285 - Best of 2025! How to Think Like AI with John Long

0:00I came across a really interesting study that was published by Harvard. They studied 776 people at Procter & Gamble. They had individuals using AI, individuals working by themselves, teams using AI, and teams working by themselves. But the individuals using AI were happier, produced as high or higher quality of work as the teams working by themselves without AI, multitudes better than the teams working by themselves. What it proved was AI absolutely should be a team member. It is something that very soon we will not be able to function normally without it.

0:38John, I'm stoked to learn something from you today. I think I'm going to learn a lot. Tell me really quickly, what's your business? Yeah, so I'm one of the co-founders of think.ai, T-H-Y-N-K.ai. And we provide businesses with AI agents that can help automate processes for them. Our agents can make and receive phone calls, send and receive text messages, emails, hold meetings, schedule meetings, interview people, variety of things. And so we help automate processes that involve communication typically with customers or sometimes internal communications as well. Okay. I'm really excited to check this out because there's a lot of talk from people who are like, oh, AI is coming.

1:19AI agents are here. I've played around like with operator and different agents. Usually when people talk about agents right now, it's like robotic process automations that have been rebranded as agents. But I'm excited to see what you've put together. And you're not just an entrepreneur who's using AI. You're an entrepreneur who has a company that helps entrepreneurs use AI. So it's pretty meta. And we use the AI to build our business, right? We are our first customer. We use what we sell and it works to help put us in touch with people. So it's awesome. Dude, that is awesome. All right. So let me ask you the first question that I ask everybody.

1:53What models do you use and for what purposes? Yeah. So we use a variety of different tools. When you say what models, that's kind of a, you know, a LLM type of a question, which is great. We do use chat GPT on the backend. We're using 4.0 for some of our conversational things, but I mean, in our business, we're in perplexity. We use Clawed occasionally, you know, chat GPT, a variety of different. Are there different use cases you've found that are better for some than others? Yes, a little bit. I'm a chat GPT user. It knows me well. I've been using it for a long time. And so I don't, you know, my prompting there is pretty easy because I can ask it something and it knows so much background about me that it's able to answer so much about my family, about my business, about, you know, the way that I, that I operate in life, et cetera.

2:45We have an AI podcast that's completely AI. We use Notebook LM for that. I don't know if you're familiar with that. That's a Google product. And so that's more of just kind of a fun thing that we do. We actually, we're using it to stay up to date on all of the things going on in the world of AI. And we were like, well, we're making this. Why don't we publish it? And so, you know, we use that. We don't do a lot of coding. My business partner is more on the operations side. So he gets into that a little bit more. And I think he uses perplexity a lot more than he's using ChatGPT. Generally speaking, these are the use cases.

3:18Chat CPT is used as a general thought partner to sort of ideate with. Claude is used a lot by coders who are writing code. Perplexity is used a lot for research because of the real-time access to the internet. Gemini is used when people are trying to tie the Google suite together. So there's a whole that I'm seeing so far. It's like, oh, these are kind of the general use cases that people use for these items. Notebook LM is interesting, though. You're the first person to talk about Notebook LM. I've used personally use notebook LM for my podcast because what I've found is I'm going to use words.

3:50I don't know. So you'll correct me, but it's kind of like a closed sandbox in the sense that I don't get a lot of hallucination when I ask it to go in and pull a transcript and maybe do an analysis for me or write a summary. Whereas with chat GPT, I'll get a lot of hallucinations unless I'm very specific with my prompt. What do you use notebook LM for? Like I said, we, we have a podcast that we produce. It's a weekly podcast. It's on the news of the AI world. And so we collect recent publications, news stories, et cetera. We load all of that into Notebook LM, and then we use its podcast feature to create a podcast that we publish every week.

4:30I haven't seen the podcast feature. No, that's not true. I have seen the podcast feature. I've never used it, though. It's pretty cool. Yeah. Is it good? Yeah. Did you clone your voice for it, or is it using its own? We have it just using its own. You can clone your voice and you can actually come in as a guest on the podcast and the AI hosts will interview you. And, you know, there's a lot that you can do with it, but we use kind of the basic features of it. And again, normally, you know, initially we started using it just to help us stay up to date on the, all of the happenings in AI. This industry is moving so fast.

