← Mondays with Matt
Mondays with Matt
AI for Entrepreneurs: Practical Tools for Decision Making & Problem Solving
00:48:22
AI is revolutionizing the way entrepreneurs make decisions, solve complex problems, and streamline business operations. Whether you're looking to enhance productivity, gain deep insights, or automate critical processes, AI can be your most powerful tool. In this episode, we explore the best AI-driven strategies and tools that every entrepreneur should leverage to stay ahead in an increasingly competitive landscape.
Loading summary
Transcript132 lines
- [00:02]
Matt Reynolds
You're listening to the Build you'd Business podcast powered by Turnkey Coach, where we help business owners find freedom over fear. I'm Matt Reynolds and I'm his brother Chris Reynolds.
- [00:12]
Chris Reynolds
Join us as we help build your business and move from fear to freedom. You're listening to the Build you'd business podcast with your co hosts, Matt Reynolds and Chris Reynolds. I'm Matt, he's Chris. What's up man? How's it going?
- [00:35]
Matt Reynolds
Going well.
- [00:36]
Chris Reynolds
It's been a crazy week. I lost our last grandpa on Saturday morning. Lost our dad, grandpa was our dad's dad three years ago Sunday. And so it's just been kind of crazy. And we've been at the Arnold and done lots and lots of follow up calls and all kinds of work and I know you're crazy busy too, but we're going to class today on AI and this is going to be good for me because last episode, by the way, this is going to be a three part series, almost like a little course, a mini course on AI. Last week we talked a lot about just the large language model, the AI kind of text based applications which I do a ton of. And so if you haven't listened to that, that's a great place to start. And part two, today we're going to talk about how then you use those things as well as other resources in the AI community and the AI industry to help with decision support as a business owner. And so with that, you are the AI expert, I will turn it over to you.
- [01:38]
Matt Reynolds
Yeah, so I think one of the first things you know that everybody has to do with these AI tools is just start playing with them. We talked about this on the last podcast. Pick a few. If I was going to pick right now for you, I would probably pick chat GPT, the pro plan for $20. And then I would also probably start with Claude, which is a tool by Anthropic and that one's I think also $20. And both of those tools do slightly different things. They compete with each other. So in some ways they, they cover, you know, similar types of, types of things, but they do them differently. And I'll today we'll talk a little bit about how I use them and, and how you can use them to do, to do better stuff. So the first thing I'm kind of curious about, Matt, I don't have the $20 chat GPT account. I have the $200 chatGPT account because I'm using a bunch of these pro tools that are, that are in there. One of the things I'm curious about is you're using the $20 version, correct? Okay, can you tell me, do you have ChatGPT4.5 with Deep Research, do you have 4O?
- [02:52]
Chris Reynolds
And that was actually one of my questions was right now I think my ChatGPT will only do text based application. And so I've seen the demos from, from ChatGPT, from OpenAI, about, you know, being able to actually watch video, help solve math problems. You can, you know, write on an iPad, voice recording. It will actually go voice to text. As far as I know, I don't have access to that. And so now I might, but I don't think I do. And I've played around with it a lot even in the last couple weeks, not just for this podcast, but because of the things we're doing in the business. And so that's kind of where I'm at is I think where the $20 version gets you is a really good text based application. But video, music, sound, you know, voice to text, all of those sort of things I don't think comes with the $20 version.
- [03:42]
Matt Reynolds
So the, the multimodal is what you'll hear people talk about. And that multimodal just means it's the ability to, to do, you know, more with images, more with sound, more with, with all those things. The $200 a month version, which obviously is, is quite, quite expensive, I think for most people to think about that like, whoa, that's, that's a, that's a big business expense. But, but actually when you weigh that against what you could do with, may not be for me, for me and our business, that is a small amount to pay for the value that I get out of it. So it makes a ton of sense for us. It's obviously a business expense and it's a business expense that adds at least three times that value, maybe four.
- [04:25]
Chris Reynolds
So quick question for you. Are you the only one that has it on your team or do you have a few more that also have that higher price version?
- [04:33]
Matt Reynolds
Yeah, actually at this moment I'm the only one that has it on our team. And my business partner, I've told him get it whenever you think you'd use it. And he just hasn't had the need to use it. But part of the reason for that is that when we get into the more advanced tooling, I'm going to describe how to use that and that takes away some of the value that you would see with the $200 a month version of ChatGPT. But, but I think maybe the best way to describe this right now is just. I'll just tell you a few examples of the way that I use it. So the app for Chat GPT on your. On your phone with this subscription allows you to click a button and just start talking to it. Now you actually have that in the $20 version too.
- [05:19]
Chris Reynolds
For sure. You do, right on the phone.
- [05:20]
Matt Reynolds
So you. In the $20 version, you can talk to it and it will talk back to you in whatever voice you want. And that is by itself a very interesting one that we're going to talk about for just a second. But I'm going to quickly just step one step forward and tell you when you go to the more advanced version, you have the ability to use a thing called deep research for ChatGPT 4.5, which is the newest model of that.
- [05:47]
Chris Reynolds
So I actually have access, I'm looking right now on my phone because I'm actually primarily using it on my computer because I'm on my computer all the time.
- [05:54]
Matt Reynolds
Yeah.
