← Using AI at Work: AI in the Workplace & Generative AI for Business Leaders
Using AI at Work: AI in the Workplace & Generative AI for Business Leaders
106: Using AI Assessment Tools to Reveal Hidden Automation Opportunities with Corey Ganim
01:05:28
Loading summary
Transcript184 lines
- [00:00]
A
You also don't have to be an expert to go and sell these services. Your clients that are paying you 50, $100,000 for an engagement, they expect you to be an expert. I just have to be literally one week ahead in terms of knowledge for those people to trust me and to be able to do a good job and deliver roi.
- [00:15]
B
So let me ask you, are you a vibe coder?
- [00:18]
A
Yeah, definitely. I'm an aspiring vibe coder and I have no technical background. I got an F in computer science 1, but I was still able to figure this stuff out.
- [00:26]
B
Nobody wakes up thinking I need an audit today. Right, Right.
- [00:30]
A
We actually called an AI tools assessment.
- [00:32]
B
Interesting. But the people that you're talking to, what is the ideal state or the ideal outcome from this assessment?
- [00:37]
A
We have a guarantee around it too. Right. So the guarantee is, hey, if we can't identify at least 5 hours per week in time saving opportunity, based on what we find in the report where you implement AI, save at least five hours a week, then we will refund 100% of your money.
- [00:51]
B
What is the realistic discovery like? What are people typically coming in at?
- [00:55]
A
So the average is about six hours per week and opportunity. So if we prescribe say three tools, the average tool cost total across those tools is going to be $40 a month. If you get nothing else out of this other than this is to create.
- [01:11]
B
Corey Ghanim is helping non technical business owners cut through the AI noise, uncover hidden automation opportunities and use practical tools to save hours every week. Welcome to Using AI at Work. I'm your host Chris Staigle. Each week we'll be learning how today's business owners, entrepreneurs and ambitious professionals are getting more done with smart use of tomorrow's tech. Let's get started. Right now, every business leader is asking the same question. What are we going to do about AI? If this is you, chiefaiofficer.com has the answer. We give you a simple path forward where we provide executive and team training so your people know exactly how to safely use generative AI in their day to day. We also manage the deployment and implementation to make sure tools actually get adopted and deliver results. And we'll also guide company wide transformation so AI becomes part of your operating system, not just another shiny object. The companies that act now will increase productivity, cut costs and grow faster than their competitors. Those that wait will get left behind. So if you want to make AI work in your business, visit chiefaiofficer.com and see how we're helping companies of all sizes finally get results from AI Hi everybody, and welcome to another episode of Using AI at Work. My name is Chris Daigle and I'm the host. And rarely do we have the opportunity to be in the same room as the guest. Last time we had that was probably episode 30 something. And we were at our studio in Bastrop, Texas, which is just outside of Austin, the home of Starlink and boring company, Elon's and I guess now X. So this is a podcast where we aren't necessarily talking about theory. We're not talking about. I've got this great idea. We're talking to people who are doing the thing. And that seems to be kind of like the, the, the common universal element of all of our guests. And today, no different. Corey Ganim is our guest and we'll be introduced, letting Corey introduce himself in a second. But I want to give everybody a backstory. So in February of this year, beginning of February, I guess it was, I was had the opportunity to go to Fort Lauderdale in some big, you know, Pablo Escobar, Tony Montoya kind of mansion with about 30 other early AI founders. Fantastic group. And one of the individuals that was there is our guest today. And pretty quickly I realized, okay, there were still some people that were kind of exploring that were in that room. How they got in there, I don't know. But Corey was one of those people who was actively touching, breaking, bending, seeing what was working. What's, you know, what does it take to make this viable for a business and that sort of thing. And left paid attention to what he's been putting out. And his content is very accessible and, you know, like, certainly good for all levels of individuals who are wanting to discover how do I use AI at work. So before we get into the conversation, Corey, welcome to the show. Thank you for taking the time out of your day. And maybe just we'll pull an Eisenberg. What do you want people to walk away from this episode with?
- [04:14]
A
Yeah, Chris. So first of all, thanks so much for having me. And I do really enjoy getting to do these in person too. I feel like it's a totally different vibe than, than online. So, yeah, I can, I can go into my background, but as far as what I want people to get out of this, I mean, if anything, just a better understanding for how AI works, which tools are working at this point in time. Right. We're May 2026. That's always going to change, but also the underlying principles, like the tools are going to change, but then there are certain principles, certain ways of using AI and applying it That I think is never going to change. And then as well as, you know, I'm sure we can go into the, the AI assessment business model, why that works, why it can be sold as a service, whether you're starting out or, you know, all the way to the highest levels. I mean, something you guys are, are actively doing in your business is assessing businesses and telling them where they can implement AI. So there's a lot of different directions that we can go, but I think first principles is a good place to start.
- [05:11]
B
So let me ask you, are you a vibe coder?
- [05:15]
A
Yeah, definitely. I'm, I'm an aspiring vibe coder. So I, I know, I, I know how to be dangerous with, you know, Codex and Claude code. And I've used all the agents, whether it's Claude, Cowork or openclaw or more recently, Hermes. I spent the whole morning configuring some new Hermes agents. But yeah, I, I know how to vibe code with. And I have no technical background. Like, I literally, I tell people this all the time. I got an F and computer science won my freshman year of college. And, you know, so if that tells you enough about my technical background, hopefully it does, but I was still able to figure this stuff out.
- [05:51]
B
So let me ask you, are you an influencer?
- [05:55]
A
I don't really like that word.
- [05:56]
B
I don't like it either, but yeah, like, how do you describe the impact that you're having?
- [05:59]
A
Because when I hear influencer, I feel like, you know, I feel like. Or just like hot girl who's like, pedaling. Yeah, like a makeup brand. Yeah. But like, so I don't know, I guess if you want to use that term, that's fine. But I mean, really my goal with, well, one with content creation, I do it because it's fun and I genuinely enjoy it. Like, that's one thing that I really enjoy throughout my week is, you know, if I come up with an idea for, oh, like this post, this would be a good idea or this would resonate or this would really help somebody. I'll like, file it away in a notion document and then sit down one day and just create all the content. Right. So I enjoy it. But in general, I feel like it's. I feel like I have a teaching, I guess, spirit, for lack of a better term. And I feel like I have the unique ability to convey things in a way that's easy for people to understand. So I feel like I'd be doing people a disservice by keeping it to myself. And so the fact that I Enjoy it, I think is kind of the icing on the cake.
- [06:56]
B
So content creation, especially with AI, like, like it's nothing has a long shelf life. Right. It's going to go stale within a week or a month or whatever the case might be. So what are you, are you creating the content as you're exploring the tool? A new tool comes out. Hey guys, I'm going to check this thing out kind of thing. Are you getting your chops with it first and then saying, hey, this is what, this is what I like about it this way. I don't like about it, like for somebody, for our audience, if they're not familiar with you and they want to start consuming your stuff, which I do. Right. And one of the things I like about the way that you put it out is I don't have time for the big, the big things. Right. But you're giving very good, like TLDRs or executive summaries of this is what the tool is this how I'm using it. Here's a few ideas. And I find that type of content very like, boom, got it. On to the next one. So what is the content creation process look like for you and how do you integrate it into what, like the deliverables that you're doing for clients?