5:03It's a little crazy. Right. And then we were like, let's just share this. Why not? How has the podcast done? I'm so curious. How many downloads do you get a week or how many listens or how many people are? Gosh, I, you know what? I don't even pay attention to those numbers. So I honestly don't know. I apologize, Nick. I should know those numbers. You don't have to apologize. You apologize for not knowing your numbers. I'm just curious. Are you looking at it as like a potential funnel or is it just, Hey, this is cool. Why don't we just put this out there? Yeah, a little bit of a funnel, but not something that we really push a lot of content to or market in any way.

5:38I think it's something that we mentioned in a newsletter that we send out to our email list each week just as a giveaway. Yeah, that's cool. I like that. So talk to me about Think.ai. You said it helps business owners automate their business. What does that mean? I'll tell you the history, and that will help explain what we do. and what our AI agents actually do. So my business partner and I had spent a number of years in technology and in sales. And for years have talked about, you know, hey, we need to go on our own. We need to go on our own, start our own thing. And so about a year and a half ago, we started going down that road more earnestly with a technology that was in the physical security space.

6:25And as we were starting up that company, we said, gosh, AI is really coming on. Wouldn't it be cool if we could find an AI solution that could help replace the pre-sales motion, right? The prospecting and the qualifying of leads, warming things up, discovery calls, et cetera. And wouldn't it be awesome if it just put appointments on our calendar with prospects that, you know, had already gone through a little bit of a - We're qualified to some extent. We're qualified, exactly. We looked and looked and looked and looked, couldn't find anything that did everything. We found tools that would do the texting and automate the texting, but it wouldn't do the phone calls or technology that would do the phone calls, but it wouldn't do the emails and nothing that did everything.

7:09We did a lot of research. We found bits and pieces of different tools. We put them all together, put together an automation using a CRM, high-level automation backend, and we built it. and initially built it for us to use. And then as we started to recognize, hey, this is actually a pretty cool tool. We said, this could be a business. Maybe this is something that we should actually be pushing. And so we pivoted away from the physical security side and went all in on AI. And that's what we've been doing now for about the last eight months or so. So it started as a sales assistant. I don't know the right word, but for the sales process, is it strictly sales still or have you expanded outside of sales?

7:48So it can do a lot more than that. the same platform is used, but it can do customer service. It can do HR. A month or so ago, we used it to conduct like 60 or 70 interviews for us as we were going through a hiring process. So it can be used to do a number of different things, but anything related to communications. Are you still using other technology and this is a wrapper for it, or have you developed your own text-to-speech model? It's a mixture of things. So we're using other technologies underneath. underneath the hood, but we've been able to stitch all of those together and have them work in a, in a unified kind of omni-channel fashion.

8:26So. So it's interesting. I used air.ai a couple of years ago when it first came out and it was, it was cool. It was a really, so like you knew the future was coming, but there was still kind of a lag. It didn't feel great. Has it progressed? Has it gotten a lot better or is it still laggy? Like do people, do you have to tell them upfront, for example, hey, this is an AI agent that you're talking to. It depends on the person. We have people. In fact, we had someone just the other day on one of the phone calls. He had been on with our AI for like five minutes and he goes, wait a second, is this a human or is this AI?

9:00And we've trained our AI. Sometimes people say, oh, just lie, tell them it's a human, right? And we don't do that. We're very upfront with people. And so the agent said, yeah, I'm an AI agent designed to help answer your questions about Think. And the guy was just blown away. You could hear over the phone, him kind of gasp and go, oh my gosh, you are the best AI I've ever met. Oh, that's awesome. So that was, that was cool, but it, it is still AI. You know, you still have latency issues sometimes where it pauses a little bit longer than maybe as natural, or it comes in a little bit sooner than as natural.

