A (3:55)

And what did you do before you got into AI?

B (3:58)

Yeah, I was doing enterprise sales consulting. So I grew a group from 0 to 30 million and ARR in five years was only four salespeople. And so a lot of people wanted me to teach them how to do that.

A (4:09)

So. So you started using ChatGPT pretty much right when it came out or you had early access to it. Right. And what are you doing now?

B (4:16)

Yeah, it's evolved a lot, man, because I've been helping companies implement AI ranging from like startups all the way to $20 billion companies. And so within that, I mean, I've shifted more into the orchestration aspect where I'm using agents across different connected systems to do the work for me instead of just reactively continually asking for it. So that's like my core focus right now and I think it's going to be table stakes in 2027.

A (4:45)

I love it. It's fascinating. You come from this background of sales and now you're getting into like some of this really exciting stuff in the frontier of AI when it comes to real time data because we're kind of here to talk about a little bit. What's some of the biggest misconceptions that you see that a lot of people have right now when it comes to AI and data?

B (5:04)

Yeah, I think one of them is that like it's only for coders, right? So it's one of those things where folks think that it has to be reactive and they don't understand that these coding models like Codex 5.3 from OpenAI or Cloud Code are insanely powerful. And so you could actually use those for a lot of non coding tasks with real time data. I know there's a lot of marketers that listen here and entrepreneurs and that's a lot of what I use it for personally. And I'm just amazed. And I can give tons of examples as we kind of go through it, because like I said, it's one of those things that just, it's like another version of it like blowing my mind, you know what I mean? Like from the original ChatGPT instance, it's

A (5:46)

fascinating because so much of the AI world right now is, and has been for the last three plus years since ChatGPT came out, is highly technical. And the moment you hear some of these phrases like Codex and code immediately, like people in marketing and sales and entrepreneurship that aren't don't have a technical background, their eyes kind of glaze over a rollback, you know. And it's interesting that you say that you do not need to be technical in order to be able to benefit from this stuff. So we're going to get into some of this today. But when data is properly integrated into AI, what can that unlock help people understand? Because many people right now, like me, for example, we're uploading PDFs, you know what I mean, to projects in Claude or maybe in Gemini. Right. And it gets outdated. And what's the unlock, you know, that comes to so many of us that maybe we don't even realize when you have access to real time data and you couple it with AI.

B (6:46)

Yeah, great question, madam. And I can give examples across a couple models. So for example HubSpot, there's a connector in ChatGPT as well as a connector in Claude. They're also both actually working on apps and are releasing app connectors as well, which I think is the next level of MCP model context protocol connections, which is really just a fancy word of like, you know, not having to go through a really complex connection of one software tool to another.

A (7:14)

Before we get into the examples, just tell me what the upside is because we're going to get into the examples for sure. But I want everybody to understand the unlock that happens when you have real time data. Can you conceptually introduce that just a little bit?

B (7:25)

Well, here's what I'd say. I think there's two benefits. It lets you operate at the speed of thought. Oh, that's one. Okay, right. Then the second is it reveals patterns and opportunities that humans can't see most of the time unless they spend hours and hours and hours and hours and hours and tons of decades of experience doing it. So that's two. And then the other is, you know, it saves time. Right? So like hours of time as well. So I would say those are kind of like the three unlocks.

A (7:58)

Explain to me the speed of thought thing just because I want to hear more about what that means.

B (8:01)

Yeah. So I have a CEO operating system. And so with the speed of thought, like I have it connected to HubSpot, I have it connected to my note taker, I have it connected to Notion, I have it connected to my Google Drive. And so I could say, for example, and I did this today and I said, and this wasn't even anything that like I pre built. And this is why I was so excited about it is I'm like, I was like, I gotta create content for social media, right? That's a little meta. But it was for, for LinkedIn X. And I was like, all right, well you know, a lot of what I do often is sharing things that I actually do with clients in terms of my content. Cause I want it to be experiential. So I go, give me the top five posts and write them out with a hook, make them super tight for LinkedIn and I want that created. Anyway, they look through like all my transcripts over the last week. It looked at interactions with clients, it looked at other and it pulled out nuggets and it created those posts that were really good in terms of what it chose. And then the other thing that was really good about it is I didn't even have to sanitize it. Like it took out the customer name, right. So it didn't reveal anything as well. So it was like really freaking impressive. And that wasn't something that I premeditated. I just said, boom, I want to create five posts from. It went out and did that without having to like pull up all the transcripts, look through the transcripts, try and feed it into another model. I just gave it a command and it executed.