- [05:54]
Chris Reynolds
So I do have the speech to text on. And I can also just toggle onto the 4.5 research preview version, which is probably a dumbed down version of the one that you pay for. Then there's no.
- [06:04]
Matt Reynolds
It's not. Actually the 4.5 preview is where everybody is right now with Research Preview. So that's good. You probably only have so many that you're allowed to do in a month, which should be completely fine. I think I can do an infinite number of them. And the big model that I had that I hadn't seen anybody else get yet, if you weren't on Pro is, is the 01 Pro mode, which is a really long thinking and somewhat slow model, but it is exceptionally good at solving certain kinds of programming problems.
- [06:31]
Chris Reynolds
Got it.
- [06:31]
Matt Reynolds
They've come out with some new ones as well that are also good. But let's go to a second. Since you do have the interactive mode, I want you to imagine that you are. And this is the most practical application of this that I can think of. I was at an Airbnb and there was a coffee maker that didn't work. I woke up in the morning telling.
- [06:53]
Chris Reynolds
This story last week and you were like, you just punched it in and it told you how to fix it.
- [06:56]
Matt Reynolds
Well, more importantly, let's talk about what it actually did. So what I didn't tell you, I don't think, was that there is a video mode that's on there that you can. You're talking to it from your phone and, and showing it live video from your phone.
- [07:11]
Chris Reynolds
From your phone.
- [07:11]
Matt Reynolds
So it's.
- [07:12]
Chris Reynolds
Turn on your camera. Okay.
- [07:13]
Matt Reynolds
So I have my camera on, I have my phone on. And I'm like, hey, this thing isn't working the way that I expect it to work. And I kind of go around it so it can see the model and see all the various things of the, of the coffee makers. Like, oh, yeah. Okay. Well, the problem here is you probably have air bubbles that are inside of the tank. This was like a Keurig style. It was a, it was a dual mode. So you could actually run drip coffee through it too. Yep. And it said, we probably had air bubbles from doing this thing. Pull the tank off, empty it, put it back in with, you know, load it up or whatever, and it'll probably start working. And I did that and it worked. And I mean, it's talking to me in real time.
- [07:48]
Chris Reynolds
Sure.
- [07:49]
Matt Reynolds
And if I said, can you see the model number of this? It's like, no. Could you scan in a little closer? Uh huh. Here you go. Like, okay, I got the model number. This is the thing. Right. So your days of going to Google and looking up the, the documentation for your washing machine or for, you know, figuring out. I think I talked about the tire issue where you're trying to figure out like, is this thing safe? Like, here's what's going on. You kind of have this little elf in your pocket that you can be like, hey, here's some video of what's going on. Let's just talk through this.
- [08:21]
Chris Reynolds
Yep.
- [08:22]
Matt Reynolds
And that's a mini version of why it's super useful to founders. So let's talk a little bit about how you can use this same concept to help you make better business decisions.
- [08:35]
Chris Reynolds
Yeah, I was going to say. So this is a great microcosm, your coffee maker story.
- [08:39]
Matt Reynolds
Yeah.
- [08:40]
Chris Reynolds
That is decision support.
- [08:41]
Matt Reynolds
Yes, it is.
- [08:42]
Chris Reynolds
So you had several options. You could like, I can try to figure this out myself. I can go on Google, I can do research, I can go on YouTube, which was probably the primary way that our generation is now doing research to figure out how to do a thing.
- [08:54]
Matt Reynolds
That's right.
- [08:54]
Chris Reynolds
Or you could have gone and just bought a new coffee maker and sent to Airbnb owners like, hey, we just went ahead and bought a coffee maker because the coffee maker wasn't working.
- [09:02]
Matt Reynolds
Right.
- [09:02]
Chris Reynolds
But Instead you utilized ChatGPT and the video and, and audio versions, which is that part of deep research or. No?
- [09:12]
Matt Reynolds
No, deep research is a different thing. We'll talk about it in a minute.
- [09:15]
Chris Reynolds
So you use that then for it to Help you make that decision. Support like here's probably what's wrong and here's probably what you should do.
- [09:22]
Matt Reynolds
Yep.
- [09:23]
Chris Reynolds
I think you said within just a couple minutes the thing's fixed.
- [09:25]
Matt Reynolds
Yep.
- [09:26]
Chris Reynolds
You didn't have to spend in far less time it would have taken for you to research that on Google or YouTube or wherever. You were able to just talk to ChatGPT, show it the video and it told you the fix or the prop, the high probability fix.
- [09:42]
Matt Reynolds
Right.
- [09:42]
Chris Reynolds
And then it fixed it. And my guess is if it didn't fix it, you would have said that didn't fix it. What's the next highest probability issue that is going on? And it would say, okay, let's try this.
- [09:53]
Matt Reynolds
Yeah.
- [09:54]
Chris Reynolds
So I think creates a smooth bridge from text based only applications of AI to now utilizing the voice. And it's actually interesting. I've certainly used it on my phone, but like I said, I don't use it a lot. There are a lot more toggles and dropdown on my phone than what I tend to see on my laptop. So now my guess is it's there on the laptop. But because you don't walk around with a laptop showing video, typically the screen on your laptop is pointing forward.
- [10:23]
Matt Reynolds
I've never used it that way on the laptop. Honestly, I think the app's a little different on the phone. I think, I think it's pretty mobile.