9:31That's getting better. There are models out there right now that very difficult to tell. And so we're watching those very closely. And as those get tested to the point where we're comfortable, we'll, we'll continue to upgrade and, you know, make it better better so that hasn't really been an issue for you the ai component to it no that's surprising to me because the few times that like i tested it out as a cold caller where we were cold calling mobile home parks off market and then other businesses off market it was very polarizing some people were like oh this is cool yeah we'll talk to it and then other people were like click once they realized it was an agent it surprises me to hear that like it's not an issue for your sales teams to have that first call go through an AI agent?

10:12No, it's not. And in fact, it's awesome to watch. You can see as leads come in and the AI agent starts to engage, you know, it'll call them, it'll text them, it'll email them, it'll have this conversation. And as that conversation progresses, you can literally watch on the back end as that happens and see this conversation emerge. And the agent is masterful at getting people to eventually book an appointment with us. I guess if the first interactions with it are text-based, that's very different than the latency issues that might come from having a conversation with them, right? Yeah, and ours does call.

10:51So if someone responds to one of our marketing pieces, our agent will call them. That is our first outreach call and text. And so we don't hide from a phone call by any means. It's great. Do you track metrics that are like, hey, human versus AI, these are the conversion stats, maybe to show that there is or isn't any difference between the two? We're 100 % AI. And so we don't necessarily have the ability to compare like this half of the house and this half is human. But there's probably some like industry standards or benchmarks that you can compare against, right? So I actually, I came across a really interesting, it's perfect timing.

11:32I came across a really interesting study that was published by Harvard. And they did a study with Procter & Gamble. I just saw this this morning. They studied, I think it was 776 people at Procter & Gamble. And they had individuals using AI, individuals working by themselves, teams using AI, and teams working by themselves. And the individuals using AI, now this isn't exactly phone call only, it was using it in other areas as well, but the individuals using AI were happier, produced as high or higher quality of work as the teams working by themselves or without AI by themselves. And then the teams working with AI were like multitudes better than the teams by themselves.

12:20Yeah. So that was one of the first truly research-based studies from a very reputable non-AI entity on the effectiveness of AI. And it was fascinating to see. I mean, what it proved was AI absolutely should be a team member. Not necessarily replace the team, but have it be a team member and you'll instantly elevate your team to levels that you haven't seen before. That's awesome. What have you found amongst your customers? Is there kind of like an ideal customer that this works for? Like, oh, home services with a sales team or tech companies with a sales team, or is it not even sales related? That's the best use case for it.

13:01So it doesn't have to be sales related. That's because that's where we came from. That's kind of where our focus has been. And so really our ICP customers have three characteristics. Number one, they have processes that they follow, right? So they're mature enough in their business that they have an ABC sort of a process that they go through. Number two, they have people helping with those processes right now that could be better utilized elsewhere. And so people. And then number three, they are spending money on those processes. They're spending money on marketing to bring leads in or they're spending money on moving things through those processes.

13:39We can help in those areas. We can help reduce costs. We can help reduce time. and we can help processes move more smoothly. That's an interesting framework because you're not just saying, hey, anybody that has sales in their business, you can use our AI agent, right? Like you've got a pretty defined customer. From everything that you just said, my guess would be like, that's a company that's at least a million dollars. You know, maybe there are some that are less than that, but these are companies that have some infrastructure, some leadership that are putting real dollars behind it. And they're probably, from your perspective, that they're going to be more successful because they have developed internal systems.

14:16Whereas smaller companies, you know, they see these solutions all the time. They're like, oh, this will be the fix and they implement it, but they don't actually have any systems. They don't actually know how to utilize it. And they, they're not set up for success. Yeah. Now that being said, we work with a number of customers that this is becoming the process, right? They didn't have a process, but AI and their use with the things that we're providing has become their process. And that still works very well. And so if you have a, you know, know, a smaller home services company, I've been kind of a serial entrepreneur, a little bit like you, Nick, I owned a heating and air conditioning company at one point.

