff. You can actually go in and here's company data. But if I go back in, let's, here you go. Let's, let's find my GPT that we just created as Curo North. If you go in, okay, it's actually, I don't know why it's not here. But if I, let's just try a new one. I don't know why it's not connecting on ours. If you go here, company knowledge, it gets to know your company really good. You can actually start connecting with stuff. So if you go into apps and you actually go in the entire app store, you actually can start connecting to say Google Gemini Analytics. You can start connecting all of these apps where your data can start being fresh. There are ways, and if you want to build like a dashboard for your company, like go to Replit or Lovable and you can build a dashboard and then you can run API feeds directly in there. So hopefully that answers your question. But if you're looking to keep data fresh and, and you know, real time as best you can, you can do that within a custom GPT. You just have to start connecting the apps.

Yes, well, I think one thing I'd add to this is quite a lot of the, the technology that drives things like this where you want to pull information from a separate source and have it be live is MCP. So model context protocols. So this is actually something that, that we've released or with we're just about to release as part of DataCamp as well for our business users, whereby you can pull your DataCamp information to your own LLM, whether it be Claude or Gemini or ChatGPT. So you can actually, you don't have to go into DataCamp to check on your, anyone who's kind of learning on your enterprise account. You can actually get that data fed into your, fed into your AI. So yeah, there's, such a deep rabbit hole to go down in terms of the tools you're able to connect to your, uh, LLMs these days.

Yeah.

One thing I will also mention is Claude Cowork as well. Um, I've personally found that to be an incredible tool recently. Quite a lot of, um, previous kind of AI workflows and things like that, you had automation software such as Make.com and Zapier. Those tools have kind of been folded into things like Claude Cowork recently, where you can kind of go through a process once, even if it means, you know, Claude's using your browser to log into a piece of software and update a few fields. You can fold those into one process, build it once, and it will repeatedly run again. You don't need to learn about APIs if you want to set up automation workflows these days. All of these tools are becoming embedded inside your LLM. So Yeah, I would definitely explore Claude Coworker if you haven't, explore MCP if you haven't as well. There's lots of good articles on the DataCamp blog about those as well. And again, if you're doing software development or anything to do with code data, then use one of those coding agents like Claude Code or OpenAI's Codex as well.

Yeah.

Brilliant. I think that's it for questions, so I'll call it there today. If you have any other questions, leave them in the chat and we'll get back to you in the chat. But Len, thank you so much for running us through that today. Really informative. And yeah, everyone in the audience enjoyed that. If you enjoyed it in the audience, leave a nice comment for Len. We're going to be sharing the link afterwards. But yeah, thank you so much, Len. Thank you to everyone who joined us today. And yeah, we'll see you again soon for another one. Thanks, everyone.