- [10:29]
Chris Reynolds
First and I can actually toggle on deep research too. So now what will be interesting for us to talk through between this week and next week is the difference between the 20 version I have, which is now turned to 4.5 and I have deep research on my phone compared to what you have access to. Okay, so now we know it'll help you with decision support. How do you then transition that from fixing a coffee maker to. To fixing whatever in the business?
- [10:56]
Matt Reynolds
Okay, let's start with the simplest example of this, and that is you fairly easily envision a world where you're dealing with some kind of problem mentally in your head. This is the world of founders. Right. Like you are constantly thinking about how do I solve this problem or how do I communicate this thing or how do I do whatever. Right. So there's a lot of value in being able to work through these kinds of problems. Now you specifically Matt, are pretty interesting this way because you vote. I know, I know. Known you for, you know, your. My whole life anyway. Not your whole life, but you process a lot of things by talking about them. Right.
- [11:36]
Chris Reynolds
So you have an external processor for sure.
- [11:39]
Matt Reynolds
So a lot of people are this way. And I actually think that even if you aren't that way, I still think there's a lot of value in doing it anyway, just to see, you know, sort of where the value is. So what you would do is whenever you're trying to solve a problem, envision that you would pull out your phone, flip it into I'm going to talk to ChatGPT mode and you just start talking to it. So you say, yeah, like it's a.
- [12:02]
Chris Reynolds
Business mentor, like here, like here's a guy who's got all this experience and which it doesn't because it's a computer, but it does have all of the entirety of the world's knowledge and you can just start asking it questions, hey, I'm struggling with this. And you can start to prompt it to speak. Does it literally speak back to you or talk back to you?
- [12:21]
Matt Reynolds
Oh, it speaks. You hear whatever voice. And yeah, you got like eight voices you can pick from or something. And so you know, what you would do is you would start by simply talking it out. So you're going to talk about it, it's going to give you responses. Now what's cool about this is this all is going to feed into our next conversation with once you've had this back and forth with Chat GPT number one, you probably have some pretty interesting insights. So I've done this a lot in the car. When I'm in the car by myself and I'm going to pick up a kid from baseball practice, for me, that's a 15 minute drive. And in a 15 minute drive it's a great time for me to talk to Chat GPT about whatever thing it is that I'm currently thinking about and have it come back at me with, well, have you thought about this or have you thought about that? Or I say like what are five ways to solve this kind of problem? Like, or you know, what am I not thinking about? Or you know, whatever it is. And what you get is this back and forth conversation that has occurred where all the context of your head has made it into the ChatGPT chat and all of its responses have come back and all of that information is available inside of ChatGPT to be exported as text.
- [13:40]
Chris Reynolds
Okay.
- [13:41]
Matt Reynolds
Which is.
- [13:41]
Chris Reynolds
So it still transcribes the entire conversation. Yeah. And then you can and can be spit back out in transcription mode.
- [13:48]
Matt Reynolds
That's correct. And so that's useful for many, many upcoming reasons that we'll talk about.
- [13:54]
Chris Reynolds
Turn it into a stop or an actual plan that you communicate that you that you email or put on a.
- [13:59]
Matt Reynolds
Google Doc or whatever.
- [13:59]
Chris Reynolds
The thing is.
- [14:00]
Matt Reynolds
Yeah. Even if you didn't think about that at all and the only thing that you were thinking about was how can I get good answers for what I would call right now like really good general knowledge. I think I said this last time but I think about these LLMs like they are the smartest, the smartest friend that I have, the most well read friend that I have that maybe doesn't have a ton of emotional capability. So it's like zero. You were almost or something. Right. That kind of guy.
- [14:35]
Chris Reynolds
And so actually my original podcast partner was exactly that, that human.
- [14:38]
Matt Reynolds
There you go.
- [14:39]
Chris Reynolds
Shout out to Scott Hambrick knows everything. He was the original AI but was able to make clear decisions with no emotion attached whatsoever. So yeah, worked great. I even told you before the podcast there, there are things that I want to talk about some today and I want you to give me some assignments to do off the air because also I think it's sort of high pressure to do a thing for the very first time like on the air. Plus if you're not watching this on YouTube, you should be watching on YouTube then it would not make for a great radio as they say, great podcast.
- [15:12]
Matt Reynolds
But.
- [15:12]
Chris Reynolds
And so I sent you several things like SOPs that we're working with but let me actually talk a little bit very briefly and we'll stay at the 30,000 foot view on some financial things. So we're recording this. It's early to mid March which means the end of the quarter is coming up. Yep, we are as I've told most people, so we're a large online coaching business. We have a flagship B2C product or service that has been around for nine years. We have our SaaS product that we now license out our, our online coaching software to other coaches. And then really utilizing that SaaS software we get government contracts. And the government contracts have done a wonderful job of helping pay for the build out of the SaaS. We have millions of dollars, but in the single millions of dollars, not over 10, but quite a bit of proposals that are out there right now, one of which we were supposed to hear back from last week, but the government's slow and so in the next few days we're going to hear back about it. And so I'm talking to my CFO yesterday and we're talking through the decisions of if we don't get this contract, couple million dollar contract. Let's round numbers. What are the changes we need to make in the business to make sure that the financials stay strong. And essentially there's only two primary things you can do. You can get more revenue, you can grow more revenue via increasing prices, more clients, ltv, whatever, or you can cut expenses. So I could have a conversation, 30,000 foot view with ChatGPT on my phone, 12 talking through that and say, if we don't get this next contract, what are my options? And again, we have some advantage because there is so much information out there about our company and the amount of content we've built. So it actually knows so much about us already. I don't have to give it massive amounts of context. What are my options that maybe I haven't thought about? About how to increase revenue in the B2C side of the business and in the B2B side of the private sector business.
