So I think that's where the challenge with like, chat GPT comes in, is that if you are going to use it and get content that is not going to get AI detected, you have to put so many parameters in place when you're prompting and you really do need to learn how to become like an expert, prompt engineer in order to get good quality results every time. And that is really time consuming and it is a process that just takes hours and different models give you different results for things.
Mmm, that's good. To a new episode of digital coffee Marketing brew. And I'm your host, Brett Dysart. If you could please subscribe to this podcast and all your favorite podcasting apps and leave a review really does help. But this week I have Maneksha Stuart with me and she's an award winning AI and marketing specialist and the creator of marketing magic. Because we all need a little marketing magic within our marketing strategies and plans.
But it's basically an AI powered business growth tool that helps thousands of small business owners save 40 plus hours a month, which is always very helpful as well. But welcome to the show. Thank you so much for having me. I'm excited to be here. I love nothing more than a good nerdy chat about marketing and AI, so I'm excited to chat with you more. All right. And the first question is all my guest is, are you a coffee or tea drinker?
I'm british, so I am absolutely a tea drinker, but my family are turkish and so when they do have coffee, it's like really strong espresso style coffee that leaves coffee grounds in your mouth. Maybe I just need to be introduced to good quality coffee that isn't quite as strong as that. But I do love a cup of tea. Yeah, you're basically almost describing what we call in the United States cowboy coffee cowboys. Back then, they boiled their grounds within the pot and then they poured it.
And so you would get both. Yeah, it's super, super strong. And my aunties will turn your cup over and then read your coffee grounds to tell you what tragedy might beset you in the next few months. I teach this own. I love both. But do you have any specific teas that you actually drink? Yes, I really enjoy a good english breakfast tea. So just a nice english breakfast tea with a little bit of milk and it's a great way to start the day. Nice. So I gave a brief summary of your expertise.
Can you give your listeners a little bit more about what you do? Yeah. So I started SEO in 2005, working in for $3 an article to rewrite medical articles with keywords attached to them. I have spent many years working in SEO. Subsequently to that, and then in 2017, I moved more towards being a freelance head of marketing for thought leaders.
So worked with quite a few people who have published books and have got podcasts and have got large online audiences and a body of content, and was doing a lot more around funnel building and the sort of digital sales side of things for a variety of business models, but ultimately a lot of knowledge, commerce, and from there just really have spent a lot more time on the sort of strategic elements of marketing and kind of enjoyed that a lot more. SEO is great. It's very repetitive.
And then with how technology has been changing since 2017, when Google really brought its first big machine learning update to its algorithm, how keyword intent has changed things just in terms of genies have exact match keywords anymore? No, you don't, not really. So what does that mean when you're doing keyword research?
Actually, you're more looking at customer journey and what the customer experience is and what needs and problems your customers have at each different stage of their online journey, and then how you can make sure that your content or messaging is meeting them at the point of their express need. So the interest in AI had continued to build and grow from that. And obviously then OpenAI's API became available in early 2021, I think it was.
And just being able to look at generative AI and generative AI tools a little bit more, and then just thinking through the overlap between how do we make sure that the AI content that we're creating retains the value and the context that our customers need, so that we're not contributing to noise on the Internet. We are instead being really intentional about the quality of marketing that we're putting out there and how we are showing up and serving our audience.
So that's a little bit of a sort of summary of some of the key areas of interest, really, that have shaped my career on the Internet. So SEO and AI, it seems like it's a match made in heaven in a way, because like you said, SEO is very repetitive and AI can help really well with a repetitive task. Creativity needs a little work still, but. So how can PR people and marketers merge those twos and use them effectively?
Yeah, basically what we see with, so the first thing is that generative AI in particular is not able to replace place search engine functionality because of limitations around how large language models are built so it can connect in. So we see that with chat, GPT's like recent update, where you now can get bing results pulled into the chat interface. But ultimately it needs to be able to refer to the search algorithm because of things like recency and also for factual information.