14:48And I know how it is when you're busy, when you're running and the phone rings, you know, I've got to answer that call or I'm going to miss an opportunity. I'm going to miss the next job, but sometimes you can't, if you've got your hand stuck inside a furnace, or if you're, you know, already on a call with customer number one and customer number two is calling at the same time. And you sometimes just run out of capacity. And that's where something like an agent can step in. And with our agent, you could have 300 calls come in at 5 a.m. on Christmas morning, and every call will get answered. Every call will be personalized to the person that calls.

15:26Appointments would be scheduled. You wouldn't lose it, anything, right? And it could all happen at the same time. I talk about this a lot. The first, in my opinion, big hurdle that entrepreneurs face or challenge as they're growing their business is the ability to systematize and then effectively communicate that system. Many entrepreneurs get to the point where they're only ever a one-man shop or one-woman shop. I mean, that's okay if that's what they want to be. But if you want to grow, you have to be able to create a system and then scale. So maybe that's the better criteria for you is like, hey, you have to have some systems or at least willingness to systematize in order to use this?

16:02It really is the difference between owning a job, right? There's a lot of people that simply own a job, right? They own a heating and air conditioning company, but really it's just a job. They are a technician and the calls happen to come to them and they do the work and then owning a business where you have that structure that you described. Yeah. And one of the things that I've been thinking about AI that I talk about a lot is bringing on these AI agents or bringing on these AI solutions. They're like an employee and you need to treat them like an employee. You need to have very clear expectations, job descriptions, and then like a very clear onboarding process, quality check-in performance management.

16:42Like you don't just set it and forget it because if you do, it's not going to work out. You have to be very clear with what you're expecting from them because then you'll get more out of them by being clear. But many people don't treat them as employees. They're just like, oh, well, it's just solution. It's technology. It should just work. Well, and it's a lack of understanding. And I don't mean that in a bad way. It's just, this is new, right? For sure. And people need to understand that while AI could be a silver bullet for their business, it's not a silver bullet that doesn't need training and guidance and direction, right?

17:12Totally. Can AI revolutionize businesses for people? Absolutely. You know, you've got a lot of entrepreneurs that watch these, these podcast episodes and they're looking for that thing to set them free. And AI may be exactly that, but it doesn't set itself up. So you still have to provide kind of that structure. A lot of times when we teach people about AI, we teach them the 10, 80, 10 principle. And maybe you've heard this before, but the first 10 % has to come from the humans, right? The setup, right? The structure, the direction, the, you know, the end goal that you're, that you're shooting for.

17:46The middle 80 % can come from AI and it can do it very, very quickly and it can do it at a very, very efficient way. But then the last 10%, that polishing step, that final product still typically has to come from humans. Right. And so, especially, you know, if you're using like an LLM or, you know, a generative AI type of product, agentic agents tend to get a little closer to that a hundred percent mark on the backend, but they've been really well polished before they can get to that point. I freaking love that. I love that framework. 10, 80, 10. I've never heard that before, but it makes total sense.

18:25You have to prompt it, train it the right way, let it do its work. And then you still have to do something on the backend in order to make sure that it's, it's producing the type of output that you want it to produce. But that framework, the 10, 80, 10, that's, I love that, John. I freaking love that. Yeah. Well, steal it. Use it. Can I try out your, your software? Yeah. You want to do it? Heck yeah, man. Let's try this. Okay. All right. So. All right. Now's the part of the show where I feel the most uncomfortable, but my therapist says I need to face my fears. So here we are. I've started a newsletter and I want you to subscribe.

18:53And what you're going to get every single week are the aggregated conversations from that week that I have on this podcast with an overview of what their business actually looks like. I'm also going to throw in a review of one or two businesses that are listed for sale. I'll give you my opinion on whether or not the EBITDA multiple is good, or there's customer concentration, or there's red flags or green flags. And then lastly, I'm going to give you one piece of actionable advice every single week on how to buy your first business. So So click the link below, subscribe to my newsletter, and let's get back into the show.