- [07:51]
A
Yeah, so I think one, or at least how I approach it is I like to be, you know, I like to obviously know what I'm talking about before I go out there and talk about it. But it's like, what's the minimum amount that I could have used this tool or done with this workflow to be able to talk knowledgeably. So, for example, like, Claude Design just came out, you know, a month ago at this point. And I purposely didn't put out anything about cloud design on day one or even day three, because I hadn't used it yet. So I waited until I had at least uploaded a brand kit and built, you know, one or two slideshows before talking about it. So I guess the, you know, the moral there is like, I need to know enough to be dangerous. But that's a lot, like, a lot less than most people realize before I go out and talk about something or try to teach it because of how quickly, you know, how, how many things are coming to market on a weekly basis and how, how quickly stuff's changing.
- [08:44]
B
Yeah, you know, I, when I'm in person with clients and they're asking about tools I take them to, there's an AI for that dot com.
- [08:51]
A
Oh, I tell people I plug that site all the Time. You would think I was an affiliate. Yeah.
- [08:54]
B
The reason I do it is to show them how many tools are out there. And their database, last I checked, which was last week, was like almost 50,000 tools.
- [09:01]
A
And it segments it by industry. Right? Yeah. So that's pretty cool.
- [09:05]
B
If you haven't checked it out, audience, go check it out. It's good.
- [09:07]
A
Yeah, I tell people that all the time. Like, you know, when we do, when we do assessments, for example, for clients, you know, obviously we prescribe very specific tools for their workflows, but a lot of times I encourage them. I'm like, look, we found these tools on. There's an AI for that.com or Futurepedia IO is another similar, basically competitor.
- [09:27]
B
It'll be in the show notes.
- [09:28]
A
Yeah, yeah. And it's like, look, if you, you know, the tools that we gave you are great and they're going to help with what you're doing specifically. But if you're just curious, if you want to know what else is out there, go poke around for yourself and go sign up for the free trials or, you know, sign up for a month and test it out. I mean, there's, you could, you could go two lifetimes of testing AI tools at this point, and by the time
- [09:49]
B
you finished, all the ones that you had already tested were new again.
- [09:52]
A
Right. And that's why I think it's important too, is like, I think you need to talk about these tools and start putting out content before you're like truly a quote unquote expert. Because by the time you feel like you're an expert, either the tools outdated or it's changed completely. Like, it's almost like you have to get to an MVP level of knowledge and then immediately start talking about it or it's. You've kind of lost the opportunity.
- [10:16]
B
So for the listener, the takeaway for that, and I feel the same way, like, you don't need mastery with any of these tools. They're designed to be easy to use. But also if you're like, if you see something, you don't have to get 10,000 hours on claw design to be able to put it to work after two or three sessions. And if you listen to other podcasts and they're talking to the, you know, the tool expert. I was listening to one the other day, a big name podcast, and Guy was like, yeah, I've been using it for three weeks. Right now he's doing like a big podcast. So as a listener, I don't want you to think that it takes a lot more Than that once you, once you got like, get your head right about thinking in AI, we call it, right? That, like, please don't, please don't think that there's some big learning curve.
- [10:59]
A
Use the tools and another way to look at that too, right? And you know, obviously you, you and your team, you guys are experts. You guys have been in the weeds of this stuff for a long time. Like well before it was cool, even well before Chad beat. But a lot of times, like the, the demographic that I'm speaking to are these people that are newer to AI or they're newer to selling services or being a consultant. And the, the point that I try to drill into their heads is like, look, yeah, you don't have, you certainly don't have to be an expert to talk about this stuff or create content about this stuff. You also don't have to be an expert to go and sell these services, right? Like, you know, your clients that are paying you 50, $100,000 for an engagement, they expect you to be an expert. But when I'm out there selling, you know, an AI assessment to a small business who, who has never used chat GPT, I just have to be literally one week ahead in terms of knowledge to, for those people to trust me and to be able to do a good job and deliver roi. I actually had somebody push back on that last week. He was like, well, I don't, I don't agree that, you know, I think you need to be an expert before you sell these things and before you do these things. And I'm like, dude, you're, you're wrong. This is an industry that it's so new that there are no true quote, unquote.
- [12:13]
B
Bingo.
- [12:14]
A
Yeah. So, you know, if you put in a week, if you know how to build a cloud skill, if you know how to create a context file, if you know what markdown means, like all of these things that are so basic and take you 24 hours to learn makes you qualified to sell and to
- [12:30]
B
talk about, you know, it's a couple of things. It's interesting because we've been training teams for three years now, I guess, and the expectation was that they could prompt right now, sure, that that's helpful. But now the reality is that if a knowledge worker, staff level knowledge worker that doesn't know how the basics of maybe Claude code, maybe cowork, maybe code, like if they don't know the basics of that stuff, if they don't know whether or not they ever run Claude code from the terminal or anything, if they don't like if they've never seen that before. Like I would say that the, the baseline for me to qualify somebody as a capable staff level person has risen a lot just in the past three or four months.
- [13:15]
A
Yep, 100% agree. And knowing the right tool for the job too is kind of one of the, like a point that you could draw out from what you just said. I think that is as valuable of a skill as any right now is knowing which tool to use for the job. Right. Like, for example, there's, there's plenty of little scenarios where it's just a, you know, one input in, one output out scenario. And it's like if somebody wants to go and automate that, creating a clawed skill might be overkill. Right. That might just be a simple zapier zap, which is not even an AI workflow. It's literally just a workflow automation. And then there's things that are more technical or more complex. And then for those workflows it's like, well, what makes more sense here? Do we turn that into a skill? Do we turn it into five individual skills with an orchestrator skill that calls each five in parallel? Does it, should it be a Claude code routine or do we, do we build this in Codex because of X, Y, Z? Right. Like there's, I think, certain decision making points that again, if you haven't been in the weeds and you don't know what each tool can do, it's harder for you to make those judgment calls. So you just, either you freeze and you don't do anything or you make the wrong decision.
- [14:26]
B
So for those of you listening, if you're not familiar with this concept of a skill file or a skill MD file, it's kind of like for us, we did a lot of work in ChatGPT for a number of years because it was the better model, right?
- [14:38]
A
Yep.
- [14:39]
B
And we built, so as a result, we built a lot of custom GPTs that were in the SOP environment. Hey, this is how, this is how the company wants it done. Use this as your, your co pilot, your guide, right?
- [14:50]
A
Yep.
- [14:50]
B
And then we kind of made internally and clients even made a switch to Claude over the past several months. Now I'm kind of going back to GPT 5.5 a little bit. But skills are essentially, you can use them the same way that you would a custom GPT set, instructions, anybody can call it and get, you know, a likely result.
- [15:12]
A
Yep. And the kind of how I explain it to people is so a skill is kind of like a recipe, right? And so for people who are listening, if you're familiar with the concept of an sop, a standard operating procedure, a skill is kind of like an SOP where it's, hey, this is how we do a task, right. It's, this is the checklist. We're doing that task and a skill just turns that checklist or turns that recipe into a repeatable process that you can just hand off to the AI. So it's like we turn the SOP or the recipe into a skill and then we hand that skill to the AI agent and then the AI agent can go and execute that skill. And what's cool about skills is that the AI agent is going to do that process the same way every time. Right. There's, there's no deviation. Whereas if you were just to prompt it or if you were to, you know, just do it from scratch every time, it might get a step wrong or you might forget to include step seven or, you know, more is left up for interpretation.