Large language models hallucinate a lot and have a tendency to make things up. But what we are seeing with, like the AI technology is a really, it's a natural lead on from the search engine technology and the algorithms that we've been seeing in terms of updates from Google. The ability of a search engine to understand and determine context from the searches that we give to it has been a part of that machine learning.
And it's this idea that a search engine is trying to provide somebody with the best quality answer to the express needs that they've given at that time. What we have seen with the Internet is that because of how quickly the volume of content has grown, people are actually getting incredibly overwhelmed by the amount of information that there is when they're searching for things.
And so we're seeing like 16 million results for, if you Google SEO checklist, you're getting 16 million results to filter through, and you're having to filter your own context into each one of those results to determine, oh, is this actually what I need? Is this what I need? Is this what I need? And so people's overwhelm levels, I think, are the reason that chat GPT took off so quickly, because people want to be able to search for something and actually have a contextual answer given to them.
And search engine is, search engine technology is going to change because it has to, but also because users are just getting increasingly more aware of how much energy they spend being their own filter on the Internet for what they're looking for.
So when we look at using AI as marketers, as pr professionals, as people in the business space, we're really looking at like, how do we use this AI tool to help us create contextual experiences for our customers that mean that they have what they need as quickly as possible. They have the answers, they understand the impact as quickly as possible without having to search for a needle in a haystack.
And so I think, like, one of the trends that we are going to see in marketing is people just being much more comfortable with the idea of niche and exactly who it is that they serve, and understanding that, like, their customers are going to need very specific things from them. And the people who can best anticipate the needs of their customers and respond to them are the ones who are going to have businesses that are sustainable in the next era of the Internet.
But as far as using it within our from an SEO perspective, one of the results of the generative AI trend has been Google's latest big algorithm update, which happened in March, which really targets very obviously spam content, like spam content. So spam brain is the algorithm and it's picking out anything that looks like it's been mass created or created on content farm and it is removing it from getting search visibility.
So one of the things that as we are moving forward is, yes, we're using AI as a tool, but if we are not using AI as a tool and providing a contextual framework to guide the AI, our results are going to struggle to get any traction or visibility organically because it is so easy to spot when AI content has been put into a blog post or published on a page without any sort of human interaction.
So Google's already starting to penalize some of that activity, and it gives us an opportunity in the marketing space to really be able to pull together our messaging, our positioning, our filters when it comes to what information is necessary, what information is needed, and like, how do we create really good quality experiences for humans? And I think the way that I describe it is AI helps with the functional so you can get back to the fun.
So there's always a prescriptive format that you might need to have for things like press releases, for blog posts, for emails, for launch assets where certain pieces of information need to be described in a way that is clear and easy to understand. And that's where there's a little bit more prescription in terms of format. But then the creativity is like, how do we do advertising? How do we format our content?
We're talking in a podcast interview now, but if we were using, let's say we've managed to streamline a whole lot of stuff in our business that allows us to have a little bit more space or a little bit more profitability. What would it look like to have an illustrator animate sections of the podcast video to promote on social media or to send via ads?
So what we are able to see is that actually by automating some functional elements of our marketing and our sales in the online space, we can open up space for creative ideas and for reformatting things to make sure that our message lands with our ideal customers in a much more impactful way. And using AI as a tool as part of the ideation process is great, but we are ultimately still looking for value for our ideal customers as well. That is, that goes beyond just getting a checklist from chat.
GPT. Gotcha generative AI is also infecting, not infecting, but merging in with the creative side. So you have Adobe putting it into premiere pro, you have Davinci doing some of it, but not as much. Adobe is all for it. Adobe's like full bore. We're gonna, we're just, all our new features is AI, which teaches our own. It could be good, it could be bad. So like you said, Google is really targeting the farming of content from it. So how do you do it effectively?
Because I'm a one man show and sometimes I get a little lazy with it and I'm like, well, it's good enough. Like, I don't have time to like really fine tune and edit all this stuff. So how do you like, how do you make that balancing act? Because it's a balance between using it and then using yourself and injecting both of them. So how do you do it as a one man show effectively?
