Episode Transcript
[00:00:00] Speaker A: What we have to look at with AI is it's not to entirely replace us, it's to help us. You kind of look at Tony Stark if you look at the Avengers, and isn't it cool that he's got Jarvis to tell him what he's got to do for the day, give him an answer to something and guide him? We can all look at this in the wrong way. We need to look at how we're going to work with it.
[00:00:22] Speaker B: Welcome to the Agency Hour podcast where we help web design and digital agency owners create abundance for themselves, their teams and their communities. This week I'm joined by Thomas Amos, the forward thinking founder of Design Box Studios. Thomas is a leader in the design industry, leveraging AI tools to drive innovation and efficiency in his business. And he's also one of our coaches here at Agency Mavericks. In this episode, we'll explore how AI is transforming the way agencies work. We'll dive into why AI is here to assist and not replace and how to embrace it as a creative partner. We dive into real world applications of AI, from SEO blog writing to SOP creation and digital roadmaps and tools like ChatGPT, Gemini, and Fathom that are revolutionizing processes and productivity. Thomas also shares his vision for the future of AI, including advanced video generation, voice command integrations, and even how AI could assist with everyday tasks like driving us to work. Know it's coming. If you're ready to discover how AI can save you time, streamline your workflows, and keep you ahead of the curve, this episode is for you. I'm Johnny Flash. Stay with us.
Thomas, great to see you, man. How you doing?
[00:01:37] Speaker A: Good, thanks. Thanks for, thanks for having me on.
[00:01:40] Speaker B: Yeah, man. Excited. Always excited to get to hang out with you, whether it's online or in san Diego or D.C. or whatever. Um, always a great time. So, hey, I know from just other conversations that you and I have been having lately that you have really taken hold of AI and so I would love to just dive into AI and I, I want to first just start with, you know, I think there's some concern, right, because the AI keeps getting smarter and smarter and like, am I going to even be, am I even going to be needed, Thomas, in another, you know, year or two, am I going to just be fully replaced? Is my team going to be replaced? So why don't we start there? I think that's a good, good place to start.
[00:02:20] Speaker A: Yeah, I think that's a great point. And, and I can understand that from everyone's point of view. It's quite scary. You got Elon Musk having little robots now that are going to, you know, help us do a cleaning. Cleaning around the house. And there's.
[00:02:36] Speaker B: My kids are excited about that one.
[00:02:38] Speaker A: Yeah, I mean, that one, that one I'm certainly looking forward to for sure. But, you know, we, we don't want to be replaced. You know, certain people might want to do their cleaning up at home, but. But yeah, I feel like what we have to look at with AI is it's not to entirely replace us, it's to help us. You know, you kind of look at Tony Stark if you look at the Avengers, and isn't it cool that he's got Jarvis to tell him what he's got to do for the day and give him an answer to something and guide him. So, like, I feel like we can all look at this in the wrong way. And yes, it could write a blog post for someone if that could sound kind of scary. But generally, if you're trying to get it to do a really big task, it's never a hundred percent correct. And this is where it can get a bit of a negative as to the output of AI right now, obviously in years time, this is going to be a lot better.
But it needs a person to really understand what it's trying to achieve and what it's trying to output. So we really don't need to be looking at AI replacing us.
We need to look at how we're going to work with it.
We could look at the things such as the businesses 10, 20 years ago, like Blockbuster. Used to love Blockbuster and used to go down. We used to rent a movie every Friday night.
Especially if it was a new one, you'd find that someone else had already taken it out, so you couldn't rent that movie till the next week. So they're giving it back. And then Netflix come along and we've got all of them at our doorstep. And it's. And it's the same thing we've got to look at with A.I. you know, block, fit, block. We don't want to be Blockbuster in our business. We want to be Netflix. We want to be. We want to be innovating and using the technology that's in front of us. So yes, I guess it can be scary.
You know, innovation is always scary, but it's kind of looking at the tools that you've got and looking at how that can better deliver the service that you're providing within your control.
[00:04:48] Speaker B: Love it. Such a great analogy there. Let's talk about. And I want to get into how you're using it in your agency. But the thing that I always first, and maybe this is just because we're doing a lot of SEO right now, but I always just go to like, you know, how much should agencies be using Chat GBT AI? Pick your favorite bot or whatever, you know, for SEO it, like, do we just use it for research? Is it okay to have it write an article? Like, and obviously I know that we need the, the human element there, but there's a lot of time saving, you know, and potential. Sometimes there's complex topics even where it's like to find the writer that knows this medical thing or this thing really well versus like asking Chat GBT to write about it. Like it, it knows that topic actually pretty well. So what, what have you seen there in terms of just like, specifically, I'm thinking like writing for SEO, you know?