- [16:07]
B
Yeah. Or if you're just leaving it up to the individual to be like, oh, I don't need a list.
- [16:11]
A
Yeah. Or I can just, I can just prompt it. And it's like, well, that's. Yeah. Skills are just more efficient and from like a token use perspective too. Right? It's, it's way more efficient to just. For the AI or the LLM to just call a skill. It knows exactly what it needs to execute. It does it the same way every time. It's just cheaper overall too, you know,
- [16:30]
B
and you bring it up like another thing that I hadn't heard of or really wasn't a big deal, even just maybe two months ago, a month ago was token budget.
- [16:40]
A
Yeah, right.
- [16:40]
B
Like, and all of a sudden if you're training these people, like you go back to that concept where I said at a base level, they should know how to do this stuff. If they're doing that stuff, they're probably plugging into an environment where tokens are getting burnt. So now all of a sudden you've got all these people that are creating all these experiments that costs money, not the $20 a month subscription costs real money. So kind of explain to the listeners why this, this token budgeting matters and what, like, what can happen? What's the downside of.
- [17:10]
A
Yeah, so again, for the, the folks in the audience here who maybe you've been on a free plan or you've been on like a $20 a month chat GBT or Claude plan, and you've never run into, it's called a rate Limit, Right. And all a rate limit is is it's basically you ran out of your usage for your subscription. So when you run out of usage, you have to wait a certain number of hours or a certain number of days for your usage to reset. Once it resets, you can start using the AI again, whether it's Cloud or ChatGPT. And so, you know, how do people run into those limits? I mean, simple enough, they're just, they're just using it a lot, right? And why is that important? Because a lot of times people are using the AI in a way that's not token efficient. So for example, and I was guilty of this early on when I didn't know any better, I'm sure a lot of people, their default way of using AI is they just have one long chat thread, right? So it's like anytime I need to ask Claude a question or chat GPT a question, I just ask it in the same thread that I've asked at a hundred other questions. And the issue with that is every time you ask another question in that chat or you put another prompt in that chat is it has to go back and process every other earlier prompt in that chat all over again. So, you know, by the time you get 10 prompts deep in a single thread, it's having to process 10 prompts every time instead of just the one. Right? Your most recent question. So like the, the takeaway for the audience, that's probably the simplest thing you can stop doing to save tokens is every time you, you know, you switch topics or you log into cloud or chatbots again, just open up a new chat and you know, I, I'm not going to pretend to be like an expert on token efficiency or, you know, these micro optimization strategies, but overall it's like I tell people, you don't need a higher tier plan. Like a lot of people, they start off with AI and they're like, oh, I'm just going to get the $100 a month plan, I'm going to get the $200 a month max plan. And I'm like, that's not necessary. Like just start off with a 20amonth and once you start hitting rate limits consistently, then upgrade to 100amonth. And then once you start hitting rate limits consistently there, then upgrade to 200. But you know, you only want to, I guess, upgrade when you need to. I feel like a lot of people are so gung ho, they want to go like max plan from the start.
- [19:28]
B
So great advice for sure with that. Nobody needs that mega Thread that's got everything from vacation plans.
- [19:34]
A
Yeah, yeah, that's how I used to do it.
- [19:36]
B
Yeah. But so if you're an active user, I guess this makes, this is where projects would make a lot more sense, where listeners, if you're not familiar with it, project is, you know, multiple chats around a certain subject. And it's not necessarily calling the context of every chat, but if it needs to, if you prompt it to go out there and like, evaluate what it's done, you've created this kind of contained memory on that subject without the mistake of this massive. A token eater of a thread.
- [20:05]
A
Right? So one, again, the way that I like to describe projects to people. And so again, just to kind of differentiate. So ChatGPT has what are called custom GPTs, and Claude has something that are called projects. They're the same thing, right? They're just. ChatGPT's version is called custom GPTs, Claude's version is called projects. And so what I tell people is think of a project or think of a custom GPT as just siloed context, right? So, for example, I'm a big fan of Russell Brunson. I know you're familiar with Russell Brunson. So I wanted to create a Claude project that's basically Russell Brunson's brain. So that way, anytime I ask a question inside that specific project, I'm getting an answer based on Russell Brunson's teachings or his content or whatever. So how I set that up is I took the PDF copies of all three of his books, Expert Secrets, Traffic Secrets and.com Secrets, and I turned those PDFs into, I think it was like 11 or 12 individual context documents. So basically, Claude took the full, you know, PDF transcripts, chunked it down into these individual context files, and then I saved those context files as the. What are called the project knowledge inside that project, right? So you've got the project knowledge, which are the context files. And then to tie it all together inside the project or inside the custom GPT, you can add what are called project instructions. And all that is, is that's like the, the system prompt for the project. That is what tells Claude, hey, Claude, Here are the 11 context files that are inside your project knowledge for this project. Here's what each one means, here's when you would use each one, right? Here's what each one does. It's what ties everything together. So once you have your, you know, your project knowledge and then your project instructions, then every time you ask a question or put a prompt inside that specific project, it's going to function based on your project instructions and based on the context documents that you gave it. So now, for example, inside my Russell Brunson project, anytime I go to write a new email sequence or fix my positioning or write a landing page or anything like marketing direct response related, I always do it inside that project because it will give me the output based on the, you know, the project knowledge, all the Russell Brunson information and it will reference it specifically. It'll say, hey, know, you need to write the email sequence like this because this is how Russell structures his soap opera sequences. Or this is how Russell, you know, leaves a cliffhanger at the end of each email to get somebody to open the next email. Right?
- [22:56]
B
Yeah. So couple things that I'm taking away from that. I wasn't using project instructions like that. So how I'm seeing this as you're describing it is almost like a skill file where you open up the folder and it's like, here's the instructions, here's what's in this, this skill. Use this if you need that. So it's almost like a same mechanism, right?
- [23:14]
A
Except it's almost like a broader skill, right, Where a skill is for a very specific business process. Like for example, I could have a skill that writes email sequences and I do. But if I use that skill inside of my Russell Brunson project, it's going to kind of take into account like, hey, here's like the strategy behind this email sequence which is going to be at the project level and then the actual like physically writing the copy is going to be done by the skill. Right? That's like the very specialized. This is how, you know, this is a subject line we use. This is all the various process specific information is going to be handled by the skill.
- [23:52]
B
So you mentioned three books, but you turned it into 11 context documents. Why?
- [23:57]
A
Because Claude told me to. I, yeah, so I went into Claude and I said, I literally when I was coming up with the idea for this project, I told it exactly what I wanted to do. I said, hey, I want a project that is my Russell Brunson brain. And I even asked it, I said, well, hey, I have a PDF copy of each of his books. Obviously they're like 250 pages each. I was like, claude, can I just include these three PDF files as the project knowledge and then you can just use that. And it was like, well, no, that would actually be a bad idea because for one, you're asking Claude every time it gives you a response, it's going to have to read, basically read three books and then give you a response which is going to eat up your entire context window in one prompt. And two, it's obviously going to be very expensive. So the, the 11 or 12 context files are what Claude recommended. It was like, well, hey, based on, you know, the content of each book, we need these context files. One is like positioning, two is email sequences, three is, you know, your icp. Right. And it kind of just broke it out.