So I think that's where the challenge with like chat GPT comes in, is that if you are going to use it and get content that is not going to get AI detected, you have to put so many parameters in place when you're prompting. And you really do need to learn how to become like an expert, prompt engineer in order to get good quality results every time. And that is really time consuming and it is a process that just takes like hours and different models give you different results for things.
With marketing magic, like the vision behind it was that we have a product where you don't need to provide all of that context every time you need to do something functional in your marketing.
So it really has pulled together my experience in product management and marketing, and helps small business owners create the sales and marketing assets for different elements and projects within their business, but also to repurpose content so that they're not having to spend hours and hours creating brand new, fresh content for their marketing because they might already have a body of work or they might already have a whole lot of stuff in one format that can be repurposed into another
format. So the first way that I, the first thing that I built out actually was back in 2022, and it was like an AI tool that basically was built out in a Google sheet. And it took transcripts from 130 interviews that I'd done for summits and it repurposed them into like social media posts and blog posts using generative AI.
And it was a task that I had briefed my team to do, but it was so repetitive that it just kept getting deprioritized and I was like, there has to be an easy way of doing this. Yes, it's repetitive, but ultimately it's a reformatting activity. And so that was where it started.
But the issue with a lot of AI at the moment and a lot of AI tools is that people are focusing on maybe enterprise level projects and companies and tools, where it's like how you can build AI into your systems and your workflows.
The question I wanted to solve for marketing magic was, how do you make AI sound like you, so that you don't have to manually make all of these edits to your content in order to get past AI content detectors and to make sure that it sounds like you as the person, the business owner, the author, the thought leader, the creative. And it replicates like 80% of what you need to do.
And then that last 20% is you adding your sparkle to it on your unique anecdotes and the things that AI doesn't know about you and shouldn't know about you. And so the question really with your workflow is, okay, what do you need to do? What absolutely needs your original creative input to guide it. And then for the other things like launching a podcast episode, there are assets that you need to create.
And actually, once you understand what that looks like, you understand what the tone of voice is. There is a set standard that you have for your business that can be automated and that can be put through tools that help that preset to guide the tone of voice and outputs. The context is the thing that allows you to then take a piece of content that AI has created and either tweak it a little bit or not at all.
You can, like with marketing magic, you can put copy straight out of it into an AI content detector, and it thinks that it comes out as 100% human generated. And some of the reasoning for that is because AI uses very AI detectors are looking at things like the inconsistent use of language. The way that logic is AI deals with logic isn't great punctuation and a little bit of chaos, and then overly perfect spelling and grammar.
Often in some instances, when you're prompting things to replicate your way of speaking. The amount of energy that takes when you're doing a task by task basis almost doesn't always save you a lot of time. But that is really the framework for creating. It comes down to context. Context is the thing. Like, how do we build that in? And how do we make sure that what we're getting out is contextually relevant?
And then that context is the thing that is, makes it undetectable to AI detectors because AIH only allows you to build out context if you've given it the framework within which to function. And we see that with Adobe as well. A good friend of mine, one of my like, really close friends actually, but she's an AI. She's a, sorry, she's an ambassador for Adobe, and we talk about AI all the time.
And the other day I was like doing something in AI, and I was asking it to extend with the extend function, and I gave it a prompt based on the image, and it was, it was like a statue. And I was like asking it to add legs to the statue. And all it did was just replace the whole image with some sideways legs of a statue. And I was like, that is so far off. Okay, it's an image, but it's not at all close to the image that I wanted it to be. And it's like ultimately unusable.
And so we're seeing things roll out. And I think with the generative visual stuff, there are a lot more concerns around the intellectual property of designers and creatives because the data sets that it's being trained on are just much more minimal. People are often finding that prompting those requires you to use actual artists names as a reference point. And then you're seeing a lot of plagiarism in terms of style and all of that kind of thing.
But with Adobe, it's fascinating because it's really asking the question like, what is the limit for us to alter reality for people and basically create something that's completely fictional or imagined? And then what is our responsibility to disclose that to the people who are consuming it, whether it's on social media or it's in an art gallery?