A (9:25)

I want to kind of deconstruct what I'm hearing you say Because I think this will be helpful for people. The way that so many of us have operated in our careers, especially those of us that have had previous careers that are 10, 20, 30 years of experience, that we put in the various knowledge domain expertise that we have. Typically when we do something, we have to go out and we have to remember all sorts of information, right? We have to collect our best practices, we might have to find templates, we might have to search databases. And this is a really, really manual process. Right. And what I'm hearing you say is now literally, because AI will have access to all of this relevant data, it can do it instantly. It's almost as if you have a superpower. Would you agree with that?

B (10:09)

Yeah, I mean, that's the way I feel, man. I was super jazzed was. This was yesterday. I spent a lot of time and it was kind of funny, man, because like, I've been looking at forecasts and pipeline reports my whole life, right? So not my whole life. That's an exaggeration. My whole professional life, if you will. And it's funny. Like I, I was in this habit of, of going through manually looking at progression, identifying next best action. I'm like, why am I doing this? Like, I built out some things that they're not completely built, but like a sales operating system. And so I basically just said, hey, give me a next best actions report for all the deals I have, stack, rank them, prioritize them. And then it did that and created a report. And then the other thing I said is, hey, these deals are off stages. Change them in the CRM and it changed them without me having to manually change, you know, seven different accounts in different stages. So like, that's something that I think is so freaking powerful. Another one, man. And this is like, add on to that. This is more simpler so that what I'm talking about is using Claude code and connecting it to a lot of different solutions. A much simpler approach because I want to meet people where they're at is for example, like Gemini. Or let's say I was talking about ChatGPT with HubSpot. Like, let's say you could, you could basically run a deep research report with HubSpot and ChatGPT and say, okay, identify the ICP pattern of my contacts over the last 90 days, what types of companies they are, size all those details, and give me a list of the hottest ones that I need to follow up with now, and boom, it'll do that. So, like, that's another example and that's a much simpler connected Solution with just a normal chat model.

A (11:46)

Love it. Okay. And we're going to get into kind of how to do all this stuff for those of you that might not be tracking with everything Ryan's talking about. Okay, so let's say that we want to connect real time data into our AI systems. Before we begin, is there anything we need to think about or consider, whether it be privacy issues or whether it be like mindset shifts or any of that kind of stuff? Like just to kind of set the groundwork here?

B (12:09)

Yeah, I think on the security side, don't do it on a free account. That's like one of the dumbest things you could do is have all your data. I would get pay the $20 with the $30 where they have commercial data protection guarantee, which means the models aren't training on your data. And then also go to the settings and make sure that you're not sharing your data to train the models. That's number one. And that's like the ground rules, I would say, for that. In terms of a mindset shift, it's really interesting because I've been doing a lot of work on this lately over the last few weeks, just because I even noticed with myself there's times where I'll still operate the way I've always been doing it without. But I'm like, hey, I got like basically a worker that's available to me 24 hours a day and can work, you know, a hundred times faster than a human if I asked it. Like, don't you think if you have that kind of power, you need to kind of reevaluate how you work on a day to day basis? Right. And so, so that's one of the questions that I started asking myself. And then obviously the value I have is like trying to deliver this for clients and help them along in this journey and re engineer role structures, org structures, all those areas. But I think there's a big difference between the classic mindset and an abundance mindset where you have that unlimited resource for pennies on the dollar right now.

A (13:24)

I love that. So how did you have to shift your own mindset to think differently? Because I'm sure some of my audience is struggling with some of this right now. You know what I mean? Was there any, any things that you did that you or you could share with others to help them kind of think different?

B (13:40)

Yeah, yeah. This is a real simple one and this is useful for anybody and any model that they're using. But what I would do is like you could literally ask the AI model is just like, hey, I'm trying to like deconstruct or reverse engineer, like how I could work better with agents versus the classic way that I've worked. Can you give me a side by side on how I need to change my thinking to be able to really work hand in hand with, with agents? Right. And if you're talking to an agent, you could say you. That's what I do sometimes, you know, like, hey, how can I work better with you? And like I want to. So I'll do that with like Claude code or the other agents, but that's a simple way. And what I'll tell you, man, is like, it's been super impressive with what it comes back with. And it's like, that's a really freaking good question. Like, here's how we operate differently, here's how you do it, you know, and the side by side I think is really critical because it shows the differential.