- [17:14]
Matt Reynolds
Yep.
- [17:15]
Chris Reynolds
And can workshop that out with ChatGPT. Likewise, I can do the same thing and can say, here are some of our major expenses, here's the percentage of that based on our top line revenue. Where are places that you think that I should cut and why? Yeah, and it will actually walk through those things. And then I could start to workshop that and then continue to even dive deeper until I get to a point where I'm like, I feel pretty comfortable about if this, then this. It's really, it's. You know, my kids take, I mean yours probably do too. My kids do classical education. They take a course called Logic literally every single year. So all 12, 13 years of school. And that's all logic is, is if this, then this. Okay, so if we get the military contract, how does that change the financials? Where should we be spending our money? Should we be spending it on dev? Should we spend it on customer acquisition costs and why, if we don't get it, where are the places to cut? Where we don't lose service? Where hopefully we wouldn't have to make layoffs, like whatever those things are. You're telling me that that is a conversation that I can have right now, today with ChatGPT over audio directly on my phone?
- [18:22]
Matt Reynolds
Yes, absolutely. And it's super enlightening. Whatever it is. You can ask it to go, you know, further down a particular path or whatever the key there is that, you know, because it isn't emotionally charged about any of the topics for, for it, it's sort of just, this is all information that I have in my information bank. I'm going to, I'm gonna provide you feedback based on the results. You can have it do other things too that are sort of interesting. I've started playing with this idea of asking it to answer as though it is a particular type of personality. Because one of the things that's interesting about these models is that they've been trained on the entirety of human knowledge. They at times, they are as smart as the smartest of us and as dumb as the dumbest of us. And so one way that I've started using it is I've started saying, look, you've got all of the written work and every spoken word of Warren Buffett. I want you to answer. How would Warren Buffett answer this question?
- [19:20]
Chris Reynolds
What I'm hearing you say is that it is extremely high iq. It's not infinite iq, but it is a much higher IQ than any human who's ever lived.
- [19:30]
Matt Reynolds
Yeah.
- [19:30]
Chris Reynolds
But extremely low eq.
- [19:33]
Matt Reynolds
Right.
- [19:33]
Chris Reynolds
However, if you know how to prompt it, what you're saying is it can bridge EQ to iq. So give me an answer as if you were Charlie Munger, as if you were Warren Buffett, as if you were this. So then all of a sudden it takes everything it knows about that person and it says, well, here's what Charlie Munger would probably say back to you.
- [19:50]
Matt Reynolds
Yep, you can take a book that you've recently read where you think the author knows all the things about the whatever, and now it will be better based on how much that person has been covered in the news, how much they've published.
- [20:06]
Chris Reynolds
It's not going to give you copyrighted, because I've tried this, so it's not going to give you straight copyrighted information that isn't already out there on the Internet. So you can look up Undoing Urgency. My book, and we've put out a ton of stuff on it. I've asked it to give me a breakdown of the. I'm amazed at the amount of stuff it can. It can do. But I've done the same thing with other authors that I like and love, and there's just less out there because they've been very protective of the ip, Right. Of the copyright of what they've actually written.
- [20:35]
Matt Reynolds
So, yeah, a friend of mine was talking about how there are people that have started doing this in a way where they think about it like, and you can set things up to make this a little easier, where you set up your own little brain trust board of different people, different personalities that you particularly respect. How would Elon Musk answer this question? How would Peter Thiel answer this question? You sort of like, you can build out a group of people that when you have a super hard problem and you want to talk about it and you want dissenting opinions to start sort of pushing in different directions. You can build out those personalities where you're having that conversation with that type of person. So obviously you're not really having it with that person. So all of this is be smart, take this with a grain of salt, right? Like don't.
- [21:24]
Chris Reynolds
It's a tool.
- [21:24]
Matt Reynolds
In some cases, it's a really wonderful tool.
- [21:27]
Chris Reynolds
It's not another human being, it's never going to be another human being. But it will continue to get better over time. So figuring out how to use it as a tool now, and I think this is really the main goal of this three part series, is that it is moving very fast and it's not very hard to use right now. But just like people who just never adopted the computer or never adopted Internet, you know, our grandparents generation, they were sort of left behind. And so it's going to be much, much more difficult six months from now, a year from now, and certainly five or six years from now if you haven't already started using these things. And so this is the point. Get used to the text based application version of this first move to the voice and video, then start moving to the conversational piece and the decision making like the support for decision processes within your business or your life or your health or whatever. And there is a systematic progression of how to use this and get better and better at it so that you stay maybe not right on the cutting edge where you would be or where somebody who's like a, you know, runs a development team, an engineering team would be. But you're at a point where you're far more advanced than the, than the normal founder or the normal business owner that's out there. And you're able to use a tool now that is, it's, it's like a cheat code. You got a cheat code tool that most founders aren't using because they're still too scared of it. And they're like, it's just going to be another, you know, Web 3.0, flash in the pan, you know, whatever.