- [25:03]
B
Yeah, yeah.
- [25:04]
A
And then I approved that plan and I said, okay, that looks good. Here are the books, right? Here are the PDF copies. Generate the context documents. And it did it one after the other. It generated the 11 or 12 individual markdown files to where all I had to do is just save those and then upload them to the project knowledge inside the project.
- [25:23]
B
Did you review them?
- [25:25]
A
Yeah, for sure. And like, I mean I didn't read everyone word for word because they are pretty lengthy. But like at a high level I was like, you know, it gave me the plan. It was like, hey, I think we need these 12 context documents. And I was like, okay, based on what I know about Russell, like that makes sense, right? So I approved that. And then it started. It built them one by one and like the first one I read through and they're all pretty. They just take his concepts and just segment them.
- [25:50]
B
Yeah.
- [25:50]
A
Is essentially what it's doing. So yeah, it's, it worked really well. And I, I use that project every week.
- [25:56]
B
Yeah, I bet. So when we were teaching, I guess new users about AI, we teach them this concept. I learned it from my buddy Mark moss. But it's 10, 80, 10. The first 10% is you being clear on what you want. Next 80% is after you hit enter, the model does its work. Final 10% is you putting, you know, like making, taking it from synthetic to authentic. Right. How much 10% on the back end are you doing once you've. Because this is a big investment of time and thought and building and all that. How much better is the output than if you. Obviously you're saving time because you're not having to reintroduce how you want things done. But are you spending less time on the back end?
- [26:38]
A
I'm definitely spending less time on the back end, yeah. So like, for example, right. To use that email sequence example again. So that, like I had that project, that Russell Brunson project, I had it right, an email sequence that it was a five day sequence that goes out to anyone who downloads one of our lead magnets. Right. Our Our, the template for how we deliver the AI assessment. So you know, I gave it a simple prompt essentially, hey, here's what the lead magnet that somebody's opting in for. This is what they get. This is our offer that I need you to CTA in the fifth email, write me the email sequence like Russell would write it. And then, you know, 60 seconds later I have a full five day sequence. And so that, you know, the 10% was me specifying more. So just like here's what our offer is and here's what somebody's opting in for. So make sure the email sequence is relevant. It writes the whole sequence. And then when I go to transfer that into our email marketing software, really the only changes that I'm making are around like formatting. And I mean there's some, of course I read everyone like that's something I'll read beginning to end. And there's some sentences where I'm like, this is redundant or this sounds kind of weird, like I'm just going to rewrite it. I'd say I spend, you know, 5% of my effort actually critique on the back end.
- [27:58]
B
Yeah, yeah.
- [27:59]
A
You know, dialing things in. I could have shipped it unchanged, copy paste, but it just, it wouldn't have felt right.
- [28:05]
B
It's good to know that if you needed to copy.
- [28:07]
A
Yeah, I mean, yeah, you absolutely could
- [28:09]
B
you get away with that.
- [28:10]
A
And another, I mean, another reason that I think that works, at least for my setup is because I have a really strong underlying brand voice skill and that's something that everybody needs to have. Like if you're listening to this, if you get nothing else out of this other than this is to create a brand voice skill. And so all that is is it gets called any time Claude or chatbots needs to write something that sounds like me. Anything from an email to a marketing material to a tweet to a whatever, it uses the brand voice skill. And you know how that works is and you can ask Claude like, hey, interview me to put together a brand voice skill. But essentially at its core it's like, hey, this is how Corey sounds. These are words that he likes to use. He's concise, he's to the point, he doesn't beat around the bush. These are words that he never uses. And the reference files for that skill are five of my podcast transcripts. So like I could go and take the transcript from this podcast here, give it to that skill so that it has the most up to date essentially way of how I talk. Right. As well as written Examples, So it's like, hey, also in those reference files are 10 of my best LinkedIn posts or 10 of my best tweets. And so that way it has a really good idea of, one, how I talk because of the podcast transcripts. Two, how I write because of the social media posts that it has access to.
- [29:33]
B
You know, I did so 100%, we're going to talk about your audit and how that one particular document is part of that. What I did was I had one of the models, I said, look at the last 50 Slack messages that I've sent. Yeah, right. Because my style is not formal. I. If I got to be formal with you, you know, we'll do the ceremony. But that's not going to help either one of us.
- [30:00]
A
Right, Right.
- [30:01]
B
So my style is very casual, and I had it do that, and then it built out what it saw was my style. And now, like that, that's the document that I'm using. It's nailing it. So for those of you that haven't done this, we're going to talk about it in a second, but good idea. So with that, let's talk about you. You guys have kind of hit on a. A hot point with these, these audits that you're doing. So talk to me about, like, what you've discovered with that, who's interested in it, what do they do with the information, what makes a good audit? Just kind of like the whole, like, this is obviously something you've dived into. So I want to hear your expert perspective on this paradigm.
- [30:46]
A
Yeah. So I'll kind of tell you how it came about, first and foremost. So I was having lunch with two friends of mine, both business owners. This was probably five or six months ago at this point. One of the guys, he's a very successful commercial real estate broker. And because of that, he's very busy. Right. They're always, you know, he runs a fund. He's. They're listing properties. He's got a bunch of agents under him. But he really wants to use AI and he knows he needs to use it. He just doesn't know where to start. So he was like, he kind of made it. It all started really with an offhand comment that he made at this breakfast. He was like, man, Corey, I wish, I wish I could just pay. He's like, I'll pay you $1,000 to just come into my office and just follow me around for the day and tell me where I could be using AI. And. And he like. And I know he meant it. If I would have said right there, like, okay, I'll do it. Like, he would have cut the check, but like, it, it literally, I will never forget that because I'm like, if he feels that way, right, if he's having that issue, then chances are there's so many other business owners in his position who feel the same way. So I didn't take him up on that just because right off the bat I'm like, sure that'd be great, but like, that's not scalable. So, yeah, if anything, if I want to run with this idea from the start, I need to figure out a way that allows me to do this without being on site, you know, with a client every day. So that led to the first iteration, which funny enough, was asking clients like, hey, we're going to add you to our Loom account and we're going to have you hit record on Loom on your screen while you work for 90 minutes, two hours, three hours, whatever. And then we'll take that recording, feed it to Google Gemini, which is the only model that can, that can analyze video and have Gemini tell us, hey, based on how they're working, these are the opportunities for improvement. That's a nugget, right? Well, come to find out, there was a lot of friction there. People don't, you know, they're hesitant to record all their actions or, you know, they're self conscious that they're scrolling Reddit or whatever. So there was friction there. So we're like, okay, well what if, what if instead of having them record their screen with Loom, we just did like a 45 minute Zoom call with them and we can create a question bank designed to pull out their pain points. Right. Almost like a, you know, an interview. But the, the goal of the interview is not to prescribe or tell them what they can fix. It's literally just a diagnose, right? It's, where can you save time? Where is work piling up? What, what have you tried to automate in the past that failed? Right. So we've kind of developed this question bank that allows us to really efficiently pull out a bunch of opportunities for AI in a 45 minute call. So that's the current iteration of the model. We've actually built a voice agent that can do that, that piece. But I've done a lot of those interviews myself with business owners. So we conduct the interview whether it's me or the voice agent. We take the transcript and, you know, to simplify it, I always tell people, we just give it to Claude, have Claude, go research off the shelf AI tools that they can implement to, you know, fix some of those pain points. And then we prescribe those tools. But it is a little more detailed than that. It's not just a simple prompt. We built actually a series of skills that do a very, you know, thorough deep dive to find the exact tools that meet their needs. So we, you know, we find the tools and then we compile that into a report. And this is not just like a Google Doc. This is like, it looks like a 5 to $10,000 McKinsey deliverable that we built in a tool called Gamma. So we, you know, we put that into a report that's basically, hey, here are the three to seven biggest bottlenecks that we identified. Here are the three to seven AI tools that can fix those bottlenecks. And then we schedule a 30 minute review call to go over it with them. Right. And the purpose of that review call is twofold. One, we're sharing our screen and going through the report line by line to make it so that, so that they understand the opportunity. It's like, hey, you said you had pain point X. Well, here's tool, why, this is how much it costs. This is exactly how it can help your problem. Right. So that's part one is we're trying to actually help them with their problem. Part two is that's an upsell opportunity. Yeah, right. So that's where the client, you know, they're like, well, hey, this is great, but I'd love if you could just, you know, help me out. Can you just do this for me? Or a lot of times we'll uncover bottlenecks that it's not, it's not simple enough to where, hey, an off the shelf tool can fix this. It's like, we can help but it's going to be more of a, like we need to build an agent or we need to integrate XYZ AI tool here. It's not just a download. And so that gives us a lot of opportunity to upsell additional services. So that's really the model in a nutshell.