What are the questions that we need to be thinking through around disclosing the use of AI and some of the stuff that really is changing what reality feels like for people. And going back to repurposing content. I use opus clips, for example, to repurpose this into shorts or squares for LinkedIn, because LinkedIn is still on the old school way of videos. And so is that the best use case for repurposing things, because it will add the transcripts for me.
Also, there's other things that will dub my voice. I haven't used them yet, but I've seen them dub my voice in other languages, so I don't have to find somebody to redub my voice. It does it for me. That looks natural. Are those good use cases for the generative AI? Because like, you said, we do have issues of deepfakes, like putting people's faces into something that's making them say something they didn't say. That's the cons of AI.
So are there good use cases in using that for good and not using it for tanking someone's career, for example? 100%. I think what I always would advise caution on is how readily we give our face and our voice to technology companies, because that data, once it's out there, we saw this with a 23 andme leak where it was like a million people's DNA was sold on the.net.
so I think that with giving our voices, we want to really understand what the terms and conditions are and what permissions we're granting for recordings of our voice and how they're going to be used and who they're going to be accessed by.
So I think just as far as our data security and privacy is concerned, we do need to go into adopting some of these technologies with more open eyes than we have had so far, because a lot of people would not have put stuff on the Internet if they had known it was going to be used to train an AI model without their permission. And I think that if we know that has been done, we know that OpenAI has got a lot of lawsuits at the moment. We know that is a concern.
We want to be very cautious about giving away biometric data about us as people without understanding how that's going to be used and what our right to deleting that is, et cetera. So I think that those tools are incredible from being able to build our accessibility and make sure that the things that we're doing are ultimately serving like audiences in a way that we can't do if we're limited by our own time capacity.
But I. That, yeah, the caveat for the voice thing, because I had that conversation with someone today, if you've got people who are a little bit older and they're wanting to tell stories, there's all kinds of different use cases for how that could really create special things for real humans. But there's also just the. On the flip side, what does it look like? And it's the right to be forgotten thing that Google kind of talks about quite a lot. As far as repurposing content.
I haven't actually personally tried any repurposing tools yet, which is ridiculous, but in terms of the. The tools that help with actual clips. So that is on my to do list for the next few weeks because I've just finished a big development run but with the repurposing of clips for social media, love all of that kind of thing, because it just means that you're not having to spend that time editing.
Ultimately, it's like, how can you collapse time on the things that are taking you the bulk of your energy and make sure that you're getting stuff out there without having to be chained to your laptop all the time? Within marketing magic, the repurposing is taking the sort of transcript and repurposing into other marketing materials. So it's looking at things like how can you use a podcast episode to create a, an evergreen email to nurture your email list?
How can you take it and create some blog posts out of it? Take the topics that have been discussed. Create three or four different blog post ideas. How can you repurpose the content from a transcript, for instance, into like a, a cheat sheet or a challenge or something that is actionable for your audience if you're doing any kind of online training or service provision? So that's in terms of the elements of that.
But there's also tools for like creating a webpage content and what does your homepage look like? Can you use one or all of the sections to give your homepage a refresh to make sure that your brand is positioned in the best way possible? And the idea really, with anything when we're approaching AI, is, do I, do I need to learn this now, or am I trying to learn it to make up for a weakness that I feel like I have?
And is it from a place of personal insecurity or ego that I feel like I need to learn this manually? Or can I use a tool that helps me automate this so that I don't need to learn how to do it from scratch, and instead I can do it more quickly and then spend more time in my zone of genius in the thing that I'm really good at, in the thing that I love to do? And that gives me creativity flow?
And so I think that's the question for a lot of people, is like, there are some things that you choose to do because you really enjoy it and you get a lot of professional fulfillment out of it. There are other things that you do because you feel like you should do them.
And I think the nice thing about AI is that we can take the should dos and we can think about ways to make them feel a little bit less like a chore or a burden on our to do list, and instead we can get them done quickly and move on to spending more time doing things that we know we can really have an impact with.
Yeah, it seems like for a lot of repetitive tasks, machines, AI are really good at that, and humans are pretty terrible at because we just don't like to do repetitive things over and over again because then we get bored and then our mind wanders and then we don't do it very well. Yeah, exactly. And I think what is intelligence? Right.
Artificial intelligence as we're looking at it, is ultimately automating some of the cognitive decision making processes, but they're not really necessarily big decisions to be made, often like in the repurposing of transcripts where you've already got the content and it's just a case of reformatting it as an example of that.
The way that the job market is going to shift, the way that work and the idea of work is going to shift, I think we are going to see roles that previously were around data entry, and things like that just look like a different, it's a different format, but within the AI space, we're going to see more and more jobs that are equivalent to that, but they're applying in different ways. And then also being a filter.
Being a filter and looking at how we can use human expertise to shape some of these systems and guide the quality of them as well. I always reference the data information knowledge, wisdom pyramid, because I think it's particularly good for anyone who's in the professional space to understand that AI can help with this sort of data information.
And now because of generative AI, a little bit with the knowledge, but it struggles to get to the wisdom point, because the wisdom requires us to have cultural context, emotional intelligence, and awareness of the cultural context in which we exist as business owners. And so if we are able to take our own profession and kind of look at it through the lens of, okay, what is the wise thing here? What's the wise takeaway?
What is the context relevant takeaway for this particular person audience business? That's where we continue to be able to grow our careers, our platforms, our professions, even with the use of AI, because AI is a tool that enables us to filter out some of the noise. But ultimately, we are still going to need to be the filter that ensures that the quality of what's going out there is relevant to the contexts and the environments in which we're doing business.
So let's say like a student or a marketer still in the industry, how can they start to figure this out? Because right now it's, you can use AI, but you don't need to use it, eventually it's going to be, you need to understand AI or you're just not going to get a job anymore. Yeah. So I think it's a really good thing, obviously, to be using the different models as they're coming out.
And I would always suggest that people have a hobby or a project or something they know enough about to find it interesting and to be able to understand, like, the nuance of it. And it could be, you know, it could be something that topically is completely unrelated to your professional work, but you know enough about it to be able to prompt the AI engine and see how well the results come back.
And so figuring out how to use chat GPT, being very cognizant that chat GPT, if you're just using it as is, you're putting stuff into the public domain when you do that. So you need to be super cautious about what you're putting into chat GPT. But then also try out Claude as well and see what kind of different results the large language models generate.
And then when you think about your actual day to day work and the things that you are doing day to day, I would always recommend keeping a bit of an energy audit to see what are the things that you're doing that are draining you the most. And then of those things, can you look at ways of simplifying them or making them go faster by getting AI's help with it.
So again, you need to be very careful about putting any like personal data or anything to do with like business data, customer data, your own information into some of these tools. But even if it's setting up spreadsheets and things like that, chat GPT can help with formulas and it can help with simple code fixed. If you're an email marketer and you're trying to center something, you can put the code, the HTML into chat GPT and it corrects it.
So there's little things like that where it might consume a lot of time or it might be something that you find draining. And actually you can start to build in the use of AI in your workflows. In a way that means that you're able to move on to the next thing and be able to move past the things that you don't enjoy a little bit more quickly. And you can also use it to take the initiative. Right?
If you're, if you're early or mid career, or you're a student and you're just starting out, there is information that is available to you on the Internet that people who are very senior in your industry do not know and or the amount of time that you would need to be in the industry in order to be able to get that information is really significant.
So you can use it to take the initiative if you've got big projects coming up, if there's something where you're working on a campaign, you can use it to get insights into the ideal customers, to get different ideas for how you might be able to approach the marketing or the campaign in a different or fresh way. And asking for customer insights and things like that is a really good way to make sure that your marketing, regardless of your industry size, your business size, is really impactful.
So there are ways that you can shortcut the learning process to give yourself the edge when it comes to working within a team and progressing within your career path as well. And for example, like, I use perplexity to help me with questions for like podcast guests, and it actually does pretty good job at it. So it's like that stuff would, because it's not damaging. I do like AI. Like it's very topical type of a thing. Like some of the things that would be okay for it to actually do.