- [22:56]
Matt Reynolds
It's not, not.
- [22:58]
Chris Reynolds
This one stays, right?
- [22:59]
Matt Reynolds
This one stays. And so, and so, yeah, so let's, let's talk about, maybe we'll get into a little bit more detail around how to use these tools where they support different kinds of decision making or different kinds of research and the various pieces to these puzzles. So when I use ChatGPT in particular, when I need information that is hyper relevant right now, like super current, extremely current information, I use the ChatGPT4O model with web search because it is basically constantly mapping the Internet and can give me, if I say I need a relevant answer to a thing that came out yesterday, I can get a really good answer. It's going to go very quickly. It doesn't do a huge amount of research, but it's going to very quickly go get whatever it either already has in its memory bank or it's going to do a couple of simple web searches and then come back like good, good web searches, better than you can do and then come back with a summarized version of that information. So that's a super useful trick. I haven't, I don't think I've been to Google honestly in. It's been at least a month since I've gone to. I don't go to Google ever. Like, the only thing that I do to get information that is current and relevant is use these different, these different models.
- [24:23]
Chris Reynolds
Of course, as a side note, not to just crap on Google, they're putting billions and billions and billions of dollars into this as well because they have the greatest search engine history, the like, they probably have the greatest database ever. So if they can figure out how to do the same thing, that's their leg up.
- [24:42]
Matt Reynolds
That's right. They've got, you know, and I anticipate that it's going to keep getting better. One of the things about those models is, is called Gemini. And Gemini has the largest context window that exists right now that we've seen. So what that means is the amount of text you can give it where it can take all that as input and understand it in one shot is the highest of any of the models. And it's because not only do they have this enormous amount of information, this huge database of knowledge, they also have a tremendous amount of computing power. And so because they have that amount of computing power, they weren't waiting on a bunch of GPUs coming from Nvidia. They were already there. They could already start doing some things that others couldn't do. Gemini has a 1 million token, which is about four characters, four letters is the way to think about that. That's about right. It can do a million in one model, it can do 2 million in another model. That's a tremendous amount of context. I mean that's getting up there where you could put most of the summarized stuff in your entire business in that and it would understand it. The problem is right now their models are not, I've not noticed their models being very good. So when you actually ask A question with all that context, do you get a really good answer? The answer to that is not right now, at least in our field. So not. It's not great yet in some of the tech stuff that, that we currently use it in. But it is getting better. It's getting better really fast. And I suspect that it will continue, maybe it might even continue at a faster pace than some of the others. So that's good news for everybody. I mean, that's just having some competition there across all the different companies that are doing this is just hugely valuable. Everything gets better.
- [26:34]
Chris Reynolds
Good for us.
- [26:35]
Matt Reynolds
Yeah, absolutely. So let's talk a little bit about. We talked about web search. We talked about, like, how to, how to do it. There's a bunch of different models. They're confusing over time. It's already sort of come out from OpenAI that they plan to create a unified interface. So you don't have to make, you don't have to make guesses about that. You just talk to it and it'll choose the model for you. Basically, it'll have a model that chooses. Models are great.
- [26:56]
Chris Reynolds
So as opposed to saying, I need 4.0 with recent web search versus 4.5 deep research. Right. It will automatically know based on the question which one will provide the best answer.
- [27:07]
Matt Reynolds
That's right. So that's kind of coming right now. You still have to know a little bit more about it. But the next one is, I think, maybe even to me, I think it's been the most valuable of all the pieces of this tooling. And that is the new Deep Research tool is insanely powerful, just insanely powerful. And I'll explain what it does so that you can think about all the places where the context makes sense.
- [27:35]
Chris Reynolds
This is ChatGPT. Again, this is ChatGPT 4.5 deep research.
- [27:40]
Matt Reynolds
Yeah, the new 4.5 deep research. Now, there's a lot of other deep research models that are coming out. So just be aware, you know, specifically.
- [27:47]
Chris Reynolds
The OpenAI version of ChatGPT 4.5 deep.
- [27:50]
Matt Reynolds
Research, this is the one that I'm talking about. It's the one I use most frequently. I, I've got, you know, the subscription and it was easy to play with. So what it does is it takes in a prompt, and it's not going to be a huge prompt, but it can take in quite a bit of information in your prompt where what you're doing is you're asking it to go do research for you. Okay, so imagine that you had a super, super smart top graduate from an Ivy League university that worked at your company and they could go. They weren't particularly good at giving sales presentations or closing a deal on the phone. But if you needed the best information possible, if you needed a report from one of the big consulting firms that was going to go research the Internet for you and figure out all the things, they could go do that because, you know, they did an internship there and they know that stuff like, that's who you have. Okay. So what it does is it takes a very long time. Like, its whole thing is that it doesn't come back sometimes for 20 minutes. It might be 20 or 30 minutes. So you leave. Like, you can leave, you can close your computer, you can close your laptop, and it's doing all this behind the scenes, but you can also watch it. I think it's kind of fun to watch it because you can watch what research it finds, but you can ask it questions. For example, you could say. You could say, my business is a consulting business. And we do, I don't know, we do tax consulting for, you know, gyms that are in, you know, for. For the fitness industry or whatever. Let's say. Let's say that that's a business. I'm sure it is. Somewhere. I want you to go out and find all of the other companies that do this on the Internet, and I want you to compile them in order by the largest estimated revenue of these guys. I want you to also describe for me what the ideal expense ratio is for a company like that and what are reasonably good targets for gross profit margins and net profit margins. If we ever wanted this company to. If we ever wanted to sell this company, given the specifics of the industry that we're in and the, you know, the dynamics.
- [30:03]
Chris Reynolds
Right, all this stuff, all that stuff.
- [30:05]
Matt Reynolds
All right, so you say all that and then. And then you go away and you're.
- [30:08]
Chris Reynolds
Still doing that on your phone.
- [30:10]
Matt Reynolds
You can do it on your phone. I do it on my phone all the time. Because if I'm in the middle, and.
- [30:13]
Chris Reynolds
Then you don't have to leave it up. It's running in the background and you're doing all your normal stuff. Does it ping you then when it gets an answer? And so then it's like answers here. And then you open it up and okay, it does.
- [30:24]
Matt Reynolds
And then. And then what you get is not just the answer, but you also get the citations. So it provides you with what I would. What I would suspect is probably one of the best kinds of reports you could ever have as the foundation for other things. You're going to do next. Right. So you could even envision that you went and asked it this kind of question to get all this incredibly relevant current information. You can ask it for this, by the way, I do this all the time. I say the. I want you to heavily weight the sources for what is most recent because I need. This particular thing needs to be super recent. You can envision the other way, where recent is worse in this case. And what I want is really, really good scientific research and that kind of stuff, depending on, you know, the field that you're asking it to do the research in. And when it comes back, you have the ability to essentially learn anything.
- [31:22]
Chris Reynolds
I'm imagining what this, what sort of impact this is going to have on the field of. Of law and attorneys that outside of being. I could be wrong, but outside of being an attorney in a courtroom arguing still to try to convince other humans on a jury, you can use deep research to do all of the research on case law, historical jurisprudence.
- [31:50]
Matt Reynolds
Yes.
- [31:50]
Chris Reynolds
Right. So judicial precedent, everything can be done there. And every single law firm has kids either in law school or fresh out of law school. That. That's all they do.
- [32:01]
Matt Reynolds
That's right. We don't need that.
- [32:02]
Chris Reynolds
And now you can literally just ask. I need all of the case law. I need all of the precedent that's been set. And I've wanted in historical order, either from oldest to newest or newest to oldest, or weight it based on this. And very quickly, you can build your data, you can build your model, you can build your argument on behalf of your client based on all of that historical data that's already out there. You don't have to go in. And if you've ever been in a law firm and it's just like books and books and books and books and books. Well, all those are actually now on your phone.
- [32:35]
Matt Reynolds
Yeah, that's right. And you can do things that. So when you start doing this, the first thing you're going to do probably if you're like me, is you'll ask it to do very general research about topics that you're just sort of interested in, and you'll be wowed by how good it is. Like, it's phenomenal at how good it is. But I think the place that it really has incredible application is when what you ask it for is something incredibly specific. And so I'll give you an example of this. Every so often we run into a problem technologically, and maybe a good example of this would be, let's say that you have a common application that you Use. I would guess that most people that are listening to this have Gmail or Outlook one that they use for email, let's say. And let's say that you have a problem where you would like your stuff to be categorized better whenever it comes in. Like let's say that your emails come in, they're not categorized enough. It's just chaos in your inbox and you could spend forever trying to figure out how to do it right. Or you could go to Deep Research and say, I use Gmail and I get emails all the time that are not categorized properly. Here are a couple of examples of what I get. Here's my current, you know, folder structure or whatever. Build me a step by step guide that walks me through, click by click exactly what I need to do in order to set up my inbox so that everything is automatically routed to exactly where it should go. What Deep Research does is it comes back and asks you super fast. It'll ask you like the top four or five clarifying questions. So it comes back and it goes, okay, cool. Can you give me a list of all your folders that you currently have set up or what you want to have set up? Yes. Can you give me six examples of emails that come in and your whatever. So you're going to paste it some additional context. This is hugely valuable because this is the part where this is the nice way of what I said last week, which is most of you are probably worse communicators than you think you are. Probably less clear than you think you are. I know I am. I come back all the time. I'm like, oh, why didn't I, why didn't I say that thing? That would have, that was very relevant information I should have asked for and then it'll go out and actually do that for you. So what comes back at that point is a step by step guide for Chris Reynolds to set up his inbox appropriately for, you know, my kinds of emails that come in or whatever. Like, it's just crazy. It's crazy how good it is. The actual examples that we use frequently, we have a lot of programming examples we use anytime we need to do anything that is extremely technical. And what we really want is to follow a step by step guide. It will make a step by step guide for you.
- [35:10]
Chris Reynolds
Okay, let me ask you a question.
- [35:11]
Matt Reynolds
Whatever you want.