- [35:31]
B
So nobody wakes up thinking, oh, I need an audit today. Right, right.
- [35:35]
A
And we like to call it an assessment too because a lot of, you know, businesses, when they hear anybody, when they hear audit, it's like immediate turn off.
- [35:42]
B
Yeah.
- [35:43]
A
So we like to refer to it as like, hey, it's an AI, we actually call it an AI tools assessment.
- [35:48]
B
Interesting. Yeah, so we, we, I don't talk to the sales team, I guess to get the latest language, but I think we call ours an opportunity audit.
- [35:57]
A
Okay, that's definitely better than just an audit.
- [36:00]
B
Audit. Yeah. But what are people, the people that you're talking to, what are they wanting? What is the ideal state or the ideal outcome from this assessment?
- [36:13]
A
Yeah, so, and good question. So we have a guarantee around it too. Right. So the guarantee is, hey, if we can't identify at least 5 hours per week in time saving opportunity, based on what we find in the report where you implement AI, save at least five hours a week, then we will refund 100% of your money. Right. So again, the business owner goes into it with the kind of the, yeah, the mindset of like, okay, if this doesn't work, if they can't find five hours a week, I'm out 45 minutes. Yeah. Like that's, that's the only risk.
- [36:44]
B
It's worth it.
- [36:44]
A
Yeah. So basically either I buy back five hours a week or I lose 45 minutes.
- [36:49]
B
What is the realistic discovery? Like, what are people typically coming in at?
- [36:53]
A
So the average is about six hours per week and opportunity and the average tool cost to the business owner. So if we prescribe say three tools, the average tool cost total across those tools is going to be $40 a month. Because, dude, you'd be shocked how many of the tools we, we find where it's like you literally need a fathom note taker, it's free, or you need sanebox, it's $7 a month.
- [37:19]
B
Yeah.
- [37:19]
A
And you know, we're talking like one client that we, we did one for recently. We're doing further engagements with him as well. He owns a business brokerage. This guy's drowning in email. Two hours a day of email.
- [37:32]
B
Yeah.
- [37:32]
A
And the tool that we prescribed him, sanebox, sanebox.com not affiliated. $7 a month. All it does is it puts an AI layer over your inbox and it forces you to batch prioritize your inbox. You have your main inbox, which is everything important and urgent, and then you have what's called sane later, which is anything that's not urgent and not important.
- [37:54]
B
Still a match on that.
- [37:55]
A
He came to me, he came to me a week later. So we, we always send the report before we do the review call. So I think it was like a week between like when we sent this report and his availability for the review call. But he emailed us before the review call and he was like, guys, I went ahead and got sanebox. Like I read the report, I got sanebox. It's, it's crazy. And I saw him in person two weeks ago he said it's saving him about an hour, an hour and a half a day.
- [38:18]
B
Yeah.
- [38:19]
A
And so like. And it's a $7 a month tool. Yeah. So that's a kind of opportunity there. And another great example, actually some guys that, that you met, I met at the Fort Lauderdale event. Yeah. They own a, a, I guess a E Commerce pharmacy. Yeah, yeah, right, sure. And so because of that they, because they ship products in all, I think 46 states out of the 50 states. They're responsible for filing and remitting sales tax in 46 states. That's 46 different departments of revenue. That's 46 different sales tax returns either monthly or quarterly. Right. And so Adam, the co founder of the business whose time is probably worth about $1,000 an hour, is doing this manually every single month.
- [39:06]
B
Yeah.
- [39:06]
A
And one of the tools that we prescribed him is not an AI tool. It's literally just a SaaS called taxjar.com that 100% automates the filing and paying of sales tax that was worth the fee right across every state. Yeah. And we ended up doing, we did an assessment for him, for his co founder, Taylor, and then for their head of pharmacy, Nathan.
- [39:28]
B
So you typically do it at the individual level, not at the team level or anything like that, Correct?
- [39:34]
A
Yeah, we find, because I mean most of our clients are small business owners. They're probably actually some of the bigger business owners that we've worked with because they have multiple businesses. But yeah, it's at the individual level. So we'll do it. You know, we've done them with like realtors or like a one man show. Yeah. All the way up to like the business owner themselves.
- [39:53]
B
Yeah. So this might be for if, for the listener. If they're part of an organization, they might just say, I'm going to pay for that for my own damn self.
- [40:00]
A
Yeah. And I mean it would be worth it. Now I mean, that said, we have had requests to essentially create like a bigger assessment product of like hey, can you come and assess our department or you know, our whole business if it's a eight person business or 10 person business. And that's early on the roadmap. But we've, we've had a lot of demand with this just kind of single stakeholder product that we've just been doing a lot of those.
- [40:23]
B
I'll tell you what, as we talked about at lunch, the more people, the more complex.
- [40:27]
A
Yep.
- [40:27]
B
And if you can hit that sweet spot with just the like the individual professional who wants to know how to either run my business better or just perform as an executive better.
- [40:35]
A
Yep.
- [40:36]
B
Interesting. You crack the code, turn that into the soccer mom angle too.
- [40:39]
A
Yeah.
- [40:40]
B
Then you're in business. How do I run the house better?
- [40:41]
A
Yeah. Oh, yeah. I love that. The only issue with that is, like. And like, take my fiance, for example, right? Like, she's. She knows I'm obsessed with AI, but she kind of gets not weird about it, but, like, she's like, well, you better not ever have the bot, like, talking to me, you know, Like, I feel like there's going to be a lot of that pushback. She's like, I feel like there's going to be a lot of that pushback from people who aren't in tech. Or like, you know, take like a soccer mom, for example, maybe not even in the workforce. They. They like the idea of AI, but they don't want it to replace some of that. Like. Yeah. Which, you know, of course you'd use it tastefully, but. Yeah, I just thought that was. That just reminded me of that.
- [41:20]
B
So I like this concept. So the. And it's something that. That can be done in 45 minutes of bandwidth. Because the number one thing that when we're working with companies early on, pretty much you're like, well, how much time is this going to take? It's like you're just going to take what it takes. It doesn't take a lot. I don't want the listener to think that it takes a lot. But it's also not, you know, a fairy godmother, magic wand that somebody's just like, ding, it's done. It's a process for sure.
- [41:47]
A
Yeah. And so, again, for our. Our very specific assessment product, it's 45 minutes on the front end for the business owner and then 30 minutes on the back end for the review call. Right. And then in terms of my workload, 45 minutes for the assessment. You know, it takes us roughly 30 minutes to put together the report a little less, and then 30 minutes with a review call. Now, I did mention how we do have an AI agent named Annie who can handle the initial discovery call now. Right. And so we built her.
- [42:20]
B
What's the voice agent you're using?
- [42:22]
A
Retail AI.
- [42:24]
B
Okay.
- [42:24]
A
So it's. She's built on top of retail AI. But again, the idea with that is a lot of times business owners, you know, when they're talking to me, another human being, things tend to go off track. Right. We start bantering and we're talking about whatever, but we find that when they're talking To Annie, like, they know that. Obviously, they know they're talking to a bot. It's not a secret. So they're just, like, very direct and to the point. And so Annie can get a assessment done in 20 minutes. Yeah, that would take me 45. And she's. And, and honestly, not to toot her own horn, but she does a damn good job. Like, she's. She's designed in a way where she starts off broad, ask kind of more general questions, and then she'll kind of drill into very specific use cases. Like, if. If she finds that, like, okay, well, you know, sales is a big problem here. She's going to drill into sales and we're going to. We're going to uncover a lot of very specific sales use cases. When it comes to AI, I can
- [43:20]
B
see a lot of uses for that, man. I'm kind of a. On. On repeat, let's say there's something that's been on my mind a lot since I first got turned on to it about six weeks ago was. And I don't know if you paid attention to it, but Jack Dorsey fired a bunch of people from block. Right. 4,000 people they fired is a. When you read the terms of the layoff, it was. It was a kind gesture, for sure. They're getting well taken care of. Me, I just thought, oh, AI got them. Right. They came in, they did some stuff like this. They did an audit. They identified some opportunities for efficiencies and that sort of thing. Right. But that's not what happened. Are you familiar with the paper From Hierarchy to Intelligence that he came out with about a month after that announcement?
- [44:04]
A
So I heard people talking about it, but I'm not familiar. Maybe you can educate.
- [44:08]
B
Yeah. So basically he was saying that we were AI forward. We were evaluating our company and asking the question of what does the organization of tomorrow look like? And he said, our. Our analysis, it revealed that a big portion of the middle layer of our org chart, it looks like a pyramid. Right. Big portion of their role is they get information from somewhere, either externally or internally. They do something to it, they review it, they compile it, they analyze it, and then they ship it back out. Right. Or report on it or make a dashboard.
- [44:42]
A
All things that can be done by agents.
- [44:43]
B
Bingo. Right. And he said, that's what we did. It wasn't about that, that we made a better mousetrap with the existing organization. He said, we just flattened that organization.
- [44:53]
A
Yeah.
- [44:54]
B
So I know that when we were at lunch, you talked about how when you guys are evaluating these Processes that you're looking at the processes not just as is because, you know, odds are it can be a better, There's a, there's a, there's fewer moving parts required to get the deliverable from the company. Right. So you guys evaluate the process once it's captured for efficiency in general and then introduce the AI angle to it. Are you guys taking into account like preparing companies for, I don't know, their data today, still usable but ready for ingestion in the future by some intelligence or an agent?
- [45:32]
A
Yeah. So what you kind of hit on is something I've been just kind of like talking and tweeting about recently. Talking about the concept of building an internal knowledge base. Like a second brain. Yes, kind of the, yes, the popular wording for that recently. Now again, because I'm not like a data analyst, I have no technical background. I'm sure it is more complicated than the way I'm envisioning it in my brain. Maybe not, but maybe not. You're right. But really at its core it's like. And my business partner actually created a really good pyramid style graphic to depict kind of the levels of AI. And the foundational layer is data, right? It's your internal data, your internal knowledge base. Everything from emails to call transcripts to copies of proposals to SOPs. Like everything in the company is at that data layer. And I think the companies that create a really strong structured data layer of all their internal data, it's going to make it so much easier for them to be AI forward and for them to, you know, deploy AI agents and, and start AI projects. Because again, that data is the foundation of every AI everything. Right. So it's like if that, if that foundation isn't in place, it's going to be a lot more difficult and a lot slower for them to roll out AI versus if they do the work to, to build that foundation one time, everything builds upon it and they can, you know, they can ship AI projects so much faster. So what I've been saying is one that's going to be a multi, multibillion dollar industry of people offering what I call second brain as a service, going to, you know, companies like your clients and saying, hey, we will help, we will build your knowledge base or essentially create the data systems that will act as the foundation for all your AI deployments. So there'll be, that'll be a whole multibillion dollar industry. And then two companies are going to start to do that internally too. Like if you're a, you know, one to 10 person business, you probably don't have the funds to go out and hire somebody to build that for you, but it is something you should build yourself for sure. And I was at a workshop, I think it was last week. One of the guys presenting talked about, I guess, Arthur Anderson. The, the hospital did a. I think it was Arthur Anderson or no, M.D. anderson. I believe the hospital system did an AI pilot with IBM to something around piloting their Watson AI technology. They were like, it was something like there were $58 million in and hadn't treated a single patient with the initiative. And the reasoning come to find out is because the data layer was not intact at all. They jumped straight to. In the, you know, the. Using the image that my business partner created that they jumped straight to the top of the pyramid, which is, you know, the agent harness on top, basically the automation layer and the LLM, and they completely neglected the foundation, which was the data layer. So it's like if they would have built that foundation first.
- [48:36]
B
Interesting.
- [48:37]
A
And then, you know, jump to the automation and the, the layer, they would have had a much higher chance of success.
- [48:44]
B
Well, man, I hope it. Hope that companies can work with sloppy data because, I mean, that's a big lift, especially for like a small business or something like that. Yeah, they're just, you know, they're just hanging on.
- [48:55]
A
Right.
- [48:55]
B
So for them to prepare. So I can see that certainly being a role that. So we've got this, this situation. Right. Where you're, you're talking to people, I'm talking to people, and you and I probably have an expectation that, oh, everybody knows this or everybody knows that.
- [49:08]
A
Yep.
- [49:09]
B
And what I'm realizing is that even when I try to dumb it down, the feedback that I get from smart people is like, oh, that was over my head kind of. Yeah. Right. And, and I would say that's the bulk of business people, staff level, for sure. Even executive level. That's the bulk of it. Right. So that's, that's what the American workforce looks like today. But the technology continues to improve. Right. Like, like the whole last six months with Claude code and cowork and blah, blah, blah and all that stuff. Agents, Claudebot, Hermes, all that. That what, didn't exist or it wasn't as usable. Right. So we've got these people who, they're, they're still, they still need to learn like the basics of prompting, but the technology continues to pull away.
- [49:48]
A
Yep.
- [49:49]
B
Right. So, like, where does that meet up? Where does the, does it get to the point to where it's just so easy that this, that the workforce at large can say, oh, I can just use this tool. I don't need to know how to prompt, I don't need to know how to do anything like that. Or kind of. What do you see happening with, I don't know, the trend in adoption?
- [50:08]
A
Yeah, I think it'll get easier. Like, I think it'll just get even. Like we'll get to a point probably in the next two to four years where like people's grandmas are going to have an AI agent in their pocket or on their phone. Right. Like it's going to be so user friendly where like take Open Call right now, for example. Like setting up a simple cron job sometimes takes like an hour of debugging to, to do properly or, you know, connecting. Like there's, you run into issues trying to connect it to Gmail or whatever like that. That's just not going to be an issue. Like I think it's. The technology is going to improve to a point where it'll still, it'll all be natural language in the sense that I can just tell my AI agent, hey, call this restaurant and book a reservation for 8pm and it'll just do it and it'll just work. Whereas, you know, right now it's like, well, before I can do that, I've got a, I've got to get a phone number through Twilio and then connect it and then I've got to set up an 11 Labs voice profile and then I've got to make sure that, you know, I send it the proper URL for the restaurant. All that is going to go away. You're just going to be able to tell it to do things and it's just going to work.
- [51:16]
B
So I know that when we met in February, you were already Claude Botting. Right. But my experience with claudebot was huge pain in the ass. Yeah, right. Learned a lot. But like, I think maybe I've got one chief of staff that's up and running, but I don't even use it.
- [51:32]
A
Yep.
- [51:33]
B
But you said over the past few days you've been messing around with Hermes. Better.
- [51:38]
A
Better so far. So, yeah. So that's kind of like my background with claudebot. So that's really how like when I really started getting into the weeds, it was with claudebot. And it's because my business partner, he's very technically minded. He's a developer on day two. So this was like January 3rd or something.
- [51:55]
B
Yeah, because you presented on it. Yeah, yeah, yeah.
- [51:57]
A
So we had been using it at that point. So I presented on it that was probably January 27th or 28th. I had been using it for maybe three weeks at that point only because my business partner discovered it on day two. Like it was still completely unknown. But he messaged me one morning, he's like, dude, you've got like, we've got to start messing around with this. This stuff is crazy. And he built actually Annie, the voice agent that.
- [52:20]
B
Yeah, yeah.
- [52:20]
A
The first iteration of Annie was as an Open Claw agent. So I started experimenting with her and on day two of using her, I was like, I could never picture myself without this technology again. Like, it was that impactful to me that early on. And so we just got a three week head start on everyone else, which is the only reason I was able to present on it at that event. But I say all that to say, and the reason that was so impactful is, and just Open Claw in general to like our little bubble of the world is that was the first, I think, like real powerful demonstration of agentic AI and that coming mainstream. Right. Like that was the first time where you could tell an AI agent to do XYZ and it would go do it and it had, you know, no guardrails and it was smart and it would just like it would just get the job done. And yeah. And you just. That wasn't possible before that. So, you know, I was Open Claw everything for two or three months after that. Really. I've stopped using Open Claw almost entirely over the last probably two months. Using mainly like Claude Cowork and Perplexity Computer. Not as powerful, but better user interface and they just work better.
- [53:31]
B
Yeah.
- [53:32]
A
Now that Hermes is kind of the big thing on the scene and kind of replaced Open Claw in the. I guess the vibe, for lack of a better term. Right?
- [53:40]
B
Yeah.
- [53:40]
A
Everybody's kind of switching to Hermes because like people say it's like Hermes just works right out of the box. It just works. It's not as finicky as Open Claw. It's not, you know, it doesn't get stuck on these loops or like cron jobs don't fire. Like it literally just works. So one thing that I actually spent my whole morning this morning doing. Are you familiar with gbrain? Have you heard of G Brain?
- [54:03]
B
Gary Tan.
- [54:04]
A
Gary Tans, that's his kind of open. And so Gary Tan's the CEO of Y Combinator, which is like the.
- [54:10]
B
Knows a thing or two.
- [54:11]
A
Yeah. Most prestigious VC incubator in the world. But he created this open source project called GBrain. What GBrain is, is it's just my understanding is it's, it's essentially a knowledge layer for AI agents, but it's very robust. Like it can ingest, it can ingest thousands and thousands of context files and documents and notes and everything. And it has a retrieval system that it can easily retrieve that context as needed. Right. So for example, this morning I spent the whole morning building out my G brain, connecting it to my Hermes agent. So I've set up my G brain in a way where it pulls in so, you know, when I get on a zoom call or a Google Meet call, it uses Fathom. That's a note taker that I use. Yeah, right. Every, every Fathom call generates a call transcript. So I set up my G brain to where every call transcript that gets generated from my Fathom automatically pulls into G brain so that if I like, if me and you had a zoom call today and a year from now, I can ask my Hermes agent, hey, what did Chris and I talk about on, you know, whatever today's date is? It could easily, quickly, cheaply and accurately look up that transcript because it's in my gbrain and tell me exactly what we talked about. And then I can, anytime I create a new Hermes agent, I can connect it to that gbrain and all that context is there forever. So I connected my calendar to it, my Google Drive, my emails and my call transcripts. So it's just a big shared knowledge base for all my agents.
- [55:47]
B
So for the listener, this, this is a trend you're going to start hearing and seeing a lot about where people are just, they're not using chat GPT like a drive up window where they ask for something, they get their answer, they leave. They're going to be integrating it through the connectors and through the agents and through all this, the ingestion of all of these business artifacts like your email and your Slack and your Fathom transcripts and all that sort of thing so that you're going to be able to have this, I mean literally second brain, except something that on demand or will even surface things that you're not even thinking to ask, hey Chris, maybe you should think about this kind of thing, right? That's going to be a trend that I think that the those and you don't have to be like great at quote unquote, great at AI to set that up. You can get that going. But once you do, I think that those are the people that you're really going to notice like, oh my gosh, they're like, they've got a Superpower.
- [56:38]
A
Yeah. And so to your point, Right. It's not like the kind of the benefit of having that G brain or whatever second brain you end up building. It's not so much so that if I, you know, six months from now, I need to figure out what Chris and I talked about on May 29th. It's. It's more so. Exactly like you said. It's like the AI is going to be able to draw parallels.
- [56:59]
B
Yes.
- [57:00]
A
Between inputs that you would never be able to draw. Right. It's like what it might say, like, hey, Corey, you know, based on your last 50 call transcripts.
- [57:09]
B
Yeah.
- [57:09]
A
Every prospect you talk to, every one of them has this one objection. And, like, sometimes you overcome it, sometimes you don't. But in all your sales materials, you should be optimizing for this objection.
- [57:20]
B
Yeah.
- [57:20]
A
And like, that one insight might make you an extra 50, 100, $500,000 and
- [57:27]
B
save you a ton of time.