[00:05:43] Speaker A: Yeah, yeah, so that's a really good point. So I would say certainly early days, my team tried to use chat GBT for everything and, and you know, the quality was terrible. And you knew it was Chat GPT, you know, it wasn't, you know, you knew it wasn't good. I think we have to look at things individually and break them down.
So the way that I would probably use for blog writing as just an example right now is Maybe we use ChatGPT. We brief it on the.
Let's look at the start. It's probably the best way of doing it. Let's look at the start and say, right, okay, we're looking to do a blog post. Let's do some research whether that's Chat GPT, whether that's Gemini. Let's do some research with AI as to the best topics that people are searching for. So we'd maybe use AI to assist us on some top searches, give us something to, to base that around and then ask it for inspiration. Or maybe some headlines. We're not saying do the headlines. We're saying, hey, give us 10, 20 examples of some really impactful headlines that can direct the format of the blog article.
So we've done some research for what people are asking for, what people are searching for.
We've then broken that down into a next conversation as to headlines. And then I'd probably say the next thing that I would use AI to do would be to, to put together a brief for the article, not to do the article to do a brief.
And that's now going to give you, that's going to save you so much time. It's going to set you On a path that's, you know, nicely formatted in a good way, then I feel like the human element still should do some content writing.
But now we, we can fill in the blanks. You know, we can create a couple of paragraphs on the topic, on the subject, and maybe we're stuck on a paragraph, maybe we don't want a section, or Maybe we need ChatGPT to help extend upon a section or put in there a fact that's in the blog post that would, you know, add more credibility. So you use it to kind of assist you along the journey of what the task is that you're doing. And then maybe at the end, you know, make sure that it does a little search to improve it, to make it more SEO friendly, to optimize the article, to make sure there's no plagiarism, little things like that. And you'd give it. Individual tasks where people tend to go wrong is they go, hey, create a blog post on this. And you're not giving it enough direction. You're not going through the journey. So that I feel like, look, when it comes to SEO, you kind of have to look at the individual tasks, work with it to assist you on the tasks that you've got, not do the task.
[00:08:38] Speaker B: Yeah, yeah, because we have to kind of guide it, right? I mean, it's, it's. We're sort of like, I don't know if this is a good analogy. It's just coming to me off the top. So that's always scary. But, you know, it's kind of like we're sort of like Yoda, right, Training the Jedi. And if we just like, say, hey, go out and fight the, the, you know, the Dark Force or whatever, and, and just go at it like, they're going to get crushed and it's going to be bad. Right. But if we kind of train it on how to use the lightsaber and how to, you know, all the different things, I think it is sort of like garbage into chat, gbt, garbage out. You know, if you just like, say, hey, write me this thing and write it, or if you kind of take the time and say, hey, these are the things that are important. This is the audience that they're trying to reach. This is this, you know, whatever. You can get a lot more kind of tailored content, right? Or, or parts of content or inspiration for content.
And so I think you're right with that.
[00:09:32] Speaker A: Yeah, it's. It's certainly good for writer's block, that's for sure. It's going to Give you some ideas.
Maybe you're just trying to do something in a different tone and you're not quite sure how to reword that. That's just going to assist you. But yeah, it's, it's, it's just there to assist you. And I think that's the way we need to look at it. Not to replace us. It's there to have a personal assistant and isn't that great? Like we can have an assistant and all the things that we do now rather than, rather than being replaced.
[00:10:00] Speaker B: Yeah, Love it. So what are other ways that you're using AI in your agency, like? Because I think we're all trying to find the best ways to use it and I know you've been using it in a lot of different ways. So talk a little bit about that if you would.
[00:10:12] Speaker A: Sure. So look, I could probably go down the rabbit hole of all the things that I want AI to do, but I think the easiest thing to do is to look at your day to day. So, you know what's taking you 15 minutes in a day that's really tedious that you hate doing? You know, is there something there that can help you get through that day and not have to do that tedious job you could probably look at.