A (14:32)

Yeah. This is something I want to double down on as well, because I recently wrote about this on LinkedIn. I am going to be teaching at my conference, Social Media Marketing World opening keynote, as I always am every year. That's the benefit of having your own event. But I wanted AI to tell me, like, what are the things that I intuitively do that maybe are not things that your average user does? You know, So I could develop a framework and it looked across everything I've ever done over the last 60 days. Right. And it gave me some really insightful things that I might not have seen in myself. Right. And that's also kind of another way of thinking about what you said, which is, hey, Claude or ChatGPT or Gemini, how do I need to be thinking differently about interacting with you so that I can maximize whatever your objective is? And it's really going to help you, especially the models like, like Chat, GPT and Claude, who have the ability to kind of like query all of your chats. And that is kind of one of those shifts. Right. Because we treat AI a lot like we treat humans. Right. And, and these things are so much faster in some regards. And that's a mindset shift. I love it. Okay, you've got this thing you mentioned called the Sales operating system. I would love you to like explain a little bit more about what you did just so people can understand a little bit more of what kind of stuff this could unlock. If you don't mind sharing a little bit more about that.

B (15:53)

Yeah, sure. I'll, I'll break it into like just very easy, digestible components so What I did is I kind of looked at my business in the key areas and I started with like, all right, what do I actually want as an outcome versus what do I need to build? And I'll start with there's three kind of components that I started working on. And I'm deep into CEO operating system. For me in terms of looking across all elements of myself, my belief structure, how I'm aligned with my goals, with my time and resources. I have a sales operating system which is really like how to continuously improve the conversion rates, deal sizes, pretty much anything I need, and then a product operating system. And so that's how I can continue to improve the deliverable, which I execute on, which is really like AI transformation for clients to have tons of transcripts, resources, details like that. So that's, that's what I decided I wanted to build. And then what I did is I went in Claude code and there's some technical barriers up to originally starting and getting to know this. I remember like kind of one of the things is like I made, I drew a line in the sand. Cause I'm like, I see this being transformational. And so I watched a YouTube video for two hours and only got like 15 minutes of the way through it. Because I would like watch the video, try it, it wouldn't work. Then I would go to Claude and be like screenshot and be like, hey, this isn't working. How do I do this? It's like, oh, you're not doing this. And going back and forth until I figured it out. And so once I got that set up, it was relatively easy. And some people have an easy time setting it up, but there's iterations to it. So you do that directly in cloud code or in cursor. Cursor is kind of like a model aggregator or you could do that within Codex. Okay, so that's step one.

A (17:41)

Real quick. Before you get to too deep, explain what the objective of one of these operating systems is. Just so people can conceptually understand why you're building one of these in the first place. And then we can drill down even deeper into the technical side of it.

B (17:53)

Sales, for example. It's like I could look at every single deal with next best actions that need to happen to make those happen. Because I don't deal with like small opportunities, they're larger, what we call high ticket deals. And so it'll look at the trends of like what people are saying, the questions, details like that. So that would be on the sales side. That's just like one of Many examples. The other thing, if we're looking at like the CEO operating system which I've been using the most lately is I'll do like daily check ins with it. So I have it set up a system set up where I'll do daily, weekly, monthly check ins. The cool thing is it's got frameworks embedded in there from like top leadership experts ranging from personal development like Tony Robbins to Marshall Goldsmith to other folks. So it has like maybe seven different frameworks embedded in there. And then what happens from there is it also spits out my goals on a monthly, quarterly, annual timeframe. But because things are moving so fast, I have that updated, usually monthly, in terms of how that's operating and what's kind of happening. So that's one thing. The thing that I've been most surprised with Michael on this is there's things and patterns where it'll call me out on like, I'll also do a brain dump. Like just sometimes I'll have a brain dump where I'm like, ah, this seems like this should be right. This seems like this isn't like, I think there's a big opportunity here and I'll just like unstructured dump it in there. And it's like, hey, like this is the number one thing you need to focus on. Quit jumping over here, right? So it's like good for my adult add, if you will, in terms of jumping all over the place. And it also gives you the good side too. It's like, hey, like you have some superpowers in this area. You really need to lead into these. And your customers are literally saying, give me more of this. Like, you should do that. So it's like almost like an angel and a devil on your shoulder that has access to more information than anybody else in your life or in your world, because it's looking across every single conversation. You have your brain dumps of what's going on in your head, what you don't say, and then it synthesizes that, looks at patterns, and then helps you try and optimize it. So that's what blows my mind.

A (20:00)

Love it. Okay, we're going to get to cloud code, but before we do, I want to talk a little bit about the connectors, because you've already kind of mentioned that a little bit at a fundamental level. Let's just start with how do we connect our data sources, whether they be sheets or customer relationship management CRM databases. How do we connect these to like chat, GPT or Claude and start unlocking this real time benefit because I know this is available and I know you know how to do it. So maybe you can explain a little bit there before we actually get into the more complicated parts of this.

B (20:31)

Yeah, super easy. So you just go to settings and then you look at. It'll either say connectors or apps. I think it says connectors in Claude as of today. It says it just started changed it. They just changed it. And chatgpt to connectors. You could also do the same thing in Gemini. It's called connectors in there and it'll allow you to basically connect to Google Drive and other areas. And then what I would say too is for copilot there are connectors. The good thing about it is if you have a paid license, it automatically connects to all your work. So OneDrive, like Teams Outlook, that's something that's pretty solid from that too.

A (21:09)

Yeah, Gemini does that too, by the way, for folks that have Google accounts as well. Every year the gap widens. The marketers who understand AI are pulling ahead, creating better content faster, automating strategic work and delivering results their competitors Simply can't match. 2026 could be your best year ever because you finally embraced AI. Now you're listening to this podcast, but why not join us for two days of systematic mastery at AI Business World, April 29th and 30th in Anaheim, California? Here's what Melanie Miller told us. Quote the AI teaching was mind blowing. I'm so far ahead of so many people that claim to do what I do because Michael Stelzner only brings the best to teach, unquote. Here's what sets apart this from random tutorials and scattered learning, which I know we all use. You're going to get complete workflows, not isolated tips, strategic frameworks that connect everything, implementation plans you can execute immediately, and 20 sessions dedicated to build comprehensive AI mastery. These are the best of the best that I have recruited. By the end of the event, you'll have all the AI skills that took others years of random implementation to figure out literally at your fingertips. The gap is widening. Which side are you going to be on? Learn more at aibusinessworld Live. Grab your tickets today at aibusinessworld Live. So once you connect that stuff, is it legitimately real time or is it like, do you have to like, help people understand how that works? Exactly? Does it just query it when it needs to? How does that data kind of between your stuff internally, whether it be Excel or sheets or HubSpot or whatever? How real time is that actually with these connectors?

B (22:58)

Yeah, it's 100% real time from what I've seen. Right. So it's not like lagging or anything like that. It's more so just real time. And then I'd say the way to tap into it is like you have to toggle that connector on for it to be focused because like the AI models effectively look at billions of pieces of data. So you got to give it the focus on like let's say HubSpot and then ask a question specific for HubSpot. Right. Tools like Claude though are a little more advanced than that where they'll let look across multiple tools sometimes in the same work stream. ChatGPT, I haven't quite seen do that as much, but yeah, that's real simple way.

A (23:35)

Funny enough, I have not yet connected because we have data privacy policies here at the company and I'm sure a lot of people depending on your business do have data privacy policies where you're not allowed to necessarily do this. But some of you are smaller businesses might be willing to take this kind of position where I'm going to go ahead and try it anyways. But once you toggle on this, for lack of better words, connector and you and I both use Claude pretty regularly, do you have to call it or tag it or something like that in order for it to bring it in or can you create it in like a project where it automatically knows to look there or what's, what's. Any tips or insights on that?

B (24:13)

Yeah, I don't think it has the capability where you could set it up specifically in a project, but you could call it and then dump it into a project. Right.

A (24:22)

Okay.

B (24:22)

I don't think you can set it as a default in a project, but

A (24:25)

in a thread you generally have to tag it or something like that to say like look into HubSpot or look into this particular document set in order to be able to reference that real time data. Is that generally how that works?