- [57:28]
A
Or save you a ton of time. So it's those insights. And when I tell people, because a lot of times people, when I've been kind of telling about this concept of the second brain, they have trouble wrapping their heads around, like, what's the ROI here? I was like, well, I can't tell you because I don't know what you have under the hood.
- [57:45]
B
Yeah.
- [57:45]
A
But what if every bit of information your business has ever had, if you were able to ask AI one question, it's look at all my information.
- [57:55]
B
Yeah.
- [57:56]
A
And give me 10 things that will either save me more time or make me more money based on every bit of context you have about my business. And then I guarantee you, you will be baffled. Yeah.
- [58:08]
B
Put a price on that. Yeah.
- [58:09]
A
You'll be baffled by the information that you get.
- [58:12]
B
So that's the exercise, and that's a great point. If somebody says, what's the roi? Like, I mean, I'm locked in. Right. If you're not that engaged with your career or anything, maybe it's not going to have that big of a difference. But if you're locked in and your business is important, important to you having another you, like, dude, if I was my own co pilot, forget about it.
- [58:31]
A
Yeah, for sure. I mean, and that's what you're building, right? And yeah, you can have a bunch of them. Like, that's what I like about Hermes.
- [58:36]
B
You can have a bunch of them
- [58:37]
A
is you can have. With Hermes, it's called profiles. Right. So I can have. I have right now my main Hermes agent, which is kind of my, you know, operations slash executive assistant, but on My to do list this week actually on Wednesday, I'm creating two additional profiles. One is going to be a marketing slash content profile, and one is going to be my cfo, my AI cfo. So each of those, that's essentially three individual Hermes agents, they're all going to have access to that G brain that I mentioned. So that's like the shared context layer. And then in addition, each of those Hermes agents, it's going to have their own individual private G brain. So the content agent is going to have a content gain, the finance agent is going to have a finance G brain, but they're all going to have access to that shared one, right. That has the call transcripts and the email and everything. So it's like imagine it's literally like me building out my org chart of AI agents that one get better over time because that's what Hermes is fundamentally designed to do. And then two, have shared context across everything that you're doing, you know, from a general level. And also have private contacts. So that way you're not muddying the waters with, you know, your finance agent doesn't necessarily need your content pillars. Right. And your content agent doesn't, doesn't necessarily need to know about your may piano. But for the context that does matter, they all have access to that.
- [60:03]
B
Yeah. So man, we've covered a lot of ground and it's been awesome because I don't get a chance to, to geek out with somebody that I know. Most of the people that are on the podcast, we've had a pre interview, I'm interested in their subject matter, of course, but we don't really know each other and we broke bread a couple times, all that jazz. So if you're listening to this and you're like, whoa, I want you to think about something. If I ask you, hey, where are you? On a scale of 1 to 5 when it comes to using AI, I get this all the time. I have enthusiastic executives tell me, oh for like, great, man, what are you doing with it? Right? You use it all the time. Okay, great. Well tell me what it is. I use it to write emails and summarize documents. Right. And that's great, that's an awesome use. And the fact that you're using it regularly for those things only opens the door for you to go, oh, I could also, like, it's part of the process for sure. But what we've talked about here, and this is like, like Corey is literally like on the edge of things trying to break these tools. This is where a 4 or a 5 is. They are thinking about these boring documents that are very powerful, but it's the design md, it's the skill md, it's the context document, it's the company brand voice document. It's all of those things that aren't really, like, that's not AI Right. That's you documenting the soul of your business. Really, like all these different elements. And that's great. You get one contacts document. It's better than nothing. But then you start getting into, you know, these levels of it. Now I've got an agent. Well, great. Now my agent has specialized agents that that agent can call onto, like that. And you're not doing this for a major enterprise. You're doing this to run your business, to allow you to be somebody whose influence is much greater because the ability for you to touch the full spectrum has gotten so much easier. Or it's being done on your behalf because you invested the time to think about, well, what do I want this agent to do? And what skills should it have and how should it be structured so that I'm not spending fifteen hundred dollars a month on this one agent and its tokens and all those sorts of things. So, folks that are listening, we've had a wonderful conversation, but, like, there are layers to this stuff, and we. We could have gone all. Yeah, we were toning it down a little bit. So fantastic conversation. And then now you're off to go enjoy the fruits of your labor.
- [62:29]
A
Yep, yep.
- [62:30]
B
Very cool. Cory's on his way to get his. His new Tesla.
- [62:33]
A
Yep.
- [62:34]
B
So nice. Well, thank you for taking some time out of the day here. And I want people to plug into what you're doing. So if you don't mind, maybe take a minute and just kind of share, like, where you're. Because I know how much time and energy you're putting into creating this content.
- [62:45]
A
Yep.
- [62:46]
B
I read the content. I don't waste my time. So for all of you that are listening to this, I would encourage you to pay attention to all the places Corey's about to tell us to go catch up with them.
- [62:54]
A
Yeah. So, I mean, the main place I would send people is my. So you're listening to this podcast. You obviously enjoy podcasts. Go listen to, slash, subscribe, Build with AI. That's the name of my podcast.
- [63:06]
B
We'll have a link to it in the show notes.
- [63:08]
A
Yeah. And so, you know, check it out on audio platforms, of course, but it's best experienced on YouTube because I do my podcast in screen share style. So every one of my episodes we're sitting down, we are sharing screen and we're building. It's called Build with AI. Like we're building things live or we're testing tools or we're, you know, tweaking workflows and, and you can get the, the video version of the podcast over on my personal YouTube channel, which is just my name at Corey Gannam. But yeah, that's where I'd send people. And then, you know, if folks are interested in an assessment or they want to work with me directly, they can send an email to Corey C O R y@returnmytime.com so I mainly follow you on X.
- [63:50]
B
Where else are you putting out content?
- [63:52]
A
X LinkedIn and then YouTube slash podcast.
- [63:56]
B
Yeah, okay.
- [63:56]
A
X. I mean X is my. X is definitely my most active in the sense of like when I come up with an idea, I fire it off on X and if it does well, we turn it into a YouTube video. And YouTube is where we go like super deep on a very specific. Nice build or agent or topic.
- [64:12]
B
I didn't know that. So I'm going to have to catch those as well.
- [64:14]
A
Yeah, very cool. Yep.
- [64:16]
B
Corey, thank you so much for taking the time, man and everybody, thank you for being a listener of Using AI at Work. I always strive to bring people again who are doing the thing, not just selling a product or whatever so that we can get into these types of conversations and you can see what, what it's really like to be an AI fluent, you know, ass kicking executive or business owner out there. So if you enjoyed the episode, please think about sending this along to somebody else, you know, who was on the journey and we'd love to have them as listeners. So thanks everybody. We'll see you on next week's episode. Thanks for tuning in to Using AI at Work. Don't forget to subscribe for more conversations about how to use AI at work and a special thank you to our sponsor, Chief AI Officer for Empowering Businesses with AI Education and Training. Visit their website for a free AI Readiness Assessment and AI Strategy Guide to help you get started using AI at Work. That's www.chiefai officer.com. follow us on Twitter at the handle Using AI at Work and visit www.usingaiatwork.com for free resources to help you harness AI in your role.