So for us, as you know, Jonny, we've got this incredible SOP database that the team have created over the years and that takes so long to do. And for anyone that's looking to start that now, you know, that's a big task to cover all of the processes within your company.
So one of the first GPTs we created. And so for those of you that don't know what GPT is, if you pay premium for chat, GBT or business, you can explore something called GBTS and you can create this almost AI process driven task system in ChatGPT and it will do the same thing every single time.
So we looked at that as a great tool that we could use rather than having to create the same conversation, teach AI every single time how to do the same thing again and again and again. We would then go, right, we're going to create GPT. This is how we want it done every single time. The format we want every single time so that we don't have to do it again.
So sops, if I take you back, sops, we had multiple ways we did it. Some guys were better at writing quicker and I certainly wasn't that guy. I was, I was, I would do a quick loom video, I'll do some talking, take the transcript and Then try to write a process from the transcript.
So that was how I did that. And that's how it would take me five or 10 minutes. And then I had a video as a bonus that would add into a document that someone could follow along.
Then I had my content writers that just, you know, they wanted to do it in, in writing format. So that's how they did it. But you know, creating an sop, unless you're doing the task there and then could take, you know, half an hour, an hour and sometimes you need to edit it and there's so much backwards and forwards. So now we have a GPT. I just simply go into that and say, please create a process on how to create a Google Ads account for a new client. Give it two minutes. It's in the exact format that I need every single time. Put that into a Google Doc, put it into a database. Everyone now knows how to do it. Or put it into your project management tool.
And something that now would have took an hour is fairly accurate, 100% accurate based off the data it's got online. And we're good to go. So that I've kind of broken it down in stages. So I've looked at all the things that we work on daily. To go back to your question, we've just kind of looked at all the little things that have been niggling throughout the agency throughout the whole year, such as sending emails. So you might have email follow ups, might be like unique every single time. Doesn't matter what template you've got, Someone's asked a question in a different way and you're like, oh, this doesn't really fit my template. So we would have a GPT that would follow up with clients in the same format almost every single time. We'd paste in what they'd emailed us. It'd come back in the format that looks professional and the way that we would follow up with a client and anyone can use this, they could be dyslexic and, you know, might not be great with grammar, but it's going to come back professional every single time. So certainly for emails it's been great. But again, it's an assistant, so you want to check it, you want to not just send it blindly, you want to go back and do that.
We're now doing it with digital roadmaps. So a big shout out to Josh, who's in my squadron, he did this first and he basically taught Chat GBT to create a digital roadmap based off of the process of paid discovery to do a perfect strategy almost every single time.
So something that would take me three hours, we're talking five to 10 minutes. And we've got a real solid plan for a client to propose to them and move on to the next steps.
[00:14:54] Speaker B: Love it. That's what I've been working on. I've been working on one. I spent two hours yesterday. I got to spend the last hour today before my presentation tomorrow. So I, I need that chat gbt. I need that for that because that does take some time, even after you've, you know, transcribed the workshop and everything. And so there's a lot of time saving already. But there's still. It takes time to get it in the right format and address all those different things and stuff.
[00:15:20] Speaker A: Yeah. So this, there's certainly definitely a lot of ways that you can do it, but you don't always have to look at GPTs.
There's live chat, there's.
So we now we use go high level and they've just introduced voice AI. So that's something that's now on my shiny object list that we're looking to do. Maybe we'll get voice AI in the future to start answering the receptionist calls and save the guys answering the phones and, and, and direct people that way.
[00:15:51] Speaker B: Programming in your voice with your accent and everything and.
[00:15:56] Speaker A: Oh, I couldn't do that. I couldn't do that to AI.
[00:16:01] Speaker B: Love it.
[00:16:03] Speaker A: But Tesla do it and it's really good. It's really, really good. You, you know, you wouldn't know. And to have someone that answers a perfect question every single time, well, question comes back with a perfect answer every single time. That. That is such a cool thing to have.
[00:16:19] Speaker B: Love it. So you've got your process creator GPT, you got your digital roadmap GPT. I know you and I were talking offline about the mission statement maker GPT. What are some of the other ones just kind of as inspiration for folks that are listening?
[00:16:36] Speaker A: Yeah, sure. So let's go for the list.
So when it comes to content, we've got a GBT for creating homepage content and it always kind of does it in the same layout and framework that we'd always want. So things that we always like to include is how the headline is written and how the sub headline is written and it follows a very specific format that we like to do.