B (24:36)

Yeah, exactly. Like you got to hit the plus symbol in ChatGPT and Claude I think you could hit a slash and it'll give you options to pull up different files and copilot and then same plus symbol for Gemini. And you could also cool thing with Gemini that's newer too is you can actually look at a notebook that you created and use that as a connector to a notebook. So that's pretty cool as well.

A (25:00)

Does it matter what model you're using? Because every one of these. Well, especially Claude and Gemini have different models. ChatGPT I think has got a router where it kind of decides which model to use. But does it matter? Have you found it matters whether you're using, like, Claude Opus 4.6 versus Sonnet 4.5, or do you just like to default with whatever the system comes up with? Have you found that there's a variance there in the results?

B (25:27)

I usually always use the most advanced thinking model. So like, I would use 4.6 opus. That just came out, right? 4.6 sonnet just came out for Claude or which is literally today. Literally today came out and it's supposed to be just as good, but way faster and way cheaper. So, yeah, and then like, for like ChatGPT, I would use 5.2 thinking. Right. I wouldn't necessarily use Pro in ChatGPT because it. It's almost like doing deep research where it might take 20 minutes to give you an answer. So I reserve that only for really big important things as well.

A (25:59)

Okay, so Claude code. Talk to me a little bit about that. Explain what it is for people, because let's just assume people have heard about it, but they have no idea what it can do. So talk to us a little bit about that.

B (26:10)

Basically, Claude Code is a coding agent, is what it was originally designed for. However, it has the capabilities of one of the best overall general agents I've ever seen, so you could use it for anything. And so what I've kind of discovered with that, and like let's say codecs is, and I mentioned earlier about operating at the speed of thought, a lot of times you could just tell it what to build and it'll start to build that. Now, granted, there's caveats to it, right? It's gotta be simple. There's a lot of techniques that you could use to do that, but effectively think of it as a simple system to build agents or to interact across multiple resources and data in real time to get you answers and outcomes and actually do work for you.

A (27:00)

Is it kind of like building your own little app for your own little purposes? I mean, a lot of people have used the word agent. It can mean a lot of things to a lot of people. As we know. There's a bazillion, like, it's a word that's so kind of overworked, overused, if you will. Do you find maybe developing an app, is that effectively what you're doing? You're creating a specialized application that's just for you doing a specific job, or that happens to have the ability to kind of operate autonomously, or is there like, like help. Help me wrap my head around That a little bit, because I know my team has used cloud code, but I have not personally used cloud code.

B (27:32)

So there's levels to this game, right? There. There are ways where, I mean, some people have. Have ran cloud code autonomously for 30 hours straight, right? Without human intervention. That's really advanced, and there's a lot of things you would need to do to. To make that happen. However, like, the beautiful thing about it is I could say, okay, like, here's an example. And when I say agents in cloud code, I mean like, like agents that go do work for you, not just an assistant, right? So like, an example of that would be like, all right, and maybe this is a good example, maybe not, but, oh, here, here's a good one. We'll use one for content, since there's a lot of marketers that listen to that. So I could say, okay, use three agents simultaneously or at the same time, have one, research the top 10 trending topics over the last seven days, have the second one, review those and match with what I talk about, and then have the third one create content in that specific format as it relates to the trending news content, right? Like, that's an example, and it'll do that. It'll create the whole thing. Or you could do the same thing, but then create, like, 10 or 30 scripts to YouTube posts from the same thing, right? So a lot of, like, really involved examples that it could execute on. Like that.

A (28:50)

Typically, when I think about something as complicated as what you just said, it's typically a workflow. And based on all the things I've heard from my other guests, you're using a tool like make.com or zapier or N8N to accomplish what you're talking about here. But in this case, it sounds like you're giving it just a really complex instruction set and it just figures it out. Is it that simple or are you oversimplifying it?

B (29:14)

It does a good job of figuring it out. However, there's another layer to it. So what I would say is it's called skills. And I think, just like I mentioned, MCP protocol or mild context protocol, MCP servers, which are basically just a fancy word for connectors. The other one that's becoming a dominant standard is skills. And so skills are basically like a workflow of instructions that the agent looks at and knows how to do a job, right? Because I think literally, and I'm seeing this already today with myself, is today's jobs are going to be tomorrow's tasks, right? Like, it's going to change that much that fast. And so I think that's something that, that's really strong and it's going to be cross, like OpenAI adopted it recently and they're going to start including it. From what I'm reading, Claude was the creator of it or anthropic, and I imagine a lot of other people will grab onto it as well. But it's basically like those instructions. So it is that easy if you have like the skill created for the agent to do, because it knows what to do every time and it's not like trying to pick a bunch of different tools to do it. It knows exactly what to do and how to do it well.

A (30:24)

And I should tell everyone this, that there are the abilities to create a skill and then download a skill and share a skill and it's just a little zip file and you can actually tell Claude, like, let's say you've created some projects, which I'm a bit huge fan of cloud projects because they're like, to just allow me to have different kind of responsibilities and roles. I've got one that's a copywriter for my conference, which is exceptionally good, and I asked Claude to help me create a skill out of it and it did. And then it basically gave me the little file and I shared that skill with someone else on my team who uploaded that skill and activated that skill. And now it's kind of like the Matrix. It's like, you know, now he's got this super skilled writer that he can bring into any conversation whenever he wants to, which I think is really cool. And I wouldn't be surprised if there's like a public repository out there where you can find these skills and download these skills so you don't have to create them yourselves. I know that there's some out of the box skills that are included inside of Claude, but getting back to the Claude code side of things, how is Claude code interface and user experience different than like chat GPT or Claude in general, which is a very simple chat interface? Is there anything people need to understand when they're actually loading up Claude code that maybe is going to look foreign or different to them than what they're normally used to using?

B (31:44)

Yeah. And this is where it stares a lot of people away. It looks more like a coding terminal, like what's called an ide. Right. And so all it is is it's like very text looking not fancy, not fancy visually. Right. Like it's. It looks very kind of like DOS versiony. Right. In terms of communicating with it. So, so that's what I would say, like is the big thing I have. The thing is like I have mine connected to Obsidian or you could connect it to Notion. And so Obsidian is like, it basically puts it into a much more visually stimulating readout than a normal, like super text, like very grainy looking. And then same thing with Notion. Notion is another example where you could push it to.

A (32:28)

So Obsidian md. Is that it? I'm looking it up while we're talking. So tell us a little bit about what that is because that does look a little bit like a note taking app almost. Huh? Is that what it is?

B (32:39)

Yeah, it's like a vault. Like a vault all to have context with note taking that it could look across all different segments and areas. It jives really well with the coding agents because they talk in markdown files, it translates markdown files in text. So it's, it's really strong like that, I would say. And then you could operate in the Obsidian or you could operate in CLAUDE code in terms of like updating it.

A (33:02)

So when you connect it to Obsidian, does that mean you are interacting with Obsidian and CLAUDE code is kind of in the background? Is that kind of what I'm hearing?

B (33:09)

You say I interact with both of them, actually. So like I would say when I want to manually just update it, I'll work in Obsidian and then if I want like full commands, then I'll operate that in CLAUDE code. But I just saw that Obsidian released like a way to put CLAUDE code in Obsidian, so you could actually have the terminal in Obsidian, which is really interesting. I just, it's literally only a couple days old and I haven't had a chance to use it yet.

A (33:34)

So let's say everybody. And by the way, if you have the Claude app, you can download it on your computer and then literally there's an option called code. But I would imagine you could do this through a web browser too. Or do you need the app in order for this to work?

B (33:47)

So here's the big difference and I think Anthropic is going to streamline this. But when I use CLAUDE code, it's connected to my local files as well, which is my technically cloud files within OneDrive. However, if you use the web version of CLAUDE code, that little link that you said, or the area that says code and you click on that, the files aren't going to be stored locally, they're going to be stored. You'd have to create what's called a GitHub account. Right. So within that it's basically just A think of it as files and drawers, but most people use them to store code. Right. So you have to connect it to a git to do that. It's free to be able to do that, but that, you know, might scare people away from doing it because it's an account a lot of people don't have unless you're a developer.

A (34:33)

The GitHub account, is that required if you are connecting to your Google Drive or your Microsoft database or whatever, Is it still required, the GitHub account, or can you do a connector to existing tools you're already using?

B (34:49)

Yeah, it's for the Claude code web version. It is required.

A (34:54)

I see.

B (34:55)

You have to have that.

A (34:56)

Got it.

B (34:56)

So like I said, I think if you're on the desktop app, they're starting to make adjustments. I think they'll have it so you connect to your local file folders too. So I think that's one of the things that's critical with it because effectively, like the other thing I forgot to mention is with cloud code, it has a memory file that you could set so that every time it starts, it knows what to look at with that memory file. And then you could update it over time. So it gets better and better.

A (35:21)

Okay, so we have scared the heck out of a sub segment of my audience that is not technical. But I heard you say you just watched a YouTube video to figure this out. I mean like once you get over the hump, is it pretty straightforward or kind of help people understand what they're in for?

B (35:36)

Yeah, I mean like this, I'm saying like the model companies are making it easier and easier to use. Codex is the same way. If you connect it to a GitHub account, you know, that could be the easiest way. It really operates very similar in terms of communicating just like a normal chatbot would, just like ChatGPT or just like Claude or just like Gemini or Copilot. It's like you ask it questions. The only difference I would say with like the coding agents is once you start to get them set up with skills and other areas, you don't need to tell them how to do every single thing they already know. So you could have them run multi step workflows without creating an N8N or make setup. And it's in my opinion way easier, faster and it doesn't break as much as like a, you know, a workflow chain like that.

A (36:21)

Okay, so here's what we know so far at this point. We know that you can take Chat, GPT and Claude and we can connect to Third party data sources and just interact inside that familiar interface, that chatbot that is built into these tools. We also know that you can do way more sophisticated things when you download OpenAI's codecs or Claude's cursor.

B (36:48)

Cursor, Cowork.

A (36:50)

Cowork, right. Oh, and what's, what's cursor? Yeah, explain that.

B (36:54)

We didn't really cover co work either. Do you want me to hit cowork first?

A (36:57)

But. Okay, no, up at this point we're talking about Claude code. Sorry, I'm getting all my. We got blog code and we got chat GPT Codex. Right. Which is OpenAI. Right. We got these two. If I got that right down, right. And then cursor. Yeah, Layer in cursor a little bit and layer in Cowork here a little bit. Because I think that also might be another option people might find more, less stressful.

B (37:16)

Let me do this two ways, man, because I don't want to confuse the hell out of people. So step one is chat CBT or code or agents with connectors. Right. Not agents, copilot with connectors. Right. Step two is a tool like Cowork by Claude. This is newer, it just came out on Windows, I think like February 10th of 2026 and it's been out on Mac for about a month. The interesting thing about Cowork is basically it's between like using a chatbot and using a coding agent. And basically what it does is you could ask it to do like end to end tasks and then it'll do that for you right from your desktop and it operates and it looks and feels a lot more like normal cloud or ChatGPT. Then you have the coding agents which we talked about with cloud code or Codex. And then cursor is over in the coding agent side is like almost think of like a model aggregator that could like I could work across, you know, different models depending on strengths and weaknesses of them specifically for the coding agents. And that's what cursor is. It's just like a, a centralized view.

A (38:21)

So why use something like Claude code over just using Claude Help people understand what the Claude code can do that Claude itself might not be able to do.

B (38:34)

It could do longer run jobs and automations across multiple systems. Whereas CLAUDE can just basically connect to one at a time. Right. That's, I would say, the biggest, the big difference. So long running jobs and at the same time you could spin on multiple agents at the same time. So multiple. And then it could connect to multiple data sources.

A (38:57)

Now a lot of People that aren't super active might not realize that you're going to hit thresholds with some of these tools, like Claude in particular with the code side of it. Do you have to set up a special account in order to be able to not hit? Because obviously if you're doing a whole bunch of things simultaneously, I would imagine you're going to run out of tokens. And with the cloud code you have to set up a special account in order for that all to work.

B (39:18)

You do have to have a paid account. I believe it's at least the 20 or $30 if you run up against it. Cloud's a little more strict with their limits than other folks. I have a $100 a month account, use it a ton. The same thing is like cursor, you know, it's like 30 or $60 a month and you can get a ton of usage out of that. So I think you're just starting like you just start with the lowest paid account and most of them should enable you to do that and leverage it.

A (39:44)

Okay, is there any automation here on any of these accounts where you just kind of like don't even have to worry about it? Like every week or every day it's going to send you reports and stuff like that. Is that possible with any of these solutions?

