Loading summary
Transcript71 lines
- [00:00]
A
I could use your help. Every year we put together our AI Marketing industry report and thousands of marketers use it to help determine where they should focus on their AI tasks for the next year. And I need your help making this year's the best one we've ever done. We've created a very short survey that will literally only take you a couple of minutes to take. So if you could pause this podcast and head over to socialmediaexaminer.com/AI survey, we would love it. It would just take a couple of minutes. Again, social mediaexaminer.com AI survey and by the way, it's going to inform the kinds of topics we talk about on this podcast.
- [00:42]
B
Welcome to the AI Explored podcast, helping you put AI to work. And now, here's your host, Michael Stelzner.
- [00:54]
A
Hello, hello, hello. Thank you so much for joining me for the AI Explored PODC brought to you by Social Media Examiner. I'm your host, Michael Stelzner, and this is the podcast for marketers, creators and business owners who want to know how to put AI to work. Just a little heads up. This is a really good episode. I really encourage you to listen to the entire thing because I honestly think this is going to blow your mind and open up AI agents in a really functional way for you. Let's transition over to this week's episode. Today we're going to explore using Manus to create agentic workflows. My special guest is an AI consultant and educator who helps train businesses on how to take advantage of AI. She's the founder of the AI Success Lab and host of the AI Grapple podcast. Kate Vandervoort. Welcome back to the show. How you doing today?
- [01:41]
B
I'm great, thanks, Mike. Great to be back.
- [01:43]
A
Super awesome to have you here. So we're here to talk about manness, but before we get into mannus and agentic stuff in general, I want to ask this question. There's a lot of talk about agents. There's a lot of talk about agentic AI. What do you see as one of the biggest misconceptions or a few of the big misconceptions people have when it comes to agents. AI agents.
- [02:03]
B
Yeah, look, I see a lot of people trying to compare AI agents to just large language models. So trying to compare agentic AI to just ChatGPT or to Claude. And these agents are really different and we're going to explore a whole lot about why they're really different. But I like to liken it as, you know, you show up to ChatGPT or to Claude and they say, how can I help you? And what they really mean is, you tell me what you want, I'll show you how and give you a whole lot of things. But you've got to do the cutting, the pasting, the stringing things together. You've still got to do the bulk of the work. Whereas when you show up to an agent like Manus, it says, what can I do for you? And that's really the big differentiator here that a lot of people miss. And I think the other misconception is that you've got to somehow be a coder. You know, a lot of the platforms like Claude Code and Codex, they've even got code in the word. And to build these agentic systems or agents, you do need a bit of tech cap. But so much of it is done now by just natural language. And I think a lot of people are scared about agents and agents running rogue and how are they going to build them safely. And the technology's come so far and it's just going to keep evolving to the point where we can just pull these agents off the shelf. But you know, they're very different to large language models in general, just our chat based interfaces and you don't have to be a tech expert to make them work.
- [03:33]
A
I love it. Okay, we're going to talk about what the heck man is in just a second. But when people use Manus to deploy agents correctly, and we're going to talk about how to do that a little bit today, what does it unlock, what's the upside, what's the benefits waiting for people if they pay attention and do what we're going to talk about today.
- [03:50]
B
Yeah, so what I love about Manus is it really is an all in one agent. And we'll talk about the different ways to access the capability shortly. But this platform allows you to show up and do all aspects of your business and it's brilliant for marketers. And if you install it correctly, if you train it correctly, like we do with all AI, you can literally delegate a multi step workflow or job and allow Manus to go ahead and do that for you. So we're moving away from that, having to wrangle back and forth and chat back and forth. And when you do this right, Manus shows up for you for full multi step workflows. And when we explore some more of the capability, you know, it does video, it does Images, it does PowerPoints, it does Full stack app development, it does websites. So there's not a lot that Manus can't do when you compare it to the rest of the platforms that are out there. But when you do it right, it's literally like having a fully trained marketing team, if you like, within your business. And when you're using chat based large language models, you know, it's like a copywriter handing you the caption for your social media post. When you use a platform like Manus or more of the agentic AI, it's like a whole marketing team showing up to work. Very highly skilled marketing team.
- [05:20]
A
Love it. Okay. We have had a number of people on the show talking about all the options that are out there. We've had cloud cowork, we've had cloud code, we've had Codex, we've had open claw. You know, there's so many options that are out there. Explain to everyone why they should consider using Manus, because I know you've done some great stuff. You did an awesome training inside our AI business society where you showed everybody how this thing worked and it blew my mind, which is why I'm bringing you on the show. But just kind of make the case as to like, why Manus? Because it's not one that a lot of people have heard of. Surprisingly, yeah.
- [05:57]
B
So just to be really clear, I'm not a tech implementer. I'm not particularly technical. My background is very much in marketing and communications. And so I'm always looking for what's the lowest barrier to entry, what's the easiest way to get done, what it is that I'm looking to do. And I started using Manus when it first went viral back in, you know, I think it was March of last year, 2025 for me, it was the first time I'd experienced that true agentic AI. It was as easy as using ChatGPT or Claude. So for me, I was looking for what's that lowest barrier to entry. And Manness just has this. It's the vibe of the thing. I'm not quite sure how to explain it. Manus just has this way of being quite intuitive in terms of knowing what it is that you need to do. Now there are some ways that you want to maximize your use of manness and we'll certainly get into that. But the ability to just show up and tell it exactly what it is that you want to do and then have it do that is the simplest and easiest way that I've found to use agentic AI. So when we look at all of its capability and the fact that it's all in one now, don't get me wrong, you don't want Manus to be the, the key AI workhorse in your business. Manus has got particular skills and particular strengths that you to utilize and we'll go more into that. You know, you want to be careful about credit usage, but it's just, it's far more versatile than a lot of the other platforms. Everything from pure chat based right through to whole apps and agents that can be deployed in the cloud that run 24. 7.
- [07:47]
A
Talk to us a little bit about like what it costs and maybe like a little bit of its history too, because I know there's been some important things that have happened that have brought a lot of attention to it.
- [07:56]
B
Yeah, so we'll do the history bit first. So Manus was founded in, I think it was February of 2025, so you know, 18 months ago, not even. And was founded in China. And then In June of 2025 they moved the platform to or they moved to the company, their headquarters to Singapore in what they call Singapore washing to distance themselves from China and they even blocked the platform in China just to show the world that, you know, we're not a Chinese. And then in December of 2025, so less than 12 months after being founded, Meta bought Manus for $2 billion.
- [08:35]
A
That's crazy.
- [08:36]
B
Now that says something. Yeah, Meta doesn't just buy platforms unless there's something that's worth paying attention to. And so Meta purchased Manus for $2 billion and then in April of this year, so just a couple of months ago, the Chinese government have tried to block the sale. And at the moment it's a little bit unce what that's going to mean as to whether it will remain a Meta platform or not. But it looks like that probably will be unwound. But you know, Mike, the thing is, for those of us that got on board early, we saw the benefit of Manus and we've been using the platform before it was purchased by Meta. And I think for anyone that has security concerns, you know, I'm not putting matters of national security into Manus. I don't think that, you know, I'm not concerned about my Facebook analytics and my marketing strategy being accessed by foreign agents. So I think, you know, we need to put that security in perspective. And the reality is that probably 80% of the work that we do in AI is not touching sensitive data. So you're not letting this, you know, this AI into your whole business system. And that's why we talk about the value of creating an external AI. Operating system that sits outside of whatever it is that your business is hosted in so that you can consciously choose what information you're giving to AI and then you can plug in any platform. It means you're not dependent on any one platform. And then as these changes and fluctuations happen and we're going to see a lot of them as these platforms all shift and change, it just gives you some ownership, I guess, of your data that you're wanting to work with AI on. So that's a little bit about the history in terms of the costs. Manus does use credits more say than some of the other platforms. So you do want to be conscious. And we'll talk about some strategies on how to get the most out of Manus shortly. But in terms of the subscriptions, you can start, I think the lowest subscriptions, $40 USD a month and it goes right up to you can pay thousands a month. One of my AI mentors uses thousands of dollars a month in Manus credits. And again, I think we're at a point where we need to really put the cost of these subscriptions in context. When I look at what Manus delivers for my business, it is literally tens of thousands of dollars worth of value that Manus is delivering into my business. So 100 or $200 a month in subscriptions is a no brainer.
- [11:22]
A
By the way, it's only $20 a month for their basic plan, the base one.
- [11:26]
B
Right.
- [11:27]
A
And then there's the $40.
- [11:28]
B
I haven't gone back down to that one for a while.
- [11:30]
A
Yeah, and there's do month for the next one and then there's a really high one. So for the price of a chat GPT subscription here you can get started with Manus, which is pretty cool.
- [11:39]
B
That's right. But let's be clear that some of the things we're going to talk about today you're not going to be able to do on a $20 a month subscription. So yeah, we'll talk about some of that. But I think the $100 a month subscription gives you about 20,000 credits a month. One of the things I love about Manus is that they often give you bonus credits if you're a user of Manus, they give you three credits a day. And so once you build a dashboard, it doesn't cost anything to keep that running. And then if you've got a scheduled task, for example, that costs a couple of hundred credits a day, that's going to be covered by your free credit usage anyway. So we'll talk about the Best way to maximize your managed credits, because there are certainly some ways to do that.
- [12:26]
A
Let's talk about some of the things that it can do in a way that others can understand it because obviously if they've never done any of this kind of stuff, it might be kind of confusing. But I know there's a couple of cool functions or features, if you will, that make this super powerful. So where do we want to start on that?
- [12:41]
B
Yeah, so there are four sort of key ways that you access Manus, so obviously in your browser, just as a cloud based platform, like we use ChatGPT or Claude in our browser. But Manus does have the capability when you're using it in your browser, to log into all of the platforms that you use online as you. And so just to break that down, you know, when we used to have Automations and we still do, we'd use APIs to string platforms together and that's kind of giving the technology the key to the back door so that all these platforms can speak to each other. But what we can do with Manus and with other platforms, but with Manus in particular, it's very good at. Once you log into a platform and you take control of the computer when you do that. So it doesn't see your login. But once you log into a platform like LinkedIn, like any of the other platforms you might be using your CRM, whatever you're using online. Manus has persistent login capability, so it can actually go in and manage or work within your LinkedIn account. It can go and find you the 30 best people to contact for your sponsorship proposal, for example.
- [13:54]
A
Takes over the browser. Does it take over the browser? Is that effectively what we're saying?
- [13:57]
B
It can also take over your. Yes. So it takes over your browser.
- [14:00]
A
Wow, that's kind of cool.
- [14:01]
B
It can work in two ways. It either spins up its own browser, but it can now also control your browser. So that's one really powerful way to use Manus. Then we've got my computer, which is the same as say, Perplexity Computer or Claude Cowork, where Manus, when it's installed on your computer, can work with your files on your computer. So I think most people are starting to become familiar with that kind of technology. So it's a desktop app that works on your computer. Then we have Manus Agent, which is runs through Telegram, so it's a little bit like the Claude Dispatch and with openclaw, but it gives you a chat channel to be able to speak to your agent that can then Control the various aspects of Manus that's available, explain
- [14:48]
A
why that might be important for people. Do you use that feature and function
- [14:51]
B
or No, I don't use it so much in Manus, just because most of the work that I'm doing in Manus is the more beefy, meaty kind of projects. And so I'm generally at my desktop, you know, beavering away, making, making that work. But the reason that matters is that with on computer use, your computer needs to remain on in order for that to work. And if you want to go out and walk the dog or go for a walk and listen to the AI Explored podcast, you can talk to your agent while you're out via your phone, so via Telegram. And so you can still be communicating so that if your computer that's working back in the office needs your input, you can continue to control that remotely. Having said that, what Manus has more recently released is Manus Cloud Computer. And this is quite different to most of what exists at the moment. And this is quite new. It does require a separate subscription, but what this is is basically an agent that is running 24.7in the cloud and it can do all manner of things. So it could be your social media manager, it could be always on, looking for sign, looking for trending topics, putting those into a database, creating content for you. So the cloud computer has persistent databases, which means it remembers over time. So unlike Manus, when you're using it on a daily basis, it spins up a sandbox, it does the work and then it dissolves that work environment. The always on cloud computer remembers everything and everything is in a closed ecosystem. So you can have a 24 chatbot that's answering your social media messages through WhatsApp, for example. So the cloud computer, whilst it is a separate subscription, it unlocks a whole lot of new capability. And I'm still getting the hang of that. But I've set up an always on competitor, Scan and AI News Sweep, and they kind of interface a little bit so that we're keeping an eye on, you know, where the market's shifting, but also reporting that and sharing that information with our community as well.
- [17:14]
A
This is really interesting because Perplexity, which a lot of us know as a AI search engine, also has Perplexity Computer, which is effectively the same thing. And my understanding of what these are, are they're virtual computers that live in the cloud. They're like spinning up a little computer. And it does everything a regular computer does, just lives in the cloud, which is really fascinating, which means it's kind of like you're paying a hosting fee, if you will, right. To just have this computer running all the time instead of using it on your own computer. Now, when I first got on Manus, I believe you're the one that gave me the invite to it, if I'm not mistaken. Of madness. It did have like the equivalent of a little browser that would pop up. That was kind of how it worked in the early days where it would pop up. A virtual browser. Right. Isn't that kind of how it was in the beginning? Or am I.
- [17:56]
B
It's still.
- [17:57]
A
Yeah, so it's like, it's popping up instead of like Chrome on your computer, it's popping up the equivalent of a browser within your Chrome browser. And that's, that's kind of like how I remember it. So all these kind of different, for lack of better words, formats of getting access to this thing are absolutely fascinating. And I would imagine just to begin with, people can just start with the browser, right. And just kind of go from there.
- [18:19]
B
Oh, absolutely. And you don't want to go straight to the cloud computer. You want to understand Manus as the base that it is first. One more thing to mention about the cloud computer though, that I love. A lot of the more complex coding, app building, et cetera, platforms require the cli, so the command line interface. Now, to be really honest with you, when it comes to installing that stuff and getting more technical, I'm out at that point. I'm a front end consumer user of these platforms. What Manus cloud computer does is just by you describing what you want, it will install any open source software or platform that you want to use. It will manage the command line interface and do all of the coding and everything that's required to set up its own environment. So it's super easy to use. But again, you. You want to understand Manus before you jump into the cloud computer. But it is this extension that just makes it incredibly powerful.
- [19:21]
A
You know what most marketers are doing manually? Ad creative, writing the copy, coordinating with the designer for the images, waiting on video edits, managing revisions every campaign, every cycle. I wanted to fix that for our members. So on July 16th, I'm bringing Caleb Cruz to TV Teach inside the AI Business Society. He's going to show members how to build personalized AI pipelines using Claude code that runs the entire creative process on its own. Copy, images, video, revisions, all customized to your brand. And you don't need to be a coder to do it. I'll be in the room for the Entire session. I'm in every session and I can tell you members are going to walk away with a clear buildable system that they can put to work that very same way week. If you can't make it live, the full recording and resource kit will be waiting for you. Transcript, slides, key takeaways, the whole shebang. This is what we do every month inside the AI Business Society. Come and see for yourself and don't miss this training. Join at social media examiner.com AI again social media examiner.com AI so let's get into actually using Manus, because so far, here's what we know. We know that it's this really cool piece of tech that can do all this cool stuff and it can take over your browser and your computer and all that fun stuff. But now. But we also know you can interact with it like you're talking to ChatGPT or Claude. So talk to us a little bit about like just prompting in general. Right? Like how do we start interfacing with it and interacting with it?
- [20:58]
B
Yeah, so absolutely, it works just like Claude or ChatGPT in terms of being able to type in natural language. Tell it exactly what you want and it will go ahead and do that. I will say, and this is probably my biggest tip around using Manus. If you were working with an intern, for example, you might sit down and brainstorm and throw ideas out and work through step by step to help train them. And you'd kind of work collaboratively on doing that together. If you're paying A$500 an hour consulting to come in and do some work with you, are you going to want to sit there for hours doing the preparing and the brainstorming and the just shooting the breeze because you're figuring out a task? Probably not. And so the biggest difference between working with a chatbot large language model and working with an agent, particularly Manus, is that you want to come fully ready to do your task. So I suggest that people do that planning. And I actually, I write most of my prompts for any AI in a different platform to the one that I'm using.
- [22:05]
A
What are you using it?
- [22:06]
B
Well, it depends. I love perplexity for writing my prompts, but one of the things that I always teach early on in AI is if you use another large language model, you can just do a brain dump and it will structure what you want into a really comprehensive, well written prompt that is AI talking to AI. So instead of you going into any large language model and doing all the, oh, I want to do this and now I want to change that. And maybe we should add in this. If you just do a brain dump, which I often use Whisper flow, or, you know, people can just use the microphone if they don't have whisper flow, but I just press my Whisper flow button either on my phone or on desktop, and I just do a brain dump of what it is that I want. Now, the key is you need to say, please write a highly optimized prompt for Gemini Deep Research, or, you know, Manus AI Agent.
- [23:04]
A
Okay, there you go. Okay. Okay, cool.
- [23:06]
B
And do not do the task because otherwise Perplexity goes, oh, sure, Kate, I'd love to run off and do that task for you. So you don't want Perplexity or whichever platform you're doing this in. And you get five free Perplexity Pro prompts a day. So that'd be five highly optimized. And when you look at what Perplexity delivers, it's an incredible prompt that often includes things that are far more in depth or useful than I had thought of myself originally. So it allows you just to do it. A quick brain dump. But get a prompt that looks like it's been written by to prompt engineers exist anymore, but that it's been written by someone who's highly technical. And then your starting point then is here instead of, oh, hi, Chatty. This is my idea. This is what I want to do anyway. So that's my. That's my Pro tip for any prompting
- [24:00]
A
before we get into actually, like, what to prompt. I would imagine if y' all don't want to add another tool to your tool stack other than Manus. If you already have Gemini, probably try Gemini Pro. Or if you already have Claude Opus. Or if you already have Chat GPT, use the latest version of Chat GPT. Right. They're all going to do better. Right? Because they're going to help you get a really good prompt that you can now take into Manus. Right. But the real question is, like, what do we need to be thinking about? Because most people don't even know how to create a agent. Right. So when we prompt, should we be thinking about some specific thing that we're going to task it to do?
- [24:34]
B
Yeah. So if you're just doing using one LLM to prompt for another LLM, you're right. You can use Claude for Chat jbt, Chat Gemini for Claude. So that's kind of just the general prompting now when it comes to prompting for an agent. And there was something you said just then, which was, we don't even know how to Build an agent. The beauty about Manus is the agent's already built. And so you don't have to build anything. You're just showing up and you're working with an AI agent. Now, because what makes an agent is that it can pick up and use tools. It can work independently without humans having to string together every step. It can use the web. So that makes it an AI agent. Now, when we're prompting for agents, it is different. And the number one skill that you need to be able to prompt well for agents is clarity. And that's why doing your prompt somewhere else helps you get the clarity that Manus needs. Because if you show up to Manus and you go, oh, Manus, I've got this idea, I want to build this dashboard. Not entirely sure what I'm looking for, but hey, let's have a go. It's about marketing. Let's pull in this. You're burning your credits, you're having this chat. Might be a lovely chat. Manus will be really helpful. But you're burning your credit card credits. Whereas if you do that somewhere else and I, I usually say, please help me create a full brief to give to Manus AI agent that helps me achieve the following. And then I do brain dump of exactly what it is that I'm wanting Manus to build for me, or the job or the task that I'm wanting it to complete for me. And so we're delegating to Manus instead of, of wrangling back and forth through a chat. And so you're showing up to your $500 an hour consultant with a full brief that covers everything. And, you know, there's that pro tip that lots of people share. Whichever platform you're doing this in, you say, ask me all the questions that you need to complete a full brief for Manus to achieve this task.
- [26:44]
A
Okay, I'm going to go a little out of order here because I think this is the logical place to give your example. Example. And then we'll get into like some of the skill stuff, but let's give an example just so people can wrap their head around it, because you've got a really good example.
- [26:57]
B
Yeah, so we used to manually. I laugh when I say this, but manually using AI, do this process. So we would have a potential client come to us and I would go to Perplexity and I'd say, please write a highly optimized prompt for Gemini Deep Research to research this company. Company and this person. I want to Explore what their KPIs might be what their past tech transformation history is, what their future AI ambitions might be. And so Perplexity will write me the really in depth prompt. I will take that to Gemini Deep Research. Gemini Deep Research would then create that whole deep research, so like 30 pages, maybe 40 pages on all of the bits of information that I need in order to be to. Able. Able to show up and do the first call. I then use Gemini's canvas and create a web page. And so it creates a webpage for me that has conversation starters. It maps our products and services to the client and looks at how what we offer can help not only the business, but the person achieve their KPIs. And then when I show up for that first call, I have a webpage in front of me that allows. Gives me all the information in one place that I can really. And people always say, how do you know so much about our company? And sometimes I've got bits of information like, oh, you used to be a DJ. And they're like, Ah, that's not in my LinkedIn profile. And so I'd have this web page that I would use in that conversation. And then we finish the conversation and I upload the transcript into usually clause with the deep research. And then Claude would help me write the proposal. So we would manually still be the ones cutting, pasting, taking that information. With Manus, what we've been able to do is turn that into a skill, and we're going to talk about skills shortly. But the skill packages that entire process beginning to end. So it does that autonomously. It goes and does the research, it creates the dashboard, the website that I need in order to be able to have that call. The only bit that I still need to do, and I could automate this, is that when the transcript goes into a folder, it can go fetch the folder, but I still just upload the transcript into that chat thread and it then goes and creates the proposal. And our proposals are really personalized. They're not just some generic proposal where we've filled in the blanks, they're really personalized to the KPIs of the business that we're talking to. But what Manus allows us to do is to do that whole process beginning to end without that human wrangling or intervention.
- [29:45]
A
And it's because it is effectively doing the exact same steps, because it can, Right? So it's somehow going into Google, doing the deep research, pulling it into a document, all the things that you would normally do. Right. And it's doing it pretty much autonomously, which is the key to the whole thing. Right. So that saves you. Even though it was really cool, like what you did the first time blew people's minds, now you can do it at scale, significantly faster and it opens up all sorts of opportunities for you now. Right. To basically impress the heck out of prospects. And that's really, really cool.
- [30:20]
B
Our conversion rate is almost 100% based. And I put it right down to that process and the fact that we can now do that process so much quicker. And the other thing just to add, add to that, Mike, is that anything in your business or in your role, that is a repeatable task where the steps are the same and the outcome is predictable. That's the formula. Steps are repeatable, outcome is predictable, is a perfect candidate for Manus. There are other ways to do that too, but Manus makes it super simple.
- [30:52]
A
Okay, so just out of curiosity, because we know, here's what we know so far. We know that Manus uses tokens and we also know if anybody uses Claude, they also have a token system where you get limited, you know, and you might have to buy more and stuff like that. So just out of curiosity, how much do you think it actually cost you out of pocket to automate the entire process once you figured out how to do it? And how fast is it just to help people understand, like. Because if this helps you close a sale and it only costs you whatever it costs you. That's what I'm curious about. You know what I mean?
- [31:22]
B
Yeah. So I would say that the building of it, so the actual creating of it probably cost us about $15. And then to run that process each time, we're talking a couple of thousand credits, so, you know, $5 maximum.
- [31:39]
A
Okay. Wow. And how much time did it take you to actually build the process that cost you literally $15 to build and $5 to run? Like, like, just out of curiosity, how much human time. Right. Did it cost you working with in your case? Perplexity. Right. How long did it take you to actually get that back and forth figured out and then ultimately have it be built?
- [32:02]
B
Yeah. So because we were using AI for that first iteration of that process, you know, it probably was still three or four hours worth of work. And a lot of that's the formatting and the.
- [32:14]
A
It's a one time thing though, right?
- [32:15]
B
Yeah. And you would have to do that each time for each client. So once we put that into Manus and we trained manuscript on the process, and I kind of don't want to jump to the punchline, but I think I will. Skills in Manus, which we're going to unpack it in more detail. Once you do the whole process once with one click of a button, you can turn that into a skill and it will replicate that exactly the way that you did it the first time over and over again. So all you need to do in Manus is do your process one well and then turn it into a skill.
- [32:50]
A
Okay, so we're talking about a couple of hours to initially set up this agentic workflow. And it might be more if you don't have some of the foundation stuff that you've already previously done, you know, but once that's done, it's kind of done.
- [33:04]
B
But the setup in Manus wasn't even a couple of hours. The previous process probably took three to four hours. But then you just, because we know that process, you just do that all in Manus. And because we just know exactly, you know, do the research, create the dashboard. So we had the prompts already very easy like it took. Now, Manus isn't the fastest of platforms, but the output is far superior in my experience. So if you put the effort into creating that brief at the beginning, and we're going to talk more about the value of standard operating procedures and how that fits into that brief. If you put that effort in at the beginning, beginning the time to actually do the building is very quick. But you know, some of these tasks can take and maybe I can give you a couple more examples. So I was doing some work with a large food and beverage manufacturer. I was doing some work with their learning and development team and they wanted to create an onboarding video for staff about best practice in beverage manufacturing. I was there in the room with them running the workshop and I said, so just give me a quick couple of points about what that looks like. And they said, it's six modules. We need a training manual. We need an online interactive quiz. So I just prompted Manus, we want a six module training program. Best practice beverage manufacturing training manual. So they watched it, it created a 42 step to do list and it ran for almost 50 minutes. So, you know, wasn't super quick. But what Manus identified was that they needed seven modules, not six they'd missed. And it created a seven module training program with experiential exercises, 150 page training manual to go alongside that course, plus an online interactive quiz that literally graded people's knowledge on each module of that program. And I said to the people in the room, how long would it take you to do this? And they Said Kate. We've been working on that for two years and we're at step number before and we've just paid $150,000 for an external agency to come in and finish it for us. And Manus literally did it in less than 50 minutes, all without human intervention. It just ran while I delivered the rest of the workshop.
- [35:28]
A
People are getting excited listening to you right now. Okay, Skills, let's get into Skills. Explain to everybody what the heck they need to understand about skills. We've talked about skills on the show a lot, but we, we need to keep talking about it because it is something people have to wrap their head around.
- [35:42]
B
Yeah. And they work a little bit differently in Manus too. But Skills are available in Claude, in some versions of, in some of the chat GPT versions or models, and they're available in Manus. And so anything that you do, and this is what I love so much about the way that Manus does it. Anything that you do that you get an output and you go, that's amazing. And it's a task I need to do. Again, Manus will always say to you, do you want me to turn this into a reusable skill? And you can just say, yes. Now, what is a skill? So a skill is a zip file or a folder that includes the following. So the name and the description, they're the two most important parts, and I'll explain why in a moment. But the name and the description, plus step by step, instructions of exactly what you want that skill to do. But it can also include your context files, like your brand voice or your style graphics guide, or your foundation documents that you've created for your business. And it can include templates or examples of what done or what success looks like. And it's one of the key things, I didn't quite get to that when we were talking about prompting agents. It's the key thing. If you take that clarity, you need to know what success looks like. And you want to look at that from the perspective of the user or the person who is receiving whatever it is that, that you're delivering. So what does success look like? And if you can articulate that and turn that into a skill, an agent or Manus can pick up and put down those skills at will. So it will be doing a task and it'll go, oh, I think the brand voice skill or I think the proposal generator skill will be useful for this task. And the beauty about skills is it allows you to package a huge amount of information. But all Manus reads is the name and the descriptor. And so you're not chewing through your tokens by overloading context. You've got those packaged in a skill and so it greatly, it helps to greatly maximize your token usage because you're not front loading with a whole lot of information. So it can read every skill it has access to and know which ones are most relevant to the task or the job that it's doing doing. And you can just call those skills manually if you want to as well, just by doing forward slash and then writing the name of the skill. But what it does is it means anything that you do on a repeatable basis doesn't need to be prompted over and over and over again.
- [38:23]
A
Okay, well, talk to us a little bit about some of the skills that are kind of included with manis, because I think they have some. And then maybe talk about some of the skills you've created and then any other tips that might come to your mind about skills.
- [38:34]
B
Yeah, so Manus does have inbuilt skills. They've got like a video generator skill and an image skill and a brand skill. So there are some that are already inbuilt. And of course you always want to go and look at what's inbuilt before you go start creating your own. But in terms of some of the skills that we're using, and I'm going to talk about one in particular that has changed everything about how we use AI in our business, but particularly how we use Manus. So we have an SOP generate a skill, and so the skill helps to generate a standard operating procedure. And where we're heading with AI, it is essential that you know how to create a really comprehensive standard operating procedure to give to any AI, large language model or agent. And so the importance of that is that it is documenting in great detail every single step of a process. So let's say you've got an E newsletter and you do the same thing every month. You write the introduction, you choose the blog posts that go into the E newsletter, you choose your campaign or marketing material that's gonna go in, you do your call to action if those steps are the same every single time, you wanna document those. So our SOP skill, and I'm gonna give it away for free to everybody on the podcast today. But our SOP skill skill takes you through 14 different categories of exactly what is needed for AI to replicate your process or your task in intricate detail. Now we have a secret sauce with our SOPs and it changes. It literally changed everything for us. About how we use AI. And that is part of this process is it documents the human rationale behind why you do what you do. So if AI can understand the human rationale, the reason you have a paragraph for your introduction, and it includes A, B and C, then the output is 10x what it is if it just has a task list, because AI can follow that task list. But what really makes us unique is why we do what we do. And so if AI can start to understand that the output is far more aligned. Aligned and closer to what we can generate than just what AI can generate. And so this SOP skill, and I've actually created a web page in Manus so that everyone can see what a web page in Manus looks like for people to be able to download this skill. And so the SOP skill means anytime we've got a process that we want to replicate. So it's a bit of that kind of meta, you know, looking at a business from a global perspective, what are all of the tasks that we want to provide to AI? The first step is to create an sop. And if you have an SOP that then gets included in your skill with your brand voice, your style guide, whatever other context documents you want to include. And so this skill generates the SOP that then informs any other skill that you build. Every time that we do this, the first thing we do is we use whisper flow and we do a brain dump of this is the process. So we do an E newsletter. The first step is we do this. The second step is we do this. The third step is we do this. The reason we do it this way is blah, blah, blah. And you literally just do a brain dump. And then if you run that skill, it will take everything in the brain dump and populate all of the parts in those 14 sections that you've already talked about. And it'll ask you any other questions, questions that are needed to complete that skill, that sop. And then at the end, you've got a standard operating procedure that you can turn into a skill. So you've got a podcast, SOP, you've got a newsletter SOP, you've got a how we do Facebook posts, SOP. And so those SOPs, we have what we call a business Intelligence center. That's all of our knowledge files, all of our SOPs, all of our skills, they all sit in one place so that any platform can access that information. But that's how we're using skills in our business. And we literally have the skills for our podcast, for our campaign Generation for our ads process. We've got it for literally any process that is repeatable with a predictable outcome.
- [43:11]
A
Okay, a couple things. First of all, folks, skills are transferable. Meaning you can use skills in not just Manus, but you can use skills and Claude, which is really valuable. Why you might want to. To grab Kate's skill. But another thing that we've warned about is be careful whose skills you download. Right. Talk about that a little bit because we've had a lot of guests say, hey, be very careful. Now, I'm going to tell you, Kate has been on this show multiple times. She's taught inside of our community multiple times. She's a trusted source. But give everybody the warning about not just taking anybody's skills and putting them in. Talk about that for a sec.
- [43:47]
B
Yeah. So I promise you my skill is trustworthy, but you've only got Mike's endorsement in mind to go by. The real risk, though, is that skills libraries are popping up everywhere and people are even selling skills now in these free skills libraries. You have no idea who's behind those skills. And in a lot of those zip files that you're going to download onto your computer and upload into your AI agent or your AI platforms, you don't know what's contained in those. And we have, have. There are circumstances where people have embedded malicious code that can literally take over your whole computer. But that's somebody intentionally building that skill with a malicious intent and distributing that on the Internet freely. And it's the same with social media. If you, if people are saying that they're sharing skills and you don't know the person and there's no kind of credibility behind that. I would never take a skill from anybody that I didn't know or trust or have some endorsement around for that reason. Because you know what people have put into those skills.
- [44:55]
A
Here's a creative idea I just got on the fly. You tell me what you think about it. But I would imagine you could upload those files to your preferred platform. And as of this recording, Claude five Fable, which is part of their Mythos model, which is super, super powerful, which we've already tested and experimented, you could probably, say, analyze these skill files for any malicious intent inside of it. That would probably be something that I would imagine is possible. I don't know if you've ever even thought of something like that. Is it possible?
- [45:22]
B
Possible?
- [45:22]
A
What's your reactions to that?
- [45:23]
B
I've certainly thought about it, but I wouldn't even take that risk.
- [45:26]
A
Really. Okay.
- [45:27]
B
No, I would not upload a skill from anywhere other than. And to be really honest with you, the process I've just shown you or taken you through, where you create your own sop, and that is the difference that changes these skills. If it has a really well documented sop, plus your brand voice, your style guide, your products and services portfolio, whatever those those core foundation documents are and examples of done right, make your own skills, because then it's completely tailored to your business. So from my perspective, I would always rather build my own with a really well documented SOP than take something off the shelf from somebody else that doesn't have the context of my business anyway. So I'm sure you could do that, Mike, but I wouldn't even take that risk because once that skill is unlocked in, in Claude, you just don't know what it might do.
- [46:19]
A
I see what you're saying. Really good wisdom. Kate Vandervoort, we have just scratched the surface of that brain of yours. Thank you, first of all, for opening up our eyes to madness and opening our eyes also to the importance of standard operating procedures. Because this is going to be a big deal, folks. The more we begin to rely on AI to do tasks for us, the more important it's going to be that we build these things. Right. You mentioned that you're going to give it away. So tell everybody, first of all, if they're interested in connecting with you on the socials, where do they go? If they're interested in working with you, where do they go? And then where do they go to get this offer that you just made?
- [46:51]
B
Yeah, so we have a free Facebook group called AI Success Lab. The banner has me on it. There is another one that I think is in French. That's not us. I am on the banner of our AI Success Lab free Facebook group. And then our website is aisuccesslab.com and if you go to aisuccesslab.com SME you'll be able to download that skill that we've talked about today.
- [47:16]
A
Awesome. Kate, thank you so much for your time today. We really, really appreciate it.
- [47:19]
B
You're most welcome. Thanks, Mike.
- [47:21]
A
Hey, if you missed anything, we took all the notes for you over@socialmediaexaminer.com a113 and be sure to follow the show on your favorite podcasting app. If you've been a listener for a little while, we'd love a review on whatever platform you're listening on and share this with your friends. You can find me on Facebook, LinkedIn, or X if you want to tag me, tag me on any of those. And do check out our other show, the Social Media Marketing Podcast Podcast this brings us to the end of the AI Explored Podcast. I'm your host, Michael Stelzner. I'll be back with you next week, I promise. I hope you make the best out of your day and may AI help you become more successful.
- [47:56]
B
The AI Explored Podcast is a production of Social Media Examiner.
- [48:03]
A
Before you go, if you want to stop grinding through the AD creative process manually, the AI Business Society can help. Caleb Cruise is teaching members how to automate the entire process process on July 16th. Full recordings and all the resources are included with your membership. Join at social media examiner.com AI again social media examiner.com AI.