- [35:12]
Chris Reynolds
Because here is the place where I'm straight struggling a little bit to try to make sure that it, I don't know if it can do this yet and I know it's coming so Gmail, like you said, almost everybody has a Gmail email account. So that step by step guide is. That's going to be clear. I sent you this morning. We were kind of talking through a post Arnold classic SOP where I've actually worked with different AI models to get the steps needed for the sop. What's not included is the step by step guide to to perform those steps inside HubSpot, which is the CRM we use. So you use HubSpot or Salesforce or whatever. Do you think it has the knowledge for me to go in and say these are the steps? Now what I need you to do is I need you to add what are the HubSpot exactly. How does my sales team moving someone from awareness to lead to qualified lead to follow up call to acquisition to activation to whatever, what would those steps be step by step in HubSpot specifically? Do you think it already has the knowledge of that or will it say HubSpot doesn't really put out their API, we're not really sure what's going on or what do you think it would spit out?
- [36:26]
Matt Reynolds
Okay, I'm going to answer that in two ways. So the first way is if you have a completely standard HubSpot with no additional custom fields, which would be odd, then the answer is yes, it could tell you exactly how to do it. HubSpot is a flexible framework that it can't know yet. There's a good answer here about how it can know, but I'm going to start with how it can't. Deep research can't know what you've done unless you go in and provide it that as context. So you would need to download some stuff from HubSpot to provide it to the search engine here and to make it get you what you want. So is it possible. Yes. If you, if in a generic way, if you haven't set up a whole bunch of custom fields, then it will answer you probably very close to perfectly, which is insane. But it can, because HubSpot is insanely well known and there's a ridiculous amount of written stuff about HubSpot and their API and all the other things that are that are there. So answer number one is it could give you a very good generic guide for how to do it with your sop. It could absolutely do that. You provide the sop, you say do the thing. Let's see what, let's see what comes out. Right. The better answer is the lead in to the next podcast, which is there are a set of tools now that make it Very fast. I will not necessarily say easy, you got to know some stuff so it can be a little more complex, but that make it fairly simple for you to not only take that SOP and have it produce a perfect step by step guide to how to enter everything into HubSpot. But better than that, we'll build you a small application so that all they have to do is a field by field entry that will automatically interact with HubSpot's API and we'll put all the data in there for you.
- [38:19]
Chris Reynolds
That was going to be my follow up question so we'll save that for the next one because I can imagine that for those of us who are not right on the cutting edge of this and for those of us who are not completely terrified of AI and want to figure out how to use it to help our business. You know when you were giving the example of Gmail, my first thought is why can't I just give it my login to Gmail and have it go ahead and do the thing as opposed to write me the step by step. And I know so again and I think you mentioned this last week, you can also then straw man, you can use AI to strawman that argument. I know there will be people listening to that be like I'm not giving AI access to my Gmail account, right. But, but some of like my business account is I, I probably would, maybe I shouldn't and I, I'll ask you about that on the next, on the next episode. What I'd love to do is not just give me a step by step guide on how to put things in the right folder. I want to say you put it in the right folder, right.
- [39:12]
Matt Reynolds
And so for non programmers, especially people that are maybe not even less on the tech curve but just people just want to have a simple solution to these things. OpenAI came out with a tool called Operator that does exactly that. It will absolutely just kick up your Gmail log in and do the stuff that you want it to do. That's part of my Pro account. I don't know if you have, I don't think you have that.
- [39:36]
Chris Reynolds
But again no operator. And then there's likely between this episode and the next episode I'll be getting it because if you tell me if we can figure out how much more value you can get out of that, especially if it also adds some of those safety rails to same thing for those HubSpot steps. If there's a way to connect AI to HubSpot which obviously all these massive multi trillion dollar or multi billion dollar HubSpot's a high, high, high multibillion dollar corporation. That's why they're trying to get AI to work with it so that someone doesn't have to build all these webhooks and different stuff step by step odd things and that you can just, it will do it if you can tell it what you want it to do.
- [40:20]
Matt Reynolds
Right. So there's, there's two tools that are, are really useful right now. There's actually, there's one more. Microsoft just came out with it, but I can't remember it off the top of my head. OpenAI's got one, it's called Operator and it literally will, will open your inbox for you and do all the things while you watch it. It's like an automated, it's like an automated thing computer user. The other one is from Anthropic called Computer Use. It actually came out first has I think maybe some more tools. I've used it a few times and it does the same type of thing. What it'll do is it'll, it'll, it opens a little window, it opens a browser in that window and you just tell it what to do in regular human words like log into my account and do this thing. Here's my username and password. There's a way of managing secrets safely that it has built in that you can use and yeah, and it, and it does all that stuff for you. So that tool is those tools are coming very, very fast to basically be able to use your computer for you with the movement of a mouse. And what it does is kind of give you an idea of how it works. What it actually does is it is using sort of rapid screenshots of that stuff. So it'll take a screenshot and it will process the screenshot to determine where it's at, where it needs to click and what it needs to type. The problem with these models right now is that they actually aren't super fast. So it's not like it's blazing going through, you know, clicking all the screens and doing all the things. It's doing image processing really fast and that's actually kind of, it's really fast for it, but it's slow in general. So the process is still early. It's, it's early days in that one. But definitely some good stuff coming along. It's way faster and I mean faster by orders of magnitude. If the tool set that you're using is, is able to utilize your command line, which is the like old DOS terminal For those of you thinking of it that way, if you're on a Mac, it's a what is essentially a Linux terminal and it has infinite capability of doing infinite things. And that one is so magical that it is. It's nerdy and I love it, but it is probably not for everybody. It's also somewhat dangerous if you don't know what you're doing. So a lot of things that. A lot of bad things can happen if you do it the wrong way, but one wonderful things can happen if you do it the right way. So it just, it gives you a lot. Hopefully you can kind of start seeing like, oh, wow, this is insane. And then before we go off that topic, the last thing that I just want to mention is that what Claude does, if you look up Claude. So Claude is. Is another one of these tools we talked a lot about. OpenAI. I actually use Claude almost all day, but it's the model I'm using inside of a tool that we'll talk about later. Called that. The one I'm using right now is called Cursor and I love it. The model is insane. If you just get a Claude account.
- [43:04]
Chris Reynolds
Which is just C, L, A, U, D, E. Yep.
- [43:08]
Matt Reynolds
What it does is it's the beginning of what I'm going to talk about later, which in the next podcast, which is you have the same kind of conversation that you're going to have with ChatGPT. Right. Same type, but instead of after each one, instead of saying, oh, that's good, that's information that I learned, what you're actually going to do is you're going to tell it to store its answer into a bank of answers that are files for you that. And we're going to build that up as documents that are ultimately going to turn into something or business library.
- [43:42]
Chris Reynolds
Right? Is that what you're saying?
- [43:43]
Matt Reynolds
That's correct.
- [43:44]
Chris Reynolds
And then that's one of the biggest struggles, so I'm really interested to hear that conversation, is that if we build up all of this information that isn't out there on the Internet, but that I'm able to feed it, but I don't want it to go on the Internet. What I'm hearing you say is Claude will build that library for you as a founder or in whatever area of your life you could put in your health metrics or whatever the thing is. Right. And it will build that library for you that's just for you and other people then can't access that library that's outside your business.
- [44:17]
Matt Reynolds
That's correct.
- [44:18]
Chris Reynolds
So. So as you continue to build that bank, then as you ask it questions, it's not just pulling from the Internet at large, it's pulling also from your specific library.
- [44:28]
Matt Reynolds
That is exactly right now it is the right place to take your next step as you're sort of expanding your AI universe. But it is also not the end. There is even better than that.
- [44:40]
Chris Reynolds
We're not going to get to the end no matter how many podcasts we do because this changes so fast.
- [44:44]
Matt Reynolds
That's. That's right.
- [44:45]
Chris Reynolds
We're going to try to take founders to the spot that. So what I want to do. Let's do this. So in episode one, we talked about getting really familiar with text based AI applications. Chat GPT is a great one. And again there's, there's a lot of. You should play with a lot of them. Grok is good. I mean again, the stuff that, you know the story of what Elon was able to do and buy what, what was it, ten thousand or a hundred thousand GPUs, whatever it was from Nvidia and put them together and like, yeah, Grok is excellent, excellent. It's insane, but it's really just playing around with text based app. The text based AI applications is step one, step two, once you're able to get there. And this is now the step that I'm playing with. It's using AI for decision support. Right. And so some of the things we didn't even get to on this, on this episode, but talking through things like helping me make a decision about market research, about the company, about do I need to change the direction of the company 5 degrees this way or 5 degrees this way. I might put something in it about our B2C business and our B2B business and it should say, hey, you should totally focus on this part of the business because that's where the growth is and that's where the margin is. It can help you with that, it can help you with decision making with financial modeling, which I sent you some of our financial modeling to be able to, to push that out and say like here, this looks good. Gross margins, not great on this product. Here's some ideas on how to change it. It will help you with that, those sorts of decisions. So and as well as things like the game plan or OKRs, KPIs metrics, it will help you build a game. What are the major goals, knowing what you know now about our company. And I think this will have a lot of overlap with next week's episode as we build the library and it knows more and more about the company.
- [46:29]
Matt Reynolds
You got it.
- [46:30]
Chris Reynolds
And you say what are the goals that the company should have? What should be the top three goals it can tell you? All right, well then what are the top three to five actions that we need to complete to get to those goals? And what are the best metrics or KPIs to measure that we're moving in the right direction? That's where we're going to go next for decision support. And then you're going to walk through next week the practical implementation guide of how to actually do that for your business. Not just in a great place to start in a just Internet wide everything that the Internet knows, but into how to actually make it personal for you and your business.
- [47:03]
Matt Reynolds
Yep.
- [47:04]
Chris Reynolds
And use it in a way that gives you a tool that really creates a moat in your business.
- [47:08]
Matt Reynolds
You got it.
- [47:09]
Chris Reynolds
There you go. So there's episode two on the AI class by Chris Reynolds on the Build your Business podcast. I've learned a ton. I was thinking like I'll probably go back and listen to this wave format before it even gets posted online because I've got all these notes in my head, things in my head that I want to stay focused on the conversation. But I'm like I'm going to go back and re listen to it probably today and then take notes off of it again. The nice thing is I'll just feed it into AI.
- [47:34]
Matt Reynolds
Let's make AI do it just how.
- [47:36]
Chris Reynolds
I gave me the summary in the notes. So thank you for listening to another episode of the Build your Business podcast. If you got value from this, as always, we would love a five star review that helps us so much with the algorithm. If you haven't done that yet, please do it. Share it with a friend or family member or founder that you know and love and want to see them succeed. And we'll see you guys next Friday.