We've got one for about us content for websites, we've got one for landing pages, we've got ones for digital downloads. So like resources.
That one we've actually got in two parts We've got one that's more of a checklist system and one that's more of a lead magnet. So we would go through the process of creating a lead magnet in almost 20 minutes and we've got it all laid out. It might need some work, but it's pretty close. And the checklist one is really cool. You know, it doesn't take long at all.
We've got, we've got one for email funnels. So if we're trying to create a list of automated emails based off a set trigger. So like someone signed up for a lead magnet, how we might want to follow up with a series of emails. We've got that and that will just have that written up. We're doing quite a lot of that recently, so that's, that's definitely been something that's been of great help.
We've got, we've got one for our project managers. So when we scope out a project that let's just use a website as an example and we've kind of scoped out the hours that we're going to do, what we do is we put in, we've got one for website projects, we putting our scope of what we've proposed and then it breaks it all up into individual tasks.
Now you can have templates in things like ClickUp, but there's always additional pages. There's always something that's different in every project. So we found that we create our own GPT that we can just import every single time and all the tasks are created following how we work as a company, the testing that we do and all of the things so that once it's got the scope, we give it the scope, it will come about, it will come out and export an Excel spreadsheet for us and we just import that into our project management software. So again, it's not always perfect, but it's pretty close and saves hours of work. So that's, that's been one that's been really, really fun to work with.
[00:19:10] Speaker B: Love it. Man, that's so great. What a great list of just ideas for folks as they're thinking about. And I love you and I were also talking earlier, before this episode about just, you know, I was. I've always thought of the GPTs more in like, hey, learn this client really well. So when I ask you something, it'll know all the things about the client and you've kind of taken an approach of like, hey, this is a process or a thing that we always do. Doesn't matter if it's for this client or this client or this client. We want our homepage content or our whatever thing is to be in this format or this structure, use this approach and so then you can just kind of feed it in whichever client info that you need and then kind of get the right output, which I think is, is brilliant way as well.
Yeah, so great, great stuff there.
Talk a little bit about, you know, I know there's so many different options for AI between, you know, Chat, gbt, Gemini, co pilot, name your name, your favorite one or whatever. But kind of how are you seeing those different things develop or, or, or you know, have you used some of those and find some are better at certain things than others or just talk a little bit about that if you could.
[00:20:21] Speaker A: Sure. So obviously as you've already guessed, ChatGPT is my favorite is the one that's most used.
I've certainly used a lot, a lot that are out there. One of them that I was quite fond of was Claude AI that was. I found it a lot better for projects. So they have a format where you can save a project that you're working on with AI rather than a conversation. If you like, save it as a project and you could have multiple conversations on an individual project. So I found it was really nicely organized in the way that you would use it. It was a lot better for I would say content writing than, than other platforms that I've used.
But I, I found that, you know, there were, there were some other disadvantages where you know, it wasn't always so accurate each time that made me potentially come away and go to what I'm most used to. But Claude AI is definitely one to look at, you know, for different people and different agencies. It's definitely a great one to look at. They do have a free option and a paid option as far as I'm aware.
Then I would say Gemini Google in my opinion when it first all come out, they made a real failed attempt at AI and they had Bard and Bard was not great in my opinion. It wasn't the best. Gemini is definitely a big step up. I know they scrap barred, they've started again, they bought out Gemini and it's definitely more of a.
I would say it's good from an enterprise level.
It's going to be a lot better from what we can see because it can be integrated into Google workplace.
You can use it in docs, emails and things like that. So it's going to be really, really helpful. AI that's out there, certainly something that I'm keeping My eye on, I think the benefit that you've got with Gemini is that it's connected with Google. So you're going to be getting more current results, you're going to get more current data, better search data and facts and things like that. So I would say it was a lot more better on the analytical side and I found a mathematic mathematical side from my experience. So yeah, it's kind of better from that point, but again, maybe just got comfortable with what I was used to and come back to chat GBT there.
There's another one that's fairly good as well, which is Copilot.
Personally I haven't used this enough to have a good enough opinion, but it was fairly good on, on productivity tasks and minor tasks. I just found that it didn't always do the output that I wanted, so I got very frustrated early on and didn't give it the time of day that I should have. But yeah, there's quite a few out there. There's, there's certainly some more for sure that use different eyes and different types of AI.
