Secret AI Shortcuts Your Competitors Use Daily...

There was this like extremely extremely dense influencer research and vetting process that one of our clients had. They were doing influencer vetting for brands. And so they were vetting tons of influencers all the time. And their problem, not only one, it's expensive to go through and research and make sure there's no brand safety concerns and they fit the right requirements for the campaign. All these things also takes a while. So like a brand would say, "We're ready to go. Can we start next week?" And they're like, "Oh, sorry. Our influencer vetting process takes like a minimum of 3 weeks, but they had it all documented. It's a really good process. And basically, we were able to take that and just do a little bit of translation. And then now we have AI doing the whole thing >> and hours instead of 3 weeks. >> If you're thinking about AI is purely a downsizing play and that's a little bit of a shortsighted mindset because to get AI to do work in line with how you like you want it done, like your special process. You're going to have to have people that come in and develop that stuff. Otherwise, basically you're just having the AI do like the average thing that it already knows how to do. >> You're watching from Marketing Misfits with Norm Farerrar and Kevin K. >> Senor Ferrar. How's the great white north? >> How's the great white beard? >> The great the great gay beard. >> Yeah. So, >> the great gay beard. Uh the great white north. Oh my gosh. It used to be salt and pepper. Now it's gray. >> I think I think it's from talking to you. >> It might be all the stress that I' I've put you under. Um it could be or could be it could be all those uh those uh little what do you call them? The jelly beans. No, the gummies. The little gummies that uh under your pillow. Or it could be >> it could be. So I I got to tell this story though. I got to tell the story. So Norm Norm's down in Austin visited me a few weeks ago and uh we're doing some business and doing some brainstorming and stuff and then he likes his gummies. So he likes little candies and stuff and so I was like >> not THC not just THC just regular gummies. >> Yeah. Not Yeah. Yeah. Regular gummies. Like you know like the the sugar gummies. Not Yeah. Not THC to be clear. So I I I I get come out and I have this little uh little wooden box. So you know like I don't know 3 3 in by 3 in. You're going there. >> I like I was like, Norm, I've got you I got you a little present. It's a special uh special special little treat. He's like, "Oh, all right. These must be some really nice gummies he got from some really cool place or some good chocolates or something." He opens up the the box has a sliding lid on the top. So, you just slide this little lid back. It's a small little box, you know. He slides the lid back. >> Chocolate. >> And I don't have ever heard Norm squeal as loud as he squealled. And I got called all kinds of names that start with things that we can't say here. But when he open slides that back, a spider jumps out. It's a it's a fake spider on a >> huge fake spider. >> It's a huge a huge fake spider string and it just jumps out at you and it's a plastic but it it looks real and it freaked him out so bad cuz you see Norm hates spiders. He like spiders are like just I mean there's a spider in his house. got to get his wife to come and like put the shoe on it or get it out of there. So, and it was it was it was a good one. I So, I got a good laugh out of that. But, but uh Norm got me back later on. He he said I said, "Don't worry, I'm going to get you back." And he was sitting there talking to my chef was he was cooking and something like, "What are you going to do?" He's like, "I don't know, but I'll think of something good like maybe put a Colombian in your bed." >> Right now, you have to know it's really good because I'm divorced from a a Colombian. It was really good. But but but speaking of uh of really good things and uh really cool stuff, our guest today is on the cutting edge of a lot of something that a lot of people probably don't know about. The big companies, a lot of them know about it, but she's in the AI. I think she we'll get her background. I think she started off at Facebook and did a bunch of cool stuff over there and the tech side and then she started her own agency that's helping people put in operations based on AI. And so this is not let's go use chat GBT and write an email or or NE or make some pictures for for our products or something like that. It's much more involved than that where the agentic side of AI and where you're actually using AI uh I think she calls it an operator and we'll get her definition and let her explain what that but this is something that I think is Nor and I both believe is the future and something Dragonfish our company is going to be doing. We're not doing it at this moment. We're we're we're doing a couple other things, but this is going to be it's on our on our road map to add this in, but because I think everybody's going to have to be doing this or you're going to have be at a serious competitive advantage. So, um what do we meet? We met Rachel at an event in Nashville, right, Norman? >> Yeah, we met in Nashville at uh DealCon. Dealscon. >> That's right. At deals con. And I remember she jumped in and she jumped in a text. We were going to a dinner and she's like, we we got a Uber and she's like, "Do you do you mind if I go with you guys?" And here's this, you know, young lady in in the car. And didn't you know, you never know who you're sitting next to. This hap this has happened a lot, Norm. You never know who you're sitting next to. And we're just chatting about a few things. And then later on, we find out like she's this goddess of of AI and like of agentic stuff and like the person and we had we had no idea we're sitting next to royalty. Uh so uh just the lesson there is always always uh try to connect with whoever you're in an Uber with at an event because you never know what what they might be able to do. So if you want to bring her on, do you remember how to hit that button, Norm? It's >> I can do that. I can do that's what they get paid for. >> Paid the big bucks for. >> Hey Rachel. >> Hello. Hello guys. >> Rachel, what up? >> What up? >> What a intro in a story. Oh my gosh. Uh, it's a true one, too. That's that's that's true. >> And, uh, I'll maybe replace uh, the words you use with like nerd, but we can keep going through that. >> If you want to call yourself a selfidentify, I'll let you. >> Fair. So fair. >> So, so you got your start um, doing something at Facebook, right? Um it wasn't and a lot of people when they think of they think oh she was in the marketing department or she is vetting stuff but no you were like in the tech side of things right doing something uh I remember you told me what was it you called it transformers I don't even know what the hell that is I think of the movie when when you say that but something what did you do at Facebook >> yeah um so one of those like wild things in life where you never know what you're exposed to and then later on you're like wow that was super convenient um So Transformers is the T and chat GBT and I was using that technology back at Facebook when it was still called Facebook. So I never worked at Meta. I always clarify, you know, um but yeah, I was using that technology. Um when I was at Facebook doing R&D of how to make ads work better. Um and I when I first was exposed to that tech back then, I was mind blown. And that was like the baby baby version of what we have today. And so it's just been so wild to see this type of AI just explode like it has. >> So what is a transformer? What is Can you explain like a layman's terms what is that? >> What does that do or >> Yeah, basically it's a um a type of machine learning model or AI model. There's actually tons of different types out there. Um, but it's one that's really interesting because it can learn um like representations and like understand stuff in a deeper way. And so chashy BT is a type of transformer that learns the patterns of language. And that's why we can put words into it and then it can actually understand it in its weird transformer world and then spit stuff back out to us. Um, and so yeah, it's just this like type of AI that works in this way that obviously now sitting where we're at is like crazy powerful and leading the charge of like the whole space. >> So with with ads, what was it? So it wasn't looking at language necessarily. It was looking at ad patterns or CTAs or what was it doing or what? So I think I I can't go too in depth to what I worked on because it's still in uh in uh I guess in use in the Facebook ads like platform. But um at a high level what we were doing was um helping small business ads that don't get a lot of impressions or don't learn very fast. Um we were helping those ads like kind of find other similar patterns to learn from using language. Uh, so things like your your captions are the same or your website text is the same or your products are the same, you can kind of help the ads learn faster. Um, yeah, by putting that in the massive machine learning model that predicts what ads to show you. So super esoteric at the time um and super nerdy and deep, but then it's kind of one of those like once you get a taste of something and I was just like, "Wow, this is like an amazing technology." and then I've been using it um in my last company and then obviously I have this agency now where >> So how did you go from uh Facebook with that technology to I I guess you went to two different companies your agency now and the one before. >> So um I left Facebook to actually start a like you know VC backed Silicon Valley startup. Uh we were kind of joking about wine earlier. I loved wine. So, I built a um wine specific like e-commerce stack uh that would help small wineries market better. And we were using GPT to like autogenerate the emails that they would send to their customers and the social media posts and captions. And so that's I did when I left Facebook. Um and then I ended up >> GPT2. This is like GPT2 or three before before it became a household word before this what Jasper was using behind the scenes and all that kind of stuff back in the day. >> Exactly. Yeah. Um and then again like crazy weird world I ended up um we sold that company November 2022. Uh and then everybody was like what are you going to do Rachel? What do you do Rachel? Like what do you do next? I was like I have no idea. What's the most absurd thing I could say that and uh people who know me know this was really absurd to say that I said um I'm gonna go become a Tik Tok influencer was like the most absurd thing I could think of that somebody would leave me uh not like leave me alone but you know let me go do my thing and um so that's what I did and that was November 2022 and talked about AI on Tik Tok and then December 2022 is when chatbt came out and then it blew up and so that's kind of how I got into what we're doing now which which is the agency. And then we also have the AI exchange which teaches people how to use AI and operations and stuff. >> Hey Norm, you'll love this man. I talked to a seller the other day doing 50k a month, but when I asked them what their actual profit was, they just kind of stared at me. Are you serious? That's kind of like driving blindfolded. >> Exactly, man. I told them you got to check out Sellerboard. This cool profit tool that's built just for Amazon sellers. It tracks everything like fees, PPC, refunds, promos, even changing cogs during using FIFO. >> Aha. But does it do FBM shipping costs, too? >> Sure does. That way you can keep your quarter 4 kaos totally under control and know your numbers because not only does it do that, but it makes your PPC bids, it forecasts inventory, it sends review requests, and even helps you get reimbursements from Amazon. Now, that's like having a CFO in your back pocket. >> You know what? It's just $15 a month, but you got to go to sellerboard.com/misfits. sellerboard.com/misfits. And if you do that, they'll even throw in a free two-month trial. >> So, you want me to say go to sellerboard.commisfits and get your number straight before your accountant loses it? >> Exactly. All right. So let's talk about the agency. What are you doing now? >> So um we do what I describe as AI operations. So I always kind of joke that most people don't really get lit up when they think about operations. I don't know about you guys, but is operations like your favorite thing? Okay, you do. We are the same. We are the the people. Um, most people don't like thinking about that stuff cuz it's like systems and processes and SOPs and documenting stuff, but turns out that's exactly what AI needs. You know, it needs really specific instructions and prompts and all this stuff. Um, but like agentic AI needs the same thing. It needs clear instructions to go follow. And so what our company does is helps businesses create those instructions. We call them playbooks. And then we teach those to AI and then AI can do that that playbook. >> So you create SOPs for AI task that AI can do basically. >> Exactly. Yeah. >> Okay. And so is that mostly big businesses using that or is that business of any size or what where are you seeing it being adopted first? So, our agency focuses on I'd say like still quote unquote small but small to medium uh size businesses usually. Um cuz some of our smallest clients are doing in the like 5 to 10 million a year range. Um but then AI operations is is really uh taken hold across like the whole industry. So, um, we do on the AI exchange side, we do courses and trainings and stuff for Fortune50 companies that are using these approaches. Um, and really all it is is yeah, writing an SMP that AI can can follow. And I swear it's like way sexier than that sounds, but it's what it is. >> Yeah. >> So, does that take a person to I mean, like, so when you when you sit down, you're doing an an SOP, walk me through how that works. So I come to you and I say I want to is it automate the right word or is I want to agenticize or what's the right I want to I I got three people doing this stuff and I think a robot can do it. Um what do I do? Do it to someone that knows what they're doing sit down with you and go okay this is what we do step by step by step. It's kind of like you know people say create a loom and then the loom will create a you know tools like guide or whatever that will then turn that into SOP you give to your Filipino. Is that kind of what you're doing? But from for AI and then you have a a human sitting on you have all these SOPs and you have a human sitting on top of it making sure that one of them doesn't screw up. Um is that kind of the the idea? >> Yeah, I mean pretty much exactly. So um our clients either like they have the SOP already in which case we do some uh kind of like shifting of some of the language and how it's set up to get it ready for AI. that once it's um all you need is a process documented and then you can start to teach that uh to AI piece by piece or most businesses don't have their processes already documented and so um we spend a lot of our time coming into a business doing kind of an interview or watching looms or watching videos and creating a process that then um AI can go and execute for people. Uh, and the way that we do that is this role I think you mentioned a little bit is this AI operator role which is that like processoriented person that knows how to come in and uh structure processes and then document it and then be the one managing and uh getting AI to to do that work. >> How can you trust like once you build one of your operators? Can you walk away or do you have to have people, you know, constantly watching in case it makes mistakes? >> So, it depends on the process. Um, there are, without getting like too in the weeds, the best approaches right now are that you can kind of like make a checklist of what are the things that we need it to do every single time. Kind of like a little test. And then we start running it. You can see, okay, it hits 10 out of 10 of the things every single time. or maybe 1% of the time it hit it doesn't hit that 10th thing. Then uh you can kind of see how much you trust it and decide if you want to let it just go free or if you want to keep somebody there to make sure that it that it runs. >> Chances of hallucination. This is always my problem. So I'll I'll do something 10 times and all of a sudden it'll be out of whack. Do you have to worry about that? >> I mean it depends on the process. Yeah. So, um, the way to think about it is just like if you were bringing in an intern to a business or, um, hiring somebody new, you know, you're going to, um, kind of be relying on both what they already know and what you're telling them. And the more things that you tell them about what you want done, the more control you have over how they do it. And so, um, that same thing is true for like hallucinations. the more you tell it like only use this information that you have access to or um you know double check your sources like these types of things then it it won't do the uh hallucinations nearly as much. >> Are people coming in to do this to to to down to downsize and get rid of people or people coming in to do this more to be more efficient? >> We see both. Um, our personal stance is that if you're thinking about AI as purely a downsizing play, then that's a little bit of a shortsighted uh mindset because to get AI to do work that um is in line with how you like you want it done, like your special process, your secret sauce, you're going to have to have people that come in and and develop that stuff. Um otherwise you basically are just having the AI do like the average thing that it already knows how to do. Um so like people need to be there to innovate and uh if you get rid of all your people then who's going to be innovating? Now >> as a business owner does every business out there should they be starting to look at this or is it just a specific niche? Uh, I'm just trying to I'm just trying to figure out, you know, who's going to hit you up first? Uh, is it manufacturers? Is it going to be digital marketers? What are you looking at? >> So, um, the way that we talk about this role and like we're doing is like there really is just like the next generation of operations. And so, like you guys know like in marketing like there's marketing ops, right? like that's a role of creating uh the systems and processes. Sales ops is a really strong example where there's um sometimes multiple people on a team in bigger companies that are doing like the sales ops function. Uh the same thing is true of there being AI ops like it can be there's one dedicated team for a company or we've seen it where it's embedded in a specific team. Um like uh have you guys seen the term go to market engineer at all uh kind of popping up out there? Um it's one that's starting to take uh LinkedIn by storm and the and the marketing and like uh sales world, but um it's the same thing. It's like embedding this AI operations person and mindset inside the sales team or the revenue org so they can get stuff structured and then delegate that stuff to AI so the team could work on more important things. >> Yeah, >> a lot of people stick AI on the end of their name or the end of their services and it's not really AI, it's it's if then else statements or it's it's just program programmatic stuff. What is the difference between that and true AI? I mean, nad and make.com, those are still programmic, but what's the true difference? What makes it truly AI and agentic versus just programmatic? >> Yeah, I mean, it's a good question. Um, the way that I kind of describe AI and software is like um automation is when your software does what it needs to do next like automatically, right? Like a good example is you get an email, it's in your email inbox, you open it up, you sit on it for 3 seconds, and then it goes, "Oh, now it's red." Right? That happened like automatically. That's an automation. There's no AI in there at all. Um, just if then uh statements like you're talking about. Um, AI is when your software is like choosing what to do and like making a decision, predicting what to do. It's it's not an if then automatic choice. It's actually making that decision. Um and so when you look at things like make.com or naden, they have a bunch of automations in it that those tools do, but they can also have AI steps in them. And that that's a lot of people right now are like bundling up into this like these are agents, right? Um, but it really is just like automation plus AI because it happens automatically and the AI is deciding what to do. >> Well, then the an NAN may be automation and then for the first three steps, grab this document, extract the text, whatever. Uh, and then it puts it go now go out to chat GBT and analyze this and spit me back the five bullet points or whatever about XYZ and then it passes it to the automation again and then maybe there's another step. Is that what you mean by where it's integrating the Is that agentic though? Agentic is more isn't it more like thinking on its own and like it's and modifying itself on its own and so it's not it's actually adapting to the situation and kind of um it has almost like its own little brain to an extent for a specific task. >> Yeah. Um so in the first question of the NAN workflow we're like yes it's like automations then there's the AI steps and there can be more automations. Um Agentic is a whole other rabbit hole because right now if I'm and I sure uh people listening to this like might disagree with me but um I really see agentic as being mostly a marketing term right now. there's not really like a clear technical definition of like what's agentic or not. And then I think for people if you want agentic stuff just thinking about that as like a spectrum of AI, right? Like putting one AI step in an end to end workflow that's like AI is just deciding how to do that one step. If you have AI have access to tons of steps or having it decide to make its own steps that is more um it's like further on that agentic uh spectrum but uh it's all kind of the same thing right it's just all AI put into automations and like the software tools that we use >> cuz they said they been again year this year a lot of experts came out 2025 is the year of the agents. And then I saw the Microsoft CEO change his tune this summer and say, "No, 2026 is going to be the year of agents." And then you have the Cosmo browser come out or DIA browser and then the Cosmo browser that can do some of this agentic shopping type of stuffs on on your behalf. But some people say, "Yeah, it's kind of cool, but it's pretty clunky." Uh, and it's still goofs up and you it's just like a a quarter first time um someone plays quarterback uh you know for the college team they're going to blow it because they're still trying to figure it out. I think they'll get better and better and better. But do you think we're gonna be at a point in the next few years where it's truly agentic and you you you tell it, hey, I need to do this and then it just goes off and look at like say lovable or something like or or look at a um >> or cloud code is a really good >> nano banana where you just talk to it and you just say, "Hey, do remove the uh the Coke can and replace it with a Pepsi can and it just does it." That's a little bit more towards the agentic side cuz it has to figure out behind the scenes. What were you talking about? What to go get, what to do. Is that more agentic probably uh behind the scenes on that? Uh or correct me if I'm wrong. Are we going to start seeing more of that where things are just doing stuff on their own? >> Yeah. I think it's like this spectrum, right? So like that is way further on the spectrum than what we had even six months ago. And so the way I think about it is expect the limits of how agentic or like how much control you're letting the AI decide. Like expect that to like keep going because like that it's just going to keep getting better and better. Um yeah, it's it's this this world we're in now where this is I mean people say it all the time, but like it's true. Like this is the worst the technology is ever going to be. And so like expect that it's just gonna keep getting better and better, you know? >> What do you think? So let's not think a year out or two years out. What about in the next six months? What do you see the changes being? >> So from a tech perspective, um it's always hard to predict forward, but I will say the thing that's I think um under the radar the most the last like even three months has been cloud code. Have you guys seen that or messed around with that? I hear people talking, I've not messed around with, but I hear people talking about how good it is. And I'm seeing stories now about people graduating with degrees in computer science are having a tough time getting a job because the entry-level jobs that they used to do and get paid for are now being done by claw code or or some similar version of that. >> Yeah. I mean it's um so like we develop stuff and it's completely changed our coding uh workflow and what our specifically our front end engineer even does like as their responsibilities are completely different now it's just like kind of you know last three or so months is when >> can you explain for people that don't know what we're talking about can you explain what that is and what that does >> yeah so it's basically um similar to we're talking about Kevin of you go in and you say I want you to change the color of the menu on our website to be purple instead of red. And if you give it access to your code base, so all the code that powers the website, it'll actually go in there. It'll find the right file. It'll plan out. I mean, that's a really simple task of changing the color. But let's say you wanted for it to build a whole new page on the website. It'll plan out what that page should have. It'll write all the text on that page. It'll make sure that page is linked to other pages. like it will do a full um amount of work that used to take an engineer, you know, a couple hours at least, if not a couple days. It could do that with one command. Um and so the pace at which you can now develop software is just so different. And um that kind of mentality and that type of uh capability is is just going to keep coming to like every every function and every part of where this technology is is being applied. Right. >> You're doing some of that norm. It's lovable, right? >> Yeah. Yeah. Well, no. Base 44. Oh, you did it with the cigar thing that you did. >> Yeah, I I did I couldn't believe what I could create in two, three days. >> Yeah, it was it was And I don't know like I I'm dangerous when it comes to coding. And this was like I sat there with my mouth, you know, just open like where were you? >> Yeah, he smiled. >> Norm's a huge cigar smoker. So, he created a you talked about that the thing you did for wine earlier. He created an app uh with all the cigar information so that when he's smoking a cigar, he can check it for nodes and for leaves and for tastes and smells and he can make notes in it. It spits back out and tells him about the cigar and he can log it and all this basically his own custom. >> Yeah. Your own personal humidor. Yeah. All sorts of stuff. And so when we talk about um kind of zooming back to the AI operations and like playbooks and SOPs and stuff um uh that only becomes more important when you have these agents going and doing stuff because I'll give you so how did we uh the way that you're using it norm is very kind of like uh green field right like you're like cool it's writing code the code works and that's awesome for your use case for our use case it asked a abide by the way that we do software engineering and it has to write in certain ways and in certain files and structure stuff otherwise it's not a good software engineer or software agent and so what did we do we wrote a playbook or an SOP for how to plan out in the way that we plan out how to design how to uh write the little requests to say okay I'm done with it and we wrote that whole playbook out. So now the agent knows that's it can't just do the work in whatever way it wants. It needs to do it our way so that we actually can have it be like a valuable team member. Um otherwise before we did that it wrote great code but our engineers looked at it and we're like this is crap. We're not putting this in our code base. But when we had the playbook um now it's it writes uh yeah on par with the way that we do things. So you're basically telling it the way that y'all like to do the modules and the way like you like to document and comment and all that kind of stuff. >> Exactly. >> Okay. So that a human going back into it that's used to it one way can easily find whatever it is if they need to modify it or the machine messed up or something makes it they know exactly where >> and even you know if we're going to run parallel agents at the same time like they got to work the same way otherwise they're going to overwrite each other's work you know it again you got to make the operations and the process like this rulebook and then that's how you get it to do like kind of that real business work. >> I guess expands out and you're doing more and more of these playbooks, >> the more and more complicated it is, keeping everything in sync. Like you just said about the like one overriding the other. Well, all of a sudden you have 10 of these that are depending I think they're depending or dependent on each other. Um yeah, you got to put a lot of thought into that. Yeah, I mean that's an AI operator that we have basically figured out is like the key to this is like you need someone who's nerdy about processes and documentation and figuring out what you want the to follow. >> So do you have a hub and let's let's say it's um uh I don't I don't know sales hub. So you have a major hub and underneath that hub you've got all these playbooks referring back to the hub, >> the main hub to make sure everything's in sync. >> We don't uh the way that we implement stuff right now, we're not to everything being technically uh connected in the way that you're describing. Um but uh I think the world will get there very quickly if you just have like one hub that is your company and then all your stuff is is plugged into and pulling from that. >> Yeah. >> In the next six months or so. >> That's why that's why I don't predict. So um and then you did uh so it's yeah it's just so hard to predict the um the future on the tech stuff because it's honestly it's just changing and moving so fast. Not that it's not I think it's pretty clear where stuff is going. It's just going to get better and better and more capable. But um the thing that is also really clear and then what we see is I think the adoption side of like how businesses are thinking about stuff. I'll say even like now in September is very different in terms of where people are at versus January when they're looking at AI and how they're approaching it. And so I I think the way that businesses are making progress is going to change a lot in the next 6 months. >> I see a lot of businesses you you see it every time Chat GBT or Perplexity or one of these guys makes an update. They just wipe out a thousand tools, a thousand AI tools >> that were doing a certain thing. Too many people are building businesses are building systems based on a very specific task and they therefore they don't have a moat or they don't have any sustainability instead of building it on a frontier model or something that could be interchanged and the the engine you can change out the engine and still the the core stuff on top works. So, how do you how do you build that moat on the top of AI that's every day there's something new and cool coming out that's changing. How do you build that top infrastructure so that when this gets better, you get better up here. You don't get wiped out. >> Yeah. Um, it's the same thing we're talking about. We actually have this saying we call own the playbook, rent the tech. Because what's actually really different about AI is you could take this playbook, these instructions, and you could pull them out of one tool and put them in another. And usually like a one-day turnaround for even some of the most complex stuff cuz like AI is all text based. So you just take the prompts out and put them in or you take the documents that the agent has access to out and put them in another tool. So, what you're talking about, Kevin, it's um the more that you invest in having a really good playbook, which comes from like does your business have a really good way of doing things, do you have some special sauce, some advantage? That playbook is where you should be investing as much of your time and resources as possible because then when the tech gets better, you can just boop, drop that playbook in the next tech thing and you immediately get that upgrade you're talking about. Are you looking to quickly boost new Amazon product launches or scale up existing listings to reach first page positioning? The influencer platform Stack Influence can help. That's right. Stack Influence pushes high volume external traffic sales straight to Amazon listings using micro influencers that you only have to pay with your products. They've helped upandcoming brands like Magic Spoon compete with Cheerios for top category positioning while also helping Fortune 500 brands like Unilver launch their new products. Right now is one of the best times to get started with Stack Influence. You can sign up at stackinfluence.com or click the link in this video down in the description notes below and mention Misfits, that's misfi, to get 10% off your first campaign. Stackinfluence.com. But but the tool the underlying tools could be different though. So like if let's say the underlying is chat GBT right now chat GBT 5 >> and then you take that out and you dump your playbook on top of of Gemini it's going to give you different results to a degree some similar but different results. So does the operator then have to go back in and go tweak everything again and make smooth it all out or how does that work? >> So um in some ways yes. Uh usually when we do that we have AI help so it's actually a pretty fast process but actually um if you write really specific playbooks it shouldn't vary that much of what it does in different tools. Um, and that's, you know, and most people don't kind of get to that level of specificity, but, um, think of it like you hired one person, they're really good, and you have, if you have a really, really good playbook, you should be able to hire another person of equal or or smarter caliber, and they should be able to do things the same way. Like, that's the same goal with AI is like, if you have a really good playbook, then you're protecting yourself against all of that risk. We probably have a bunch of small business owners, sellers, online sellers listening and they're probably wondering, you know, what can I do? Where where should I even start? What are some things that they should be looking for? Like tasks that they wouldn't even realize that they could implement. >> Yeah. So if you have an ops person on the team, the thing that I would have them do honestly tomorrow is look at your processes or uh whether they're documented or not and look go task by task and say could AI help us with this task? If so, let's either link that tool in this process or let's uh work on that little prompt and put that in this process. Um, but even just starting to have kind of quick wins in all of your processes of like, hey, I can do this little task, I do this little task, and save that stuff for later. That's the first and like easiest step to start um, progressing on these things. Uh, if you're the solo operator and you're running your own stuff, which I have I've been there, you're probably like, I don't have stuff written down because I'm doing all of it, right? Um, if you if you have uh something that you do repeatedly, spend a little bit of time with ChachiBT, do a loom or some sort of brain dump on what you're doing. Get Chachbt to write out your playbook and then you can just take that playbook and you can even upload it into like a custom GPT and it could start running that playbook for you and like helping you with those tasks. Um it's really like if you just kind of spend a little bit of time documenting what you want to do and kind of save that for later, then you'll start building up this um repository of these these playbooks like more quickly than you'd expect. >> So is this your process? Are you using proprietary systems and tools you've built or are you using like just know how to properly use stuff that's publicly out there? So we use a mix um because we have a really again that own the playbook rent the tech approach. So um we use like chatbt and chatbt teams a lot especially that's something that a lot of uh people like you know employees and and teams they already are using chatbt teams. So >> what is chach there >> people that haven't heard that? Yeah, it's basically just chatbt that you get um a license for like a group of people. So, uh if you have like two or three um licenses or subscriptions, then you can get um a team's account with GBT and they're all in one little closed bubble. >> Yeah. They collaborate on projects. >> So, it's got the history and everything and all in one little bubble there. Okay. >> Yeah. Yeah. >> Okay. Um, and so there's a lot of stuff you can do just in that. Uh, and if I were, so if you're new to like chatbt teams, um, and again you want like the easiest uh, highest bang for your buck tool to go try. I would really try to go learn custom GPTs, which is a small feature inside chat GBT. Um, have you guys played with that? >> Or it's like artifacts and claude. Same thing. >> Y. Yeah. And Claude. I don't know if Perplexi has that or I don't um but I know Claude and Jeb GT are the two main ones for that. >> Yeah. And then Gemini has gems, although they can do many many things. Yeah. Um, but yeah, if you again just load up again something repeatable you do, then now you have this little little AI that can help you with that task anytime you need that that task. >> Yeah. >> What are some what are in the space? What do you think is people is under the you you mentioned something else under the radar earlier, but what do you think are some other things that are happening that you're seeing that people just don't realize are happening or paying attention? Just like when you were at Facebook and you're using Chat GPG 2 uh before everybody everybody even knew what Open AI was or whatever to do your stuff. What's something else that's kind of going on the big corporate world or big underlying that just hasn't made it mainstream yet that you're seeing that's going to be like as soon as this explodes this is going to change everything? >> Yeah. Um it's a good question. I think there's a lot of stuff around AI taking over your like browser or your computer that I think some people have seen cuz there's these like viral tools and apps that will do that. >> Um, but that doesn't work super well yet and it's going to work very well like I think more quickly than we expect. So then you can think of like any website that you have a login to AI could now go log in every day and follow your playbook of I want you to scan through all of my you know um listings in this marketplace that has all these things or you know whatever kind of like task that you have to do every day that is repetitive in the browser AI can have just go do for you. So, I think that's coming. Um, also, without getting like too nerdy into it, um, there is a, uh, phrase that the, um, CTO of OpenAI used at one point where he said like the AI is, uh, extremely capable now. most of the problems that we have are going to be solved uh by engineering because they're actually engineering problems. Like basically the capabilities of AI just exploded so much that now we're the world's kind of playing catch-up. Um so I think some of these things like hallucinations uh like AI learning and remembering and all these things are just going to get better and better because there's so many engineers that are working on things. Um and so yeah, that's a very like broad sweeping thing, but I think that's going to still surprise people. >> Have you heard of motion? I I It's notion. I I think it's called motion. It's for your browser. It uh slips in as your executive assistant and as you're working, it'll take your task um and it'll slot in what you need to get done quicker. It'll go through your inbox. Hey, it's a really cool app. And when you were talking about the browser taking over, this is a start and it's not expensive. It does a fairly good job and um you know it can really uh eat in to uh your assistants um hours. They could spend it on something else because this is doing I love the idea. Um Kevin knows everything is I've got a time blocked and you know colorcoded and here's motion going in and saying well no better use of your time doing this first. It's cool. >> Yeah. Um I haven't used it recently. I'll have to check it out again. I didn't know about the browser stuff. But like that's a great example is like these companies that have these like they kind of have a playbook, right, of like how to be a really good executive assistant that now it's carrying out across your browser, which is like super cool. >> I I have a different respect now for people that are doing this stuff. I'm developing an LLM for all of my content. So I have I have a site a content hub right now that's got over a thousand pieces of content like this podcast podcast I do for another company all the podcasts I've been on all my events all my newsletters and you can query it but it's a textbased query but we're actually taking that and appending some additional data from my events and it's LLM so you can actually chat with it versus just search it and I've had to go through three people to actually get this to work right. Uh the first guy that I got it and they were playing around. I had to fire them. Then I hired another guy, actually the fourth person had to get rid of him. Then another guy came in, he he did some stuff and he got to the point he's like, I can't get this to work, right? Uh you know, and then now I've got a guy in in Tokyo that seems to know how to do all the the proper setup and the coding. And it's a it's a definite skill set that there's not a ton of people that know um how to do this really really well and really right. So are you seeing like this is this operator position? Is this someone that's technical and SOP oriented or is this something that's more they're more operationsoriented and they know a little bit of tech or is it an expensive position? >> Yeah. So what you're talking about is like super common in the space and the challenge that we kind of realized is like you either have somebody who's really technical and knows how to get this stuff to work or you have somebody that can kind of figure out what it should do, right? Write the process, write the playbook, figure out like what does Kevin even want to use this for? um not too dissimilar from like if you're making a software product, you have engineers and and uh product managers. That's kind of what we see with the AI operator. And then usually if they're doing like more advanced stuff, they'll work with an AI implement who is that engineer and that person who was in that the technical underpants if if uh have to call it that. I I like seeing uh Kevin in his technical underpants. Um you know uh could you share with us just one of your client success stories? >> Yeah. Um it's probably there was this like extremely extremely dense uh influencer research and vetting process that one of our clients had that uh when they came to us they had it all documented. They had an SOP. That SOP was like 20 pages. Like it was so detailed that their team was following and um >> this was to find and vet an influencer to promote their company, right? Just to be clear. >> So they were um they're an agency. So they were doing influencer vetting for brands for brand matching. >> Okay. >> Yeah. Yeah. Um and so they're vetting tons of influencers all the time. And their problem was not only one, it's expensive to go through and research and make sure there's no brand safety concerns and they fit the right requirements for the campaign and all these things. So, it's really expensive to do that. Um, also takes a while. So, like a brand would say, you know, we're ready to go. Like, can we start next week? And they're like, oh, sorry, our influencer vetting process takes like a minimum of 3 weeks. And so, but they had it all documented. It's a really good process. And basically we were able to take that and just do a little bit of translation to break it up. Um and then now we have AI doing the whole thing >> in like hours instead of 3 weeks or something probably, right? >> Yeah. Because you can also parallel path all the like you can bet as many as you want at a time, right? It does take each u each like vetting, you know, takes a bit because the AI is doing a lot of work like it's researching their Tik Tok and their YouTube and their LinkedIn and their Google search and like all these things. But um yeah, now it's now it's AI is doing all those steps. And so their team now has more people they can like basically they're able to spend time looking at the best of the best. So they're getting better matches for their clients. They didn't spend uh that time vetting so they can reinvest that time in other parts of the campaign. Um and they can deliver uh those matches a lot faster. So that was a that was a pretty cool one. H that's a what are what are a couple other use cases that a like an e-commerce seller or marketing person could use for this? That's a good one. But are there a couple others? Like you and I talked at the internet marketing party uh a while back where I was like telling you a little bit what Norm and I were thinking about doing where we're going to have this LLM brain and then have these agents going out and looking at uh you know through APIs or MCP pulling all this data from different tools and different systems and pulling into this brain kind of like what we talked about with the N8 thing a minute ago and jumbling it all up and then spitting out the the right answer and then sending it back out. um something something along those lines. I mean um I see I see like in the e-commerce world where right now let's take an Amazon seller. They have to do multiple tasks. They have to do product research like use a tool to figure out what are the keywords that they need to put in their listing. Then they got to do product sourcing. Go to Alibaba, go to wherever to find and deal with that. Then they got to do the marketing, create their images and create their their listing copy and all that. Then they got to do uh deal with influencers to try to get it launched. And then they got to do maintenance to make sure they're in compliance. And they got, let's call it, eight different individualized little tasks just to get this thing up and running. And I see a world where in the future where agents are doing all that. You got SOPs for every one of those things based on your requirements. And there's one operator sitting on top of it and all these agents are are doing everything. Is that realistic? >> It is. Yeah. Um, I would say I wouldn't uh expect the AI can do a 100% of all of those tasks today because there still are some like limitations to the tech. Like the browser stuff for example, Alibaba, I don't think they have an API. >> But what you're describing is where it seems very clear that the future is going. Um, and so I encourage people it's like if you don't have like an ops person or an operator yet, like I would go find one and have them start helping you document your processes. You don't need to go crazy, but just start, you know, with one part of the business. Write out the process and then try to see and test. Could we get AI to do this piece? Okay, cool. That worked. Now let's that's how we do that. and then just keep going through and and chucking along and it adds up really fast cuz now you don't have to do those things. >> And a lot of people are saying they're AI first. That's one of the things Norman and I have talked about in our is can AI do this? Do we need a copyriter or can AI do this or do we need this? Can AI do this? Um, and it's AI first. And I think that's where a was it the guy Shopify or something said before you hired an employee, you got to give me five reasons why AI can't do this or prove to me that AI couldn't do this task. And if it can't, okay, then we'll hire the employee. Um I think a lot of people aren't thinking those terms though right now. >> They're think they're still like the coolness and the novelty of AI. >> Yeah. And especially um because I hope for anybody listening Yeah. I hope that even if you're not an ops nerd, like you kind of see like see that this is um you get you get so much out of it when you can just be like, "Wow, I don't have to do that thing at all anymore." Like that example I shared, like I don't have to we don't have to do influencer vetting at all anymore. Um but I I think that because AI is just going to do it for us. But I think that um it's hard to like wrap your mind around getting there, but asking what you guys are saying, can AI do this before you even start something? Like that's that's the first step, you know. >> Can you can we do an AI system that do you do robotic stuff as well to where it can it can I don't have to talk to Norm and it will wipe the tears off of uh off of his face, off his cheek since he misses since he's going to miss me. Can Can you Can you put that I can get you the SOP? >> I I would do that anyways. I'm always crying around Kevin. >> I'm I'm sure you could make it happen as long as you got that SOP. Yeah. >> So the key the key to what you're saying that is you're not on top of this right now at least at a fundamental stage start creating SOPs for every single thing you do. Norm used to do like years ago he was I remember the first time I met him. Uh I didn't know who this bearded guy was at a conference in Cancun and but someone said that's the dude that has an SOP for how to make coffee in the office. And that's always stuck with me. Norm is the coffee SOP guy because he was very so you should be doing that in your business whether you're ready to take this leap to do what uh uh divop does or to do something else right that's the fundamental thing is start process getting everything in down so that you can try to job some of that off to AI >> yep yeah writing it down is the first step to delegating it >> what about common mistakes are there any common mistakes you're seeing businesses make right now >> with AI. >> Yeah, with AI. >> Yeah. Um, you mean just in general? >> How long is a piece of string, by the way? >> Um, I think, uh, a big one is, you know, a lot of a lot of AI projects and initiatives have failed. Um, and a lot of the root cause that we've seen is uh it's really hard to treat this as like a a solo sport. Like it really is something that you need either like almost all of the pitfalls and reasons why it can fail come back to like it wasn't kind of like an organiz the organization didn't take it seriously. So either like we hired a developer and that developer is like working on their own and doesn't have like the right requirements uh and so then that's why the project like didn't go well or didn't finish um or that you know the team's not really on board and so you build something that works great but the team never uses and um that sucks right so I think that really taking it kind of seriously as a as a as a company as a team um is important uh and then the other piece is like not treating it like magic, which is so hard cuz it feels we all feel like we're m magicians sometimes, you know, when we like go into chatbt, it's like, oh, I can tell it to do anything and it will just grant my wish. It's like so awesome. But, um, yeah, to get to like the real business value again, it's like kind of this like less sexy path, but that is just like what it takes. So, >> what does it cost to do something like with you guys to do this process? I if I let's say I have the SOPs and you don't have to come in and do the whole SOP setup and I come to you and say I want to automate this. Is it based on the complexity of the task or how what walk me through the process if I'm a new client >> uh on maybe on a couple levels or something? >> Yeah. So, um kind of depends what help you need. We like assess every business that um wants to work with us beforehand. We tell you where we think you're at. But let's say you do have all the SOPs and you're like we just want to turn this into something that AI does. Um then really uh we can provide like implementation as a service. So it's where our team will build the stuff. Um and then usually we give you guys some trainings so you kind of learn more of that AI operator piece. Um but most companies do need that AI operator and so um that's where we usually will like embed someone in your team is actually our most common model right now. we actually say, "Hey, part-time, we're going to like give you somebody. They're going to come in and they're going to be sitting next to you figuring this stuff out and driving those projects forward." Um, that's been working really well. And, uh, the like packages there just really vary based on like how big the company is and how complex the stuff that we're going after is. >> Are we looking at like 5,000 a month or 100,000 a month or what what's the range uh of of services? Yeah, we're definitely priced more in the small to medium business uh range. So like packages are, you know, 5,000 to um a good chunk or in the 5 to 10,000. And then often times what happens is things are going really well. And so we're like, cool, we can go faster if we double up on some of that stuff. And that's where we can uh be doing like multiple projects at once, all those things. So that's when you say, "Hey Kev, how long is a piece of string?" That should be your answer. >> I don't know. This one 25 right here. Right here. And then So how how are clients finding out about you? Are you are you have are you having to go out there and market? Um or is it you speaking on stages like events like Tom's event and some others? Um, is it um I don't know if you're speaking at Jordan's event or not that's happening here, but I know I saw Tom's. Um, is it is that where the clients are coming from or you got do social media marketing? Uh, or is it word of mouth or where are you getting how are you getting clients? >> Yeah. Um, we've had a wait list since we started. Uh, definitely a nice a nice place to be. Um, but yeah, we have, you know, referrals uh from our current clients. Like we'll pop those referrals up to the the top of the list cuz that's like really important to us. Um, and then a lot of it is if we're if we don't have the bandwidth to work with a client today, we really encourage them to get involved in our other business, AI exchange, that has all of these trainings and courses and it's like a lot of the same material that we train our internal team on. So at least you can start, you know, um and that way then whenever we start working together in the agency side, uh you're already like three three steps down the path and um we can go faster. So >> hey, Kevin King and Norm Ferrar here. If you've been enjoying this episode of Marketing Misfits, thanks for listening this far. Continue listening. We got some more valuable stuff coming up. Be sure to hit that subscribe button if you're listening to this on your favorite podcast player. Or if you're watching this on YouTube or Spotify, make sure you subscribe to our channel because you don't want to miss a single episode of The Marketing Misfits. Have you subscribed yet, Norm? >> Well, this is an old guy alert. Should I subscribe to my own podcast? >> Yeah, but what if you forget to show up one time? It's just me on here. You're not going to know what I say. >> I'll I'll buy you a beard and you can sit in my chair, too. We'll just You can go back and forth with one another. Yikes. But that being said, don't forget to subscribe, share it. Oh, and if you really like this content, somewhere up there there's a banner. Click on it and you'll go to another episode of the Marketing Misfits. >> Make sure you don't miss a single episode because you don't want to be like Norm. So the AI exchange is a content platform or is it it's a courses or what is what is the AI exchange? >> Um it's a membership and so if you um either if you're you know an AI operator or you're an owner who wants to learn about AI operations or um really you're just trying to learn more about how to start doing this stuff. That's why we created AI exchange. So, we have um kind of like mini courses and videos and then we also have uh some more advanced stuff in there. >> So, it's static content or that's like you going in there every month and show >> updating all the time. One of my favorite things is um all the videos are done by Rachel AI. So, it's not actually me delivering all the content. It's our AI avatar and then we have an entire set of playbooks in the back end that take all of our latest insights and convert that into the uh end product that we get. So the video, you know, planning out the content, doing extra research if we need to, scripting and then running the automation to do the avatar and then um yeah, it's pretty cool. So our content stays really up to date. Yeah, >> that's awesome. That's cool. And that's uh how much does that cost? >> That one actually >> on the spot. >> That one uh the price is uh going up before the end of the year. So um yeah, right now is is the just like the AI models. See, now Norm has taught me how to answer this one better. >> Uh no, it's we try to keep it affordable. >> It depends on how deep your pocket is. That's how much it cost. We do it as AI enforcement on you first. Like Delta Airlines, you're going to pay more for this seat. >> Yeah. Um, no, we we want all the stuff to be accessible to people because again, we can't serve everybody. And so, uh, we want to, um, make AI operations as as easy for businesses to start getting started on as possible. >> So, is that AI exchange.ai or what's where's that at? >> Um, we have the exchange.com. >> Okay, there you go. We got to get that.com. >> Now, Norm, before you before you go, she has a story about domains. >> Oh, yeah. >> You're the domain guy. >> I do. Um, it's so funny. I say the com uh you know, dots are super valuable, right? And the joke is it's so hard to get a good.com these days. Um, but as I mentioned in my um kind of like how I got here, uh, November 2022, I was like, this GPT thing is like pretty cool. I feel like this might be something. And so I bought a bunch of gpt.com domains. So I had like data gpt.com, healthgpt.com, assistantgpt.com, like I had a bunch of these. And I was just like, I'm going to probably build a company in the space. I don't know what I'm going to do. I'll just buy these. Then chatbt launched. And like 3 to 4 months later, um my email inbox was just filled with people asking to buy these domains. And so then I just was like, what's And I bought these for like $14 at the time each because they were like, who wants to buy a gpt.com domain? There was no markup on it. Um I ended up selling two of them uh for 17 grand within 4 months, which was insane. >> Wow. >> Uh because I wasn't going I was I decided what I was going to do and I wasn't going to use them. Um, so I I didn't start with the idea of like, let me buy a bunch of domains, let me sell them. But, um, then I was like, man, the ROI on this, maybe I'm in the wrong the wrong profession. So, I thought you'd enjoy that story. >> That's cool. That's really cool. >> Nora, have you sold any of yours for 17 grand? >> Uh, no. I've never sold any. They're just >> You never sold any? >> Really? No, I No, I have sold some, but No, I was just >> Well, you're paying He's paying five figures per year in renewal fees, too, to keep these things. >> Yeah. >> Mid to high five. Yeah, >> I have a lot of domains that I plan to use. Uh like I have I don't know if you're allowed uh am I allowed to curse on this podcast? >> Go ahead. Oh, I have to be >> It's an okay one. Um so I have badbitchbot.com because my personal mantra is be a bad So >> that's cool. That's cool. >> That's cool. Well, awesome. Well, this has been uh great, Rachel. I appreciate you coming on and and sharing. How do people reach out to you if they want to reach out? >> So, um I think LinkedIn is probably the easiest place to find me. If you just search Rachel Woods AI, I I'll come up. I'm pretty active over there. And then I'm also on Tik Tok. if anybody is active on Tik Tok. >> Is it really you? >> Yeah. >> Okay. Okay. >> It actually is me on the on Tik Tok at the moment. Yeah. Um so yeah, but I'd love to talk to anybody who's who's uh excited about this stuff. And again, we have an awesome community. So if you're wanting to get more involved and around fellow AI ops nerds, then um we can definitely invite you into the crew. >> Fantastic. Now, we do have one question and we ask our guests or our misfit at the end of every podcast. Do they know a misfit? >> Um, great question. And you did seed me before we started, so I did have some time to think about it. Um, I would say if you guys haven't, um, talked with or or know about Mike Stelner, um, I'm sure you might, uh, Social Media Examiner is the thing he's been doing for a while. And then he has the coolest um new endeavor in the in the AI space around making communication in a remote work environment. He is incredible. Um and I would say I don't know what your definition of a misfit is formally, but I would I would lovingly put him in that. >> Fantastic. All right. >> Well, that's it. You made it. >> Appreciate it, Rachel. This is great. Thanks for for coming on. >> Yeah, this is so fun. All right, Rachel, we will see you later. >> And don't go away. We'll just put you. We're going to remove you. >> I think if I can hit that button. >> It's the green one, Norm. >> No, no, no, it's not, Kevin. It's the blue one. >> Oh, it's the blue. You changed it. You changed it. >> Yes, of course. >> I figured you might have made it your favorite color pink. >> Uh, yes. Yes. Yes. >> Like to match my my reddish pinkish shirt. You know, I would never I ne I would never never wear pink. If you asked me 10 years ago, I never I would never wear pink. I like that's a female female feminine color. And then someone gave me a pink shirt. And I I I wore it cuz I had everything else was dirty. And I got more comments from the ladies saying, "Hey, you look good in pink. Now I have like three or four pink shirts. You see how that works, Norm? You're an old Mary guy. You don't got to worry about this stuff anymore." But you know, >> no, don't even think about it, Kev. Right. us younger whippers snappers still have to uh you know worry about those things. But no, that that was great. I think that's this is what Rachel just talked about is the future. It's something that a lot of people probably aren't paying attention to or they don't quite understand, but I think she broke it down really well and it's something that we're going to be doing at Dragonfish. Um and you know once we we get launched and going it's going to be it's on our road map of to do that not only for ourselves but to help other people. uh do that and maybe even be working with her along those ways. Uh so cool stuff. >> Yeah, I can't wait to dive into it even more. It just keeps changing every day. But uh >> yeah, that that's the problem is it's changing so fast. >> Yeah. >> But but something that also changes so fast is the the episodes. Like every single week we change out the episode to a new episode of Marketing Misfits. Every Tuesday a brand new episode just like this one comes. If you like this episode with Rachel, share it with somebody you think that might uh be interested. Forward it to or make sure you hit that like button or that subscribe button. You know what that does is we don't make any money off of that when you hit the subscribe button. But it helps it. It helps the algorithm show this to more people so that we can help expand the marketing Misfits brand. So, if you don't mind just doing us that favor, just that's all we ask. Uh just just pay us back from all the content we deliver just by hitting that subscribe button or the like button. That'd be great. And we're on a few other channels uh than whatever you're listening they're watching this on now. Maybe it's YouTube or Spotify or the talk tickers or whatever it >> we're uh and I don't know if it's us or if it's our avatars, but we are on Tik Tok. We also have two channels on YouTube. One is pretty much exclusively for long- form content, but the other, and I know we're going to be able to pick a ton of short clips from this episode, uh, is over at MarketingMisfits Clips, and that's all of our content, little nuggets, uh, three, uh, minutes and under. So, you can check those out or the just go to the website marketingmisfits.co. That's right. And we'll be back again next Tuesday with another episode. Until then, happy marketing. >> See you later. >> Ciao.