B (39:56)

Yeah, it's different and there's like a window task scheduler you could do in other areas for the automations. That's some of the stuff I'm digging in deeper right now with that. It's kind of interesting though because like, like if you look at like, like let's say those N8N or make workflows with all those different nodes and connecting the system and then producing an output, it could do that really good. This scheduling setting up or trigger based is a little bit different and so that's what I would say is kind of like the biggest change. I think it's way easier to construct a solution, put it together and then I think there's a little more nuance with like the automation triggering.

A (40:34)

Okay. If people wanted to begin experimenting with real time data in any kind of capacity, where do you suggest they start? Like what's a very simple application, for lack of better words and I don't mean like app, I mean like use case, you know what I mean, that anybody could go experiment with right now?

B (40:52)

Which tool are you talking in? Are you talking about like.

A (40:54)

I mean you just pick the tool. I mean we can talk about the one you use all the time.

B (40:57)

Yeah, here's what I would say just to start simple, just start with like your CRM or like your OneDrive or somewhere where you just have a lot of data. And then one of the best questions you could ask, and you could do this with deep research in Gemini too is just like, hey, look across my data and tell me trends and patterns. I should be paying attention to that I'm not. Or identify five unique opportunities that I don't see right now that are game changing. Right. Like you could just ask it a question like that and not every answer is going to be perfect or come back amazing. But usually like if you do that, like it comes back with like two or three just bangers that are awesome. And I was just like, wow, I can't believe I didn't think of that. Right. Such a blind spot. Right. So those are examples I would say that it, that it's, it's really, really good at. And that's simple. You could do it on any tool at the most basic level.

A (41:51)

And then if we want to go up to the next level would be like the next obvious thing we would do.

B (41:55)

Yeah, like go to cowork then, and then just try it, connect it to a couple folders. There's a section called plugins which are like, and this is what caused, I believe, a trillion dollar sell off in the SaaS market a week and a half ago is Anthropic released these plugins which are essentially like workflows. You know, there was ones that were like looking through legal contracts, NDAs, like autocorrecting them, redlining them, all those details. And so like the vertical SaaS providers were losing their mind. And so that caused, literally, I think it's almost like a trillion dollar sell off in SaaS because people are like, oh shoot, these companies like Anthropic and OpenAI are now coming for us, you know, at the application layer.

A (42:38)

Ryan Staley, thank you first of all for answering my litany of questions. And folks, I know it sounds technical, but as someone who has watched people on my team do some of these kind of things, I just think there's a little bit of a learning curve, obviously with some of these kind of things, but once you start using them, I would imagine it's going to unlock a lot of benefits, especially if there is any kind of database. Right. Of any data that's constantly being updated or changed inside your business. This could be a really big unlock. Ryan, if people want to connect with you on the socials or potentially work with you, where should they go?

B (43:14)

Yeah, I would say LinkedIn YouTube and X are the primary ones that I'm on. LinkedIn is the one in X that I pretty much publish on all the time with. Like, I've had the majority of my friends focused on LinkedIn, so if you want to connect me, just reach out. Say you heard me on the show so I know who you are. Because I get a lot of, a lot of rando connection requests. Right. But we'd love to hear from you and you know, if you're interested in working with me, that's a super easy way to just put it in the connection requests, like, heard you on the show. Looking for help with my company on this because like the results I'm seeing, I'm seeing clients like double output that they couldn't get past before, like double pipeline in like 60 days. Like it's insane. So excited about it. But thanks for having me on, Michael. I really appreciate it, man.

A (43:59)

Ryan Staley, thank you so much for sharing your insights with us today.

B (44:03)

No problem, man. Thanks for having me on.

A (44:04)

I know we covered a lot of technical stuff. We took all the notes for you over@socialmediaexaminer.com a98 and be sure to follow us on your favorite podcasting app. And if you've been a longtime listener, we would love a review. Do let your friends know about this show and check out my other show, the Social Media Marketing Podcast. This brings us to the end of the AI Explored podcast. I'm your host. Michael Stelzner will be back with you next week. I hope you make the best out of your day and may AI help you become more successful. The AI Explored podcast is a production

B (44:36)

of Social Media Examiner.

A (44:41)

What if you could get year round AI training? That's exactly what's waiting for you with our AI Business Society. To learn more, visit socialmediaexaminer.com AI.