This is the business tech playbook. Your source for it, help for your business.

Hey, guys, welcome to the podcast. I'm your host, Rob Olson, and with me I have BJ Pote, the CEO of E. Top technologies. Hi, BJ.

Hey, man. Hey, Rob's. How you doing today, sir?

You know, I'm dippity doing. I've had. It's a great day. It's sunshiny out here in Minnesota. I'm looking forward to Del Taco next week. In Florida, of all places.

They have Del Taco in Florida?

Yes, they do. I'm so excited. They won Del Taco in Orlando.

Oh, that's amazing. We know that you're taking Uber to wherever that is.

I will. I will thumb for a ride, which, in case kids don't know what that is, that's hitchhiking. I don't know what the. In the fifties, they used to call it thumbing for a ride. I don't know what it means now. Now I'm kind of scared and have to look it up later.

I'm not even sure you were born in the eighties. So, like, how do you know it's about the fifties?

Here I have a 1947 vehicle, and it had a little thing saying how to thumb for a ride inside it in a little booklet.

All right, well, I didn't realize having a 1947 truck automatically gave you, like, street cred from the fifties, but this was. Sorry, what?

This was a sedan. A Plymouth?

A sedan. Oh, yes, of course.

Dilly dilly.

Oh, shoot.

For all I know, they chased down Bonnie and Clyde in it, but no way we know, though, that would have been thirties. Yeah, see, you came here for tech, now you're getting history.

Well, and I mean, honestly, a 1940s Plymouth. I mean, you could run over Bonnie and Clyde and you wouldn't even dent your fender.

Yeah, still all metal.

I mean, you know, you might get blood on it, but you definitely wouldn't dent it.

Not. We're getting cryptic.

We. Oh.

Anyways, onward. Right, onward to the point. So today's topic we're going to be talking about is AI in business. Now, there's probably going to be multiple of these. We don't exactly know what questions are going to come out of this, but most of the sales calls that we're getting now for etop involve questions on how products such as Copilot, more integrations for automation in 365, and other questions about AI in general are affecting business. So why not bring it to the forefront? If our customers are asking questions, we're going to use that as a topic for the podcast, because why not?

Exactly. When Robbie came to me and was like, podcast topic, I'm like, well, I've been kind of avoiding AI because I didn't want to be a hanger on. But at the end of the day, like, part of the core value of business tech playbook is to explain current technology, hopefully demystify some of the acronyms and some of the b's of it, to be perfectly honest, because there's a lot, there's a huge amount. But at the end of the day, we've had three different sales meetings, both several with internal clients and several new prospects of people just going, we know that this is coming. We're a small company and we're trying to figure out how to leverage this before our competitors do so we can pick up any competitive advantage that we can. I know that internally at etop we've been diving into this for pretty close to 18 months and we've been trying to find new ways to use AI pretty much every day. And so it's been really, really enjoyable conversations when people start seeing the light of everything that you can do and then also being a little bit scared about some of the possible ramifications and issues that they could run into if they're not careful.

Well, I don't want to start with doom and gloom. I'm a positive guy. First of all, what's the from a residential standpoint, we'll start there. I know me putting work aside, I'm sick of hearing about AