How to Avoid Hallucinations With AI

In episode #459 of SSP, Steve Simonson joins us on the show to talk about artificial intelligence and which AI tools Amazon sellers should start using. Let’s discover how AI can revolutionize your Amazon FBA business as Steve shares his insights on tools like ChatGPT, Midjourney, and more! From automations, systems and SOPs creation, image creation, branding, marketing, and email campaigns for your Amazon brand.
 
Make sure to listen to the very end as we explore the impact of AI in the Amazon-selling industry, learn practical tips on how to utilize this new technology, and how to train your ChatGPT by using powerful prompts.

In episode 459 of the Serious Sellers Podcast, Bradley and Steve discuss: 

• 02:13 – How Did Steve Get So Passionate About Artificial Intelligence?
• 05:00 – How Should Amazon Sellers Use AI Tools
• 06:15 – Text-To-Speech AI Tools
• 07:19 – Tier 1 Tool To Automations & Systems
• 12:55 – Using AI Tools For Image Creation
• 17:57 – AI Features For Branding And Marketing
• 21:00 – How AI Improves Your Email Marketing Campaigns
• 23:09 – How To Improve Your ChatGPT Prompts
• 25:48 – Common Mistakes People Make When Using AI
• 31:11 – Steve’s Healthy Habits & Habits Outside The Amazon Grind
• 32:51 – Other AI Tools That You Should Be Using
• 36:12 – How To Reach Out To Steve Simonson
• 37:07 – Steve’s 60-Second Tip
• 39:26 – Catch Steve And Bradley In Amazon Events

► Instagram: instagram.com/serioussellerspodcast
► Free Amazon Seller Chrome Extension: https://h10.me/extension
► Sign Up For Helium 10: https://h10.me/signup  (Use SSP10 To Save 10% For Life)
► Learn How To Sell on Amazon: https://h10.me/ft
► Watch The Podcasts On Youtube: youtube.com/@Helium10/videos

Transcript

Bradley Sutton:

Today we’re gonna bring one of the Amazon industry’s foremost experts on artificial intelligence. He uses so many different kinds of software, so he’s gonna give you guys all of the best AI tools that you should be using to help your Amazon business. How cool is that? Pretty cool I think.

Bradley Sutton:

Do you wanna be able to sync your listings that you create in Helium 10 to your Amazon account in one click, including being able to sync subject matter, which you’re not able to even edit. Now, in most listings on Amazon, you’re gonna want to use Helium 10 Listing Builder. Make sure to find out how to use Listing Builder by going to h10.me/listingbuilder. That’s h10.me/listingbuilder. Hello everybody, and welcome to another episode of the Serious Sellers Podcast by Helium 10. I’m your host, Bradley Sutton. And this is the show that’s a completely BS free, unscripted and unrehearsed organic conversation about serious strategies for serious sellers of any level in the e-commerce world. We’ve got somebody who’s been a serious seller for years here. Steve Simonson, back on the show. How’s it going, Steve?

Steve:

It’s going great. Yeah, I was just wondering. I’ve definitely put up some serious numbers, but I’m quite irreverent. So sometimes not always serious in the boardroom.

Bradley Sutton:

Yeah, exactly, exactly.

Steve:

Serious numbers. But also I like to crack a joker too.

Bradley Sutton:

And that’s what we like. So if you we’re not gonna go too too much into Steve’s backstory, if you guys want to check that out, episode 38 is a good one to see some of Steve’s backstory. But I brought you on because I’ve been seeing you pop up a lot and you’re good at email marketing too, and you’ve been marketing to me for some webinars and stuff that you’ve been doing lately on AI and that seems to be all the rage. And I’m one of the ones who hasn’t really completely, you know, dove into it, you know, and like, let it take over my life. Like, I, I see some people, I definitely see the advantages. I see the disadvantages, you know, sometimes too with relying too much on it. So I was like, you know what, let’s bring an expert on, and let’s just talk about, I mean, first of all, like how you got passionate about it. Like what was it about cuz you know, it’s not like you’ve been doing AI for 10 years, you know, you have a lot of specialties, you know, be it sourcing and, and logistics and, and a lot of different things. What made you so giddy about, about the the AI phenomenon? I guess

Steve:

<Laugh>, giddy is a good word for it. Well, if you look back at my background and likely in episode 38, we talked about my fascination or love for systems. Like I’m a huge systems guy. The system runs the business. The people run the system. This is right out of the E-myth. So my thoughts are like, how do you make great systems? And for more than a decade, probably nearly 15 years, we’ve been using some levels of machine learning. This is a precursor to AI. Machine learning is like, Hey, take all of your reviews and detect the sentiment of those reviews. And then the ones that are positive, try to extrapolate some of the keyword, like make a word cloud out of the, you know, positive sentiments, or the same with negative sentiments.

Steve:

So you can use machine learning, and you train it to, this word is good, this word is bad, right? It’s very I would say boring and takes a long time, and it costs a lot of money. But when you have, you know, lots of call center call records, or lots of reviews or lots of, you know, inbound data that is more than a human can do, machine learning makes sense. So from my perspective, that history of, you know, how do we make the, the bots do more work? That’s always been part of my DNA. And so, frankly, in November when Open OpenAI kind of started making a bunch of noise, we said like, I don’t know if this is a beanie baby fidget spinner thing. Like, is this just some fad or is this real? And frankly, I was skeptical. I’m like, how could it be? But after a couple weeks, I’m like, you know, this is not the invention of the automobile, right? And mass, you know production capabilities. This is not the pc, this is not even the advent of the internet. This is like fire man discovers fire, or man invents the wheel. That’s how pivotal this is for humanity. For better or worse, the trains on the tracks to use a historic metaphor, and we better get on board or get outta the way.

Bradley Sutton:

Okay. So, I’ve had a few people on here talk about, you know, some of the basic things that can be done. You know, like you mentioned it before, that was the, probably the first thing that people thought to do is, hey, let’s take Helium 10 review insights and let’s download the reviews and let’s, let’s try and get a sentiment here. And, and other people have been using it to get like, you know, maybe some different brand ideas or heard of a couple people do it for potentially even product research. But now we’re in, you know, June of 2023 what are some maybe next level things that people should do outside of just these these basics here?

Steve:

Well, the, the first thing is there are many ways of viewing AI. And I tend to look at it in tiers, right? So tier one is the basic tier that we all start with, which is, how do I use this tool? Right? There are many ways to use technologies. And the first is, you know, it’s like, I’ve got a nail. Where’s a hammer? Right? You don’t wanna use a screwdriver to try to hammer the nail. So find the best tool for the job. So, some of the things that I love to do and this may be fun for you, Bradley, just this past week, I calm my voice, and then we started having transcriptions of various things that I’ve written, tried to be, you know, read in my voice and the intonation. It’s not perfect just yet, but it’s so close that you’re like, oh, in six months, this is gonna be pretty amazing.

Bradley Sutton:

That’s crazy.

Steve:

Yeah. So that’s elevenlabs.io, If anybody wants to check that out. Spectacular. And by the way, they have, they’re actually so good at reading. Their bots are not like the, I won’t throw people under the bus, but there are many people who use text to speech tools that are very robotic, and they don’t read with any level of, I’ll just say humanity.

Bradley Sutton:

Yeah, yeah.

Steve:

If you go to 11 labs and you go to their podcast, it’s now, by the way, one of the top 50 on Spotify, at least in the business area. They’re having their bots read, like old out of copyright things like Alice Wonderland or old news articles from 1899, and the different voices they select are amazing. And you can then start to say, depending on your line of business, obviously for a podcaster like yourself, lots of applicability, whether it’s cloning your own voice or reading, you know, the thing in a new voice. Many fun things you can do with text to speech. Really extraordinary.

Bradley Sutton:

Interesting. Interesting. You just mentioned one I’ve never heard of. Everybody now has heard of ChatGPT and, and MidJourney, and maybe, you know, we can probably talk a little bit about that if you have some strategies there. But what are a couple more things that, you know, have maybe flown under the radar for, for most Amazon sellers that, that you think are, are useful?

Steve:

Well, one of the things that excites me the most is the idea of taking that tier one tool and turning it into a an automation, right? So an automation would be, so as an example, we have the ability to create SOPs that write themselves, right? So you just go to a screen, you type in, you know, how to add an item to Amazon Seller Central, or how to create a new class in QuickBooks. You click the magic AI button, and it literally writes the SOP for you, right? So that’s a tool high use, right? If you need the hammer nail, let’s get the, get the big one out and, and it, it takes that blank screen and destroys it. Like, that’s the worst thing is the blank screen. But level two is things called Auto GPT, or Baby AGI these are types of terms you may start to hear.

Steve:

Lang chain is another thing where you can actually stitch together conversations, which allows more contextual referencing of the AI, and I won’t bore the details unless we get into it. Or vector databases, which allow a lot more scanning of you know, internal data sources. So imagine everything at Helium 10, all of that knowledge baked into one Vector database, whether it’s videos, audio, text, right? You could even have, you know, word and Excel whatever you wanted. And then this thing, you can just chat with it about what do I need to know about Helium 10 internal or external, or the tool, or, you know, an individual product extraordinarily powerful. So Baby AGI has now several iterations, and AutoGPT will take that tool, that hammering process and turn it into, gosh, I’ve got a little bot that’s running around doing this for me, without being a API driven, without being Zapier driven.

Steve:

It’s just like, Hey, go in, check this inbox, and then, you know, start to create replies. You can put in a human layer of validation if you wish, until you feel comfortable with it. But at a certain point you’ll go, yeah, this thing is just gonna go to, you know, this software, it’s gonna grab these stats, it’s gonna add ’em to this, you know, database or spreadsheet, and then it’s gonna carry on with its life. Eliminating a lot of this what I would call low value add work. And that’s that it means the people can be augmented and their capabilities extended into more value add work.

Bradley Sutton:

Interesting. Interesting. Now, you mentioned, you know, systems, and then you also mentioned the outset here of what you were just talking about you know, the SOPs. You know, this video you had sent me about I think it’s called, SOP Box you, you had this like kind of cool flow where I’m assuming that, hey, this is like, Hey, I’m an Amazon seller. I need to, to make sure that I’m not the only person in my company who knows how to do this. I hire, you know, staff instead of, you know, taking a full day to write out you know, some crazy PowerPoint with a million screenshots that for, for every single process that we do, you found a way that that can kind of like, make this a, a quicker process. Can you walk us through fir first of all, tell us what that program is, and maybe walk us through a sample, a sample process that we could create?

Steve:

Yeah. So first it is so exciting to me because like I say, I hate the blank page. And really every time I’m at an event and I talk to sellers and they talk about some problem, often it’s, you know, HR related or something broke, or some system, you know, failed. And I’m like, well, where’s it written down? Well, you know, I made a video about it, or, you know, I haven’t got there yet. Like, the original sin is the, the lack of creation, right? And so we want to, we want to eliminate that and allow people to document with simplicity, like super easy. So literally just about anything that is public facing. And, and by the way, this works on the, the ChatGPTs or even Bard to a lesser extent, Bard is not quite as sophisticated at this moment, but it’s coming for chat g B t quick.

Steve:

You can just type in, you know, how to add you know, an item or how to delete this or how to do, you know, whatever function you’re trying to do. And if it’s a public facing documentation, there’s a high probability the system will just give you a step by step process. We take it a step further from that documentation, and we add process, and we say, you know what? If you’re closing your books once a month, that’s a checklist of eight or 10 steps. Let’s link the, SOP what for the new people who don’t know what, how to do it, but the existing people, they just need a checklist and they need to be reminded. So we set recurring schedules that becomes, you know, so document and then process. The operationalization is where the payoff is. Now you have accountability. You’ve made life easier for everybody who works in the company because all of these processes can be shared amongst, you know, the entire company and assigned to individuals.

Steve:

So the payoff in management, and for founders, we, we often lose sight of this cuz we’re like, it’s in my head, I, I see how it should be done. The payoff is accountability, like reports that show, yeah, here’s what happened. Here’s how productive, here’s the variance on this task between team members or between, you know, each individual day. That is it’s a powerful thing. So systems are so necessary, especially as we head into times that are maybe uncertain financially. We gotta get tight and we gotta have our processes down. So having this kind of baseline is really important in my opinion.

Bradley Sutton:

What about imaging? Like, what are, you know, be it MidJourney, be it, be it something else. What can Amazon sellers do? Cause when some of these, like DALL-E first came out, you know, like, you can’t even make a human face. It was like a months, it reminded me of the goo the Goonies movie. What was his name? The, the guy, the, Hey you guys from the Goonies. You know what I’m talking about. But that’s like how, how the, you know, when it tried to create humans, that that’s how it would look. But where are we now here in the middle of 2023 with how Amazon sellers can use AI on the imaging side? Like, like, are we at the point where we can go ahead and start creating some lifestyle images or, or perhaps even parts of a plus content or, or creating Amazon post? Where, where are we at

Steve:

For sure? We’re not a hundred percent there, but we are, well, on our way. within a year, almost everything that you do, you can do with the AI assist, right? Obviously you need your product to be able to be, you know, some base photography that is going to then be inserted into lifestyle or, or what have you. But the complexity and the geometry that has advanced just in the past, you know, six weeks is unbelievably good. And the, the level of complexity there, there’s still errors. Let’s not kid ourselves, but sure. I’m always of the, I like to look ahead, right? What’s coming, because if that trend, and, and this is again, not a Beanie Baby trend, this is a life is changing trend. If that trend is gonna get better and more sophisticated, I’ll get good at it now.

Steve:

And then a year from now, two years from now, three years from now, I’ll be well poised to maximize it. And I can say unequivocally, whether it’s MidJourney or stability AI, these are diffusion engines that are exceptional and for sure for social content, you can make spectacular images, certainly lifestyle images. And by the way, you know, sellers forget that, you know, we think the star of the show is the product, right? That’s, yeah. Because we’re egocentric or, you know, we just wanna make the sale, right? And that’s fine mm-hmm. <Affirmative>, but the customers couldn’t care less about our logo. They couldn’t care less about really the product you know, how it looks. They want the benefit that the product gives ’em. So imagery should be, you know, the, the product is a, you know, I use this, and then I felt like that.

Steve:

And so that’s where you want your imagery to, you know, start to move to that emotion that, you know, evoking of, you know what, I used to hate going camping, and now I have this particular you know, tool or, you know, tent or whatever the, the product is now my life is, is pleasant. So I’ve got this beautiful sunset photo of, you know, a tent over, you know, the mountainside or whatever it is. All of these can be created. Text to image is spectacularly powerful. We use MidJourney constantly. And when it comes to Amazon posts or any social, I would say spectacular results in fact, my, my one of my kids, my son is starting a business and they’re going to have, you know, basically an entire AI setup to create multiple daily posts that are kind of brand-centric.

Steve:

So you train the AI about your brand, here’s our values, here’s our product selection, here’s all the things about it. And this is a really important part of the framework, by the way, in ChatGPT, if you train your brand or train a conversation about your brand, and then start asking the, the the AI about your brand and say, well, you know, what’s a good elevator pitch for this brand? What would be some good target markets for this brand? What’s a good avatar within that target market? You can then start to realize, oh you know, this particular avatar, they don’t care that my teapot, you know, is made of stainless steel, grade number 39, whatever, they care that the tea comes out hot, right? Yeah. And I have a pleasant morning. And so you make that the focus and AI is well equipped to deliver all of that and much more including you know, photo studio level replacement, frankly.

Bradley Sutton:

Okay. All right. What else? What’s some specific things that, that somebody, like, somebody listening to this right now, you know, without having to go and, and, you know, learn how to do something for one week, but they could literally hop off this podcast, sign up for something, you know, and then within one hour they’re gonna have this result. Like what should an Amazon seller right now do? Like which one should they, should they start with right now? Should it be ChatGPT? And then can you give them like, some homework? I want everybody listening to this like, to take what Steve is gonna tell you. And this is something that can bring your Amazon business value and, and it’ll, you’ll be able to do it within one hour of listening to this. Is can you, can you, I’m putting you on the spot here.

Steve:

Oh, no problem.

Bradley Sutton:

Can you think of something?

Steve:

Absolutely. So let’s talk about, so every seller should be building a list, an email list, right? This is something that it’s a little harder to do on Amazon, but with the, the right methodology and, and certainly, you know, broadening into other channels, you can get an email list regardless of how big that list is. You do not need a 10,000 or a hundred thousand person list to start interacting with your customers, right? So it doesn’t matter if it’s 50 or a hundred or 500, let’s get that thing going, but use the AI to get smart about it. So one idea, right? And I like to RnD stuff Bradley, you’re probably familiar with the term r and d in the traditional sense, which is fine, which means research and development. In my world, it means rip off and duplicate.

Steve:

So let’s get to it. That’s how we do it. So, okay, the first thing, you can go to ChatGPT and just type in, you know, what are some well known effective frameworks for email marketing, right? Instead of us writing the emails, let’s get some frameworks, and it will bring back things like ADA and BAB and Pass, and it’ll explain what those are, which is essentially like, Hey, we, you know, ADA’s, attention, interest, desire, and action. So when you say, I wanna write an email in the ADA framework, it’s going to follow that framework, right? And that, that gives you, you know, kind of the, the baseline. Like, how are we gonna structure these emails? We’re gonna use one of the, the blueprints that we’re familiar with. The second is, you know, hey, what’s a good target audience for, I, I like to use the example of garden hoses, right?

Steve:

Who are customer audiences for garden hoses? And it will tell you, Hey, you know, you got your people who own homes, they’re homeowners, right? You got your garden enthusiasts, you got your, you do it yourselfers and you know, pet owners and commercial customers. It’s going to tell you the types of customer segments. And as I mentioned earlier, a great idea then is like, well, what’s a, a great customer avatar for a garden enthusiast? Give me, gimme that avatar, and it will give you chapter and verse, you know, here’s a name, here’s an age, here’s, you know where they’re located and what’s their occupation? And, you know, what are their interests? What are their goals and challenges? And this is a really important point about marketing. We get, I would say distracted by the simplicity of search term sale, right?

Steve:

That kind of the Amazon method. But as we all evolve and make no mistake whether, whether or not we think we’re evolving, the marketplace is evolving, right? And the sophistication of our competition is evolving, by the way, your competitors that used to struggle with, let’s say, you know, native language in the, in the home market, whether it’s German or Spanish or English, they no longer have that problem because AI is really good at language. Really good at language. So literally, you can type in, let’s just use Mandarin as an example. You can type in, you know, what is the English way to make, you know, actually do all the conversation I just mentioned Mandarin, and then say, now write an English email that accomplishes as objective. So really, really high use case across the world. So anyway, once you have that avatar, then you might say, okay, now give me a logical email sequence over the next six months to take a cold lead who is a garden enthusiast and meets this avatar, and then turn them into a rabbit fan.

Steve:

And it will literally give you a six month plan, you know, month by month, you know, week by week, here’s what we’re gonna start doing. And then you start creating emails for each of those parts of the sequence. And, and obviously the next thing you can simply say is, you know one of the things I like to do is I like to learn from the great. So I would next say something like, who are the 10 top direct response marketeers of all time, right? Technically they’re marketers, but I’m a Disney fan, so I call ’em marketeers, but it’ll come back with a list like Dan Kennedy and David Ogilvy and Claude Hopkins and, and all the greats. And then you just pick a voice and you go, you know what? In the voice of David Ogilvy, write a cold email for the introduction to my brand.

Steve:

You know, we’ll just call it hose guy from the founder, you know, Luke Skywalker, and use the BAB framework, and it will write that perfectly even if there’s a small, you know, edit here or there, it’s far superior than the average output that I’ve seen. So that’s kind of an RnD case study. And by the way, you do the same thing. Give me 10 email headlines with emojis, and now you have everything, right? You’ve got all of your emails, you’ve got all of your planning, you’ve got your avatars, you’re, you’re now a marketing, you, you just became a CMO or the bot, the AI became the CMO.

Bradley Sutton:

Interesting. I like it. Alright. What else can you, can you talk about? So like some, some, some of this stuff I’m so new, I don’t even know the questions to ask because like, I’m not even sure what’s, what’s possible. So like, everything you’re telling me is all education to me, as I’m sure it is for much of our audience. So what should I be asking? Like what should we be talking about here along these lines of ai?

Steve:

Well, I’ll give you two things. There’s one is that I like to say it’s all in the prompt. So if you conjure up your favorite Happy Gilmore scene, it’s all in the prompt. It’s all in the prompt. He had a funny scene. And my, my message to you is that the more sophisticated your prompt is, the better your output is. So, and when I started programming computers when I was 12 years old, the seventh grade teacher said, garbage in, garbage out. They taught us the acronym giggle. And so if you have typed in prompts and you got garbage back, it was probably a garbage prompt, frankly. So how do you advance the prompt? And the answer is, you just start thinking about it as a formula. You know, like, Hey, please generate this type of content, right? Twitter post for my target audience, right?

Steve:

Whether it’s an avatar or a general audience that highlights the benefits of my blank and encourages them to take this whatever desired action, right? So it’s, it’s a formulaic approach to prompting, and this, this applies to anything, you know, can you create a you know, email or blog post that addresses the common pain points of a garden enthusiast and positions my hose guy product as the solution. The best part of this, and I, I referenced it earlier, but I wanna be sure you guys are clear about it. If you create a conversation in GPT 4, that conversation thinks back or remembers all the prior parts of that conversation. So each conversation, you start, you should be appending with a mission. You know, AI is just a giant auto suggest. It actually has no intelligence. The more, just like on Amazon or Google, the more characters you type in, it keeps trying to fill in what it thinks you’re going for.

Steve:

And that’s all AI is. It’s a giant auto suggest. So the more in the prompt you’re extending, the more precise and the more deliberate those results will be because you’ve given it more to work with. And the more you consider that auto suggest is, is not actual intelligence, it’s just kind of a database of knowledge, and it’s, it’s on a bell curve, like, like anything else, right? The fringe data’s ignored, and it’s delivering the bulk of what it considers the available data, not it, it can’t score whether it’s right or wrong. Yeah. It goes, most people say this. So if you have a, a product that goes against the, the flow, you may often get bad replies from from ai, but does that help you kind of see where I’m headed that way? Yeah,

Bradley Sutton:

Yeah. Speaking of bad results, speaking of hallucinations and different things, what, what are some ways to prevent this? What are some some common mistakes that you’ve seen people make with AI where they probably get frustrated? Oh, this is useless, but No, it’s, it’s because you’re the useless person. You’re the one who’s doing something wrong. So like, what are some common mistakes of newbies who are getting into AI and how can we avoid them?

Steve:

Well, it’s funny because I have this conversation frequently because people have used it once or twice, and they’re like, e even myself, I, I had it, I just went to chat GBT and I said, give me a bio for Steve Simonson, and I might have even included the word entrepreneur. And it came back and it said, you know Steve’s a founder of, you know e-commerce companies and a speaker and an author, and went on. And I’m like, okay, so far it’s not terrible. Although the author feels a little weird. And then it gave the birth date, it’s, which was wrong, every part of it. And then it gave two books that I had written, which I have not written, nor do the books exist. And it gave me some lifetime award from some organization that although I’ve heard of the organization, I’m unaware of receiving such award.

Steve:

So that that prompt was not enough. Now, the more I augmented that prompt with certain details, it got better. But my message to most everybody right now is the hallucinations exist not because of things that we can control. It’s just because the data set is so large and not yet well refined. The more an individual conversation becomes granular and precise, like you could say, no, that is not right. Stop using that information again, just in this conversation. Here’s the facts, then it will use those facts. But it does not like permeate through the rest of the internet, right? That’s not how it works. Yeah. So I, I would encourage people to be patient. I would say that like doing a prompt, which is like a linear thing where you type in something and the thing comes back to you. Amazing stuff.

Steve:

By the way, there’s a site, forgive me because I don’t remember the exact address, but there’s a site, you can just go have conversations where you speak into your computer, and then the bot speaks back with the personality and voice of these celebrities. So there’s John F. Kennedy, there’s Elon Musk, and they really are different conversations. John f Kennedy’s, you know, polite very political, right? Elon Musk, I ask it like, what are your hobbies? And it’s like, you don’t have a better question than that. Come on, I’m a busy guy, right? So, yeah, yeah. Crazy, crazy iterations of, you know, these, these personalities that you can train. And one, one that I wanted to share with you, Bradley, in terms of a workflow, and it, I think it applies to any marketeer, okay? So anybody who creates content for video purposes or for audio purposes, you have a big library or potentially have a big library of audio content.

Steve:

And by the way, it doesn’t matter if that content is in French or German or Spanish or Urdu or any number of other languages, you can take using a, an open AI tool called Whisper. You can take that content in with transcribing and effic efficiency and effectiveness that is beyond compare. There, there are a lot of transcription tools in the past that flatly sucked, right? They were 90 mm-hmm. <Affirmative> percent, and that was better than nothing. But this is like 99%. And if you go to the Whisper examples, you’ll hear a guy in Scotland speaking in English, and I can barely understand the guy, and I certainly can’t understand all his words, but they, it nails it. There’s another guy speaking in French, and it nails it all to English transcription, by the way. Even a K-pop song that they, they transcribe and it, it shows you both the Korean and English words and it transcribes it.

Steve:

So this tool called Whisper takes all of that audio content, makes a transcription of it. Now you’ve got that in your database. You put that wherever you store your files, and now you can say, you know, write me a blog article about that. Write me a summary of that, you know, maybe if it’s a a longer complex thing, write me an ebook outline and then write me an ebook about that. You could also say you know, write me a blog in English. Write me a blog in French, write me a blog in Spanish, right? And, and really powerfully take what I call industrial scale marketing content and make it into a workflow. And by the way, we’re, we’re all about putting that in our system. So we, we don’t have to do this individually, we just drop files like wisp the whisper function. We drop a bunch of files in a folder, and it will then run the transcriptions, and then we can go in and say, I wanna run summaries or blogs or e-books or whatever off of all of these files. And that’s the way I want you guys to think, right? It’s always a workflow. What’s the input, what’s the processing? And then what are the potential outputs?

Bradley Sutton:

Interesting. I’m gonna ask you you know, for your 60-second tip in a little bit, and, and also get your top favorite AI tools that people don’t know about or people are not using. But before we do that, something new I’m doing on the podcast this year, and you know, I ask all the guests you know, me having health issues last year and, and, and having to be more you know, top of mind and people having mental health issues, you know, during covid and stuff. Like, I ask all the guests, what are you doing to like, stay mentally and physically healthy? Like, what’s your habits, if at all? What’s your hobbies, when you, your favorite hobbies for when you want to, you know, step out of the daily grind of, of work?

Steve:

Well, it’s so I’ve also had a little bit of a health run in you know, a few years back. And the, the moment somebody tells you have stage four cancer, you kind of wake up and go, oh, yeah, I’m not a fan. I I basically just said, you know I don’t feel it. I don’t have any problems. I’m gonna carry on and do what I gotta do. Luckily, after a year or so of chemo, I’m basically clear, and all the people ask me all the time, they’re like, well, how did that change? Or How did that, it’s like, it didn’t change me, per se, but it made me more grateful. It made me more aware of our own mortality. Like, I didn’t, I didn’t wanna die yet, and I wanted my death to be random.

Steve:

Right. And I, I wish everybody to have a random death knowing that you’re marching towards this particular deadline, not awesome. Yeah. Yeah. And so, you know, I, you know, I love to get out and I you know, when I work out, I like to listen to podcasts. That’s a hobby. Cool. Every Friday night I play Battlefield 4 with my brothers and friends. Anybody wants to jump on the Xbox? We love to shoot each other. And it’s a complete disengagement. There’s no way to think about anything except running for your life and not getting shot or even worse, stabbed by your friends. And then they replay that animation over and over, over. So I highly recommend some sort of thing that disengages your brain, cuz even at night, my brain is still going full tilt.

Bradley Sutton:

All right, cool. Cool. All right. Now what about your, you know, top three outside of ChatGPT, MidJourney, the ones we’ve mentioned today? You mentioned a few more Whisper and others three more that people should maybe go, go out and take a look at, see if there’s something there that interests them.

Steve:

Well, I definitely, again, for people, especially when you’re trying to create content or take existing content and leverage it into other areas. So for example, long form YouTube content, you can use an AI called Munch that will then break those and make ’em into shorts, and it will even, you know, break in and put in the the appropriate wording and allow you to, you know, select different break points. But mo it’ll start with an automation. It’s like, here’s what I recommend. So Munch will actually take that long form content and start creating social media shorts or TikTok style real style videos, great tool. Another one similar in function called opus.pro. My friend Jamie Davidson told me about Opus Pro, and it’s similar to Munch, but it actually comes up with a score. So you can go to opus.pro, put in a YouTube, and it will, you know, take maybe 10 minutes to process, depending on the length, and it’ll come back and it goes, this score is the best and here are what the highlights are.

Steve:

You know, here’s the hook, here’s the action. And it’s quite amazing. And again, for me, leveraging existing assets is hugely important. I mentioned elevenlabs.io. I wanna reiterate that using voice, having voice read in a hu more human way, ironically, from written content is extraordinarily powerful. So imagine using ChatGPT to create blogs, but you might even say, you know, I want this to be a narrative blog that’s gonna be read aloud, and it will then modify the format, and then you have that voice read it. You could again, start a podcast like ElevenLabs did their podcast is doing really well because it sounds, well, mostly it’s a novelty, I’ll be honest with you. That’s why people are listening now. But it sounds real enough that people don’t feel like they’re being tricked. And as long as the quality content is good, you know, it, it’s adequate, you feels good to listen to it without being tricked into it, dealing with a bot.

Steve:

Honestly, there’s so many. The other one I mentioned earlier, but II should reiterate is AutoGPT, this will, this will change the world when it matures a bit. So I, I mentioned SOPs and having the idea of having an SOP written for you or having a checklist written for you that’s cool. But having the AutoGPT, where you can just go into that SOPand you go, you know what, I’m gonna make this a bot driven function, and then the bot just reads the instructions and goes and does it, you know, based on how granular the details are, that’s, that’s where we’re heading. And so, AutoGPT,s not necessary for people to try to deploy on their own, but certainly something that they can I, I think get some benefits from if you follow. Cool.

Bradley Sutton:

Cool. Now, bef before we get into our, your your 30-second tip for everybody can you, you know, I’m sure people have many more questions or they might wanna learn more from you. How can people find you on the interwebs out there to, to perhaps reach out to get some more help on this?

Steve:

Yeah, I think the best way you know, is if they go to the, the Twitters I think that’s how they call it, the Twitters. I’m at Steve Simonson that’s also me on YouTube. I think it’s also me on Instagram. I don’t get into the social media as very often because I’m just, I don’t have the time frankly, or the interest. But I, you know, anytime somebody finds me, whether it’s on messenger or WhatsApp or you know even YouTube or Twitter, then I try to, you know, I try to help entrepreneurs. Like, I love entrepreneurs, not like people say they love cheese pizza. Like, I love entrepreneurs. I want every entrepreneur who’s really risking it all. I want to try to, you know, help them and pay it forward if I can.

Bradley Sutton:

All right. I love it. Love it. Alright, 30-second tip strategy for everybody.

Steve:

So my, my primary strategy would be to go in and train a ChatGPT four conversation by copying and pasting all the MidJourney documentation. So essentially you go to MidJourney, you find the documentation, you copy and paste that to, let’s just say a text file. And then you go to ChatGPT in a singular conversation and you paste each of those training points. You then find some of the, the best prompts that you can find on the MidJourney Discord, which you can see the format of some of these prompts. They will, you know, they’re public unless you click the stealth mode, which I like to use that’s a secondary inside tip. Use stealth to avoid the secrets, but for all the people who are doing it in public, take your best prompts that include things like cameras, filters, shutter speed, lighting settings, environments, and then structure a couple example prompts, copy and paste those in as well.

Steve:

And then you can take a, a basic prompt like Baby Penguin and it will ChatGPT will turn it into like a, a mini paragraph of just deliciously rich, descriptive prompting. And you paste that into mid journey and your level of outputs will be absolutely superhuman. And that’s, that’s you know, that was one of our first blueprints that applies not just to mid journey, but to anything, you know, train that conversation and then keep, keep going back to it. You type in a simple, let’s just call it human prompt and a superhuman prompt comes back

Bradley Sutton:

Marvel or DC superhuman?

Steve:

Well, you can decide, I’m not decide the religious war over here. Yeah.

Bradley Sutton:

That, that’s some fighting words for some here. Okay.

Steve:

Right.

Bradley Sutton:

Yeah. Well, well, Steve, this is, this has been great. I guess next time I don’t even have to bring you on the podcast. I’ll get your voice clone and ask it some, some questions if it’s been trained properly. And then we can, we can have Steve Animals every week without even you physically having to be here. So but, but thanks, you know, seriously, thanks, thanks for all the, this information. I’ve learned a lot and I’m, I’m gonna be, shoot, I’m, I might try and use some of that to start transcribing our own podcast cuz we’re paying this other service that I think is too expensive. And so, like, you might have just given me some some great tips there, but I think, I know everybody’s got some great tips. So thank you very much Steven, and we’ll definitely be in touch and I look forward to seeing you at some of these events. This summer we’re going to Bali and you’ll be in Puerto Rico too for a Billion Dollar Summit, correct? Man, we’re gonna be following each other all over the place.

Steve:

I dig it. Yeah. Well, we can do this AI stuff live and it is spectacular.

Bradley Sutton:

Love it. All right, we’ll see you there.

Steve:

Take care.