19:20Let's have you go to our website. So go to think, T-H-Y-N-K dot A-I. And when you go there, you'll see we're up in the upper. Test R-I-I. Go ahead and put your information in there. And then while you're doing that, I'm going to share my screen. Now, Nick, what's going to happen is our agent's going to call you in about 30 seconds. Don't answer your phone. And I'll walk you through a little bit of a process with it. Okay, so I'm going to say I'm ready to test AI, and then I'll keep an eye on my phone. I always put on do not disturb, so I'll even take it off, do not disturb. I'm going to show you on the back end.

19:52So here we can see that you've come in. Provo, Utah. Skynet is calling me. Skynet. Oh my gosh. So here you can see on the back end, again, this is a high-level CRM, and this CRM comes with our agent. You do not have to use it, but a lot of companies love high-level as a CRM. we can connect with any CRM system that, that they're using, but here we can see it sent you a text message. It sent you an email. Oh yeah. Yeah. Yeah. Hey Nicholas, this is Eric over at think AI. So go ahead and respond to that text message, Nick, and just ask him a question. Ask, ask Eric, anything you want about AI. Anything I want about AI.

20:30Sure. Eric, is that your real name?

20:37so we're going to see that pop in here so here it is here's your question eric is watching the workflow on the back end oh you're going to see this on the back end now eric could respond instantaneously he could call you as soon as you hit that button a second ago but we've trained this to say wait a little bit give it a 30 second pause to feel a little bit more human-like right And so the AI is automatically updating this opportunity from a new lead to responded to show in our system that, hey, all right, we got a live one here, right? Nick is engaging with us. He's interested. And so here in just a second, we're going to see a text pop back in from Eric responding to you.

21:17So here he said, and this is probably hitting your phone right about now. You know, absolutely, Nicholas. Eric is my real name here at Think AI. I'm here to help you explore our AI solutions and how they can benefit your business. What sparked your interest in AI? And you can see he's already starting to engage in the conversation. I'll respond. I'll respond. I'm looking for an after hours solution to handle my inbound calls. Perfect. I don't know if you do that, but that's just what I said. Yep. Absolutely. Well, let's see what Eric says. We'll let Eric answer. I don't want to answer for him.

21:51I'm going to let Eric tell you what he can do. If Eric doesn't do well, I'm worried about him. You're not going to fire Eric if he messes up, are you, John? Well, we'll see. He knows he's being watched, so he's going to be on his best behavior right now. Now, when you trained Eric, is it Eric every time or is it like randomized? Eric and Sarah and Lauren? So for us, it is Eric every time. And here he's responding to you. but you can have different agents and each agent has a directive or an objective that it's trying to achieve. Eric's objective is to qualify people and then schedule them for a sales meeting with us.

22:29But you could have agents that will then, you know, Eric could potentially qualify you as a sales target and then pass you to a sales agent whose job is to pitch you on our products and sell you. We have to go to a live human at that point, but we could go from one agent to another if we wanted to as well. So this is interesting, right? He sends me the first message. I troll him, Eric, is that your real name? And he's like, yeah, it's my real name. What can I help you with? So then I tell him my problem set, hey, I'm looking for after hour solution. And he responds. But the cool thing about the last response is not just let's set up a call.

23:03It's, hey, tell me a little bit more about your specific challenges. Like he's further qualifying me as opposed to just sending me straight to a person, which feels less scary, right? Like it feels less like I'm getting... So let's do this then, Nick. Let's up the game here a little bit. Go ahead and send Eric a text and just say, hey, can you call me or call me back or something like that? Okay. And then when he calls you, go ahead and answer this time. I'll be quiet so that you can have a conversation with Eric, but put him on speakerphone and go ahead and chat with Eric for a minute or so. And let's see how that goes.

23:37We haven't planned this, by the way. I'm going to stop my share so we can see your side better here. Okay. Is there still that 30 seconds? Oh, no. Eric is calling. Oh, it's like a first date. Yeah. Hi, this is Nick. Hi, Nick. It's Eric from Think AI. How are they going so far? It's going well, Eric. It's going well. Thanks for calling me. I'm glad to hear that. I'm here to chat about how Think AI can help your business. What sparked your interest in AI? Yeah. So I have a company. We're a home services company and we get a lot of calls after hours, but I don't have a team that can actually handle my after hours calls.