But yeah, they're kind of the more conversational ones that are out there at the moment that you would use. And then there's, there's AI for creating images, there's AI for, for all these different things such as Fathom, we use Fathom AI, which if anyone isn't using Fathom that, you know, they're losing out big time. Fathom AI, you know, you get into a, you get into a meeting, it records your whole meeting, it transcribes the meeting.
At the end of the meeting you can ask it a question about what was setting the meeting and it can summarize it, it can tell. You can even ask it to create an email based off of the meeting. You know, is a really, really clever tool to use. So vevama, I definitely recommend for meetings going back to kind of the image side of things. At the moment we're only using Adobe Photoshop. I feel like it's, it's important to only use what you need it for. There's some real cool tools out there to create AI images, but I just found that as an agency, we didn't actually have a need for it, is my opinion. So because we didn't need to create these AI images, we didn't use the tools. But in Photoshop we might have a tool where we need to cut someone out.
I just mask them and just say remove the background and it does it for me. So that's where we're using it now to save time. But, you know, there will be a time that I'll tell it to maybe design a logo and it will be better than what I could do. So. Yeah. Yeah, there's some cool tools out there.
[00:25:04] Speaker B: Cool. Love it. Yeah. So many options that you can use. And man, you remember all the things we used to do in Photoshop like a decade or more ago, before even like the simple like content aware fill or scale. And now it's just like, you're like, get rid of this background, like, boom, it's gone. It's like, wow, this is amazing.
[00:25:22] Speaker A: Yeah, cool, man.
[00:25:24] Speaker B: I mean, I'm super inspired. I need to spend the next, I think, week just making all my GPTs that. That have been that you, you know, you've described that sound so great. And I think there's just so many ways that we can use AI to kind of level it up. What are. What are you, as the AI guy who's like, deep in the trenches, like, what are you looking to. For next? You know, either out of AI or to do with AI. Where. Where do you. Where do you kind of see it or hoping that it'll go?
[00:25:54] Speaker A: So there's definitely a lot of. A lot of things that's actually, you know, we're just. I always say we're heavily involved in AI, but already it's kind of. It's scary because it's like, oh, God, I don't know what's coming next. It's surprising me every day. So some of the things that are coming out. More video AI. There are some tools that do video AI right now. I wouldn't say they're great, but they're certainly getting better. They're getting better and better every time we go back to them.
Open Air is. Is bringing out like a pretty accurate AI for that. So that, that'll be cool. I think I'd be interested to see, you know, if I could create a video that AI is going to perfect the video and maybe add a subtitle and. And all of that. And a lot of these tools are already doing that. But there's, you know, could it actually add in a.
When Tom makes a mistake on a video, can it suddenly then slide to another to an example and do some of this editing for us? Could it add a funny giphy as. As. As I'm talking to, relate to the content that I'm talking about? Can it do that all automatically and move to the next level? That would be really cool.
I feel like it's already starting to move towards The Jarvis, if you like, side of things. So ChatGPT now has a voice command option. I would love to see how that now evolves, that I can now talk to AI and it's more accurate as to what I'm doing for the day, solving tasks on a daily basis. So I'll be Talking to chat, ChatGPT or Siri or whatever comes next, and it will tell me everything based off my Google Calendar and all of that. And I think, like, just really evolving into the whole minority report of side of things that, you know, everything's going to assist me into driving me to work and all of the things that Tesla are working on now. I think we just need everyone to catch up.
[00:27:59] Speaker B: Yeah.
[00:28:00] Speaker A: Yeah, pretty much there.
[00:28:01] Speaker B: Yeah.
[00:28:02] Speaker A: So, yeah.
[00:28:03] Speaker B: That's amazing. Thomas, thank you so much. This is so great. I could hang out with you all day, all week. Just always great to hang out with you. And thank you for just your generosity in our community, both in mavericks and beyond, in terms of sharing. All the things that you're doing and the knowledge that you have are just so inspiring. So thank you so much.
[00:28:22] Speaker A: I appreciate it. Appreciate it. Thanks for having me on.
[00:28:26] Speaker B: Thanks for listening to the Agency Hour podcast and a big shout out to Thomas Amos for joining us. If you'd like to learn more about leveraging AI in your agency or how you can grow your agency, check out the links in the description. Okay, folks, remember to subscribe and please share this with anyone who you think could benefit from embracing AI in their workflows. I'm Johnny Flash. Let's get to work.