Because like I said, there's perplexity. There's claude, there's gronk. If you have Twitter premium, there's, I think meta is slowly bringing out its own, on its own social sites as well. You have LinkedIn that has their own that are, unless if you're a premium member, you can actually use as well. Is it that type of thing? I think that LinkedIn, I know that notion connects in with Claude and anthropics model.
I think LinkedIn, I'm not sure if theirs is a proprietary or whether that links into something else. But Metas is interesting because you can actually download it and have it on your hardware. So it's completely disconnected from the cloud anyway. That's a nerdy little tech wormhole for another day. But as far as, yeah, using it, if there's something in the public domain as well, like it's not, it doesn't do any harm to put it through AI because it's already in the public domain.
So especially for things like podcast guest interviews, it's great because you can take like a whole load of text that is on someone's website and you can get some summary questions and pull out the things that are most interesting and then go back and skim read if you want to. But for things like that, it's great because it's helping with this data processing of potentially large amounts of information.
A lot of which you don't need to know, a lot of which you just, you're only processing it so that you can pull out of it the things that you do need. So for things like that, it's really incredibly helpful. And I just saw recently that Claude just brought out like their own web, it's their own website for their own prompts too. So that will help you like figure out how to do this, because it's just like searching through Google. There's correct prompts to do just for AI too.
There's correct prompts to do because if you don't know the correct prompt, you might not get the best answer. Exactly. Garbage in, garbage out. So anytime I'm like creating a prompt, the thing that I would suggest is that you are providing it with context. So what is the context of the instruction that you're giving it? And try to structure it super clearly and give it a very clear then instruction and a required format.
So you want to make sure that in your context, you're giving it the framework of what your role is, what your job is, what the task is that you're doing. If you've got raw text or raw data, be very clear about what the raw data is and then be very clear about what you want it to do with it. It will only respond to you based on the string of text that you put into it. It will respond based on the probability that you want to get that back.
So if you word things even slightly differently, it can dramatically change how the results come back to you because of the way that language models work in terms of probability. So if you're seeing like a really not great result, it's worth trying the same problem like two or three times to see how changing your language very slightly can impact the quality of the result that you get out of it when it comes to processing commands and things. All right?
And people listen to this podcast and they're wondering where can they find you to find more information about marketing magic or just AI in general. Yeah. So you can head over to my website. It's heymanexsha.com. there are no other manexhas I found in the marketing space, not in the UK anyway. So I'm pretty easy to find on LinkedIn and on social media.
My website has got some short form, like three minute podcast interviews, interviews, episodes that talk about the jargon behind AI and things like that. There's a few resources there if you're interested in hearing more about it, but yeah, otherwise find me, send me a DM I'm always happy to answer questions. All right, any final thoughts for listeners? Just thanks for listening.
If you're at the end of the podcast episode and AI has been a part of our economic system since the late eighties, it isn't going anywhere. It informs a huge amount of our logistics infrastructure for big companies like Amazon. Generative AI is obviously newer on the scene, but it is something that is the end result of a decades long development.
So even though it feels unnerving to tackle it, it is something that if you are going to be in the professional space in the next, even three years, you do want to spend a little bit of time just getting to grips with it and starting to understand how to use it and to feel comfortable with it and how your value as a human is not threatened by using AI as a tool, because the things that you bring are significantly more contextually driven.
As an embodied human, AI does not have the ability to experience things in the way that you have.
Just making sure that you feel super comfortable about the fact that you are more than your to do list, you're more than your ability to be super productive, and looking at your future professional work through the lens of the wisdom that you bring, the experiences that you bring, and the cultural context and creativity that you bring to roles is going to mean that you can use AI and feel much more confident. And thank you for listening to digital coffee marketing brew.
As always, please subscribe to digital coffee marketing brew on all your favorite podcasting apps with a five star review and join us next week as we talk to another great editor in the PR industry. Alright guys, stay safe, get to understanding AI and the different AI models and the correct prompts, and see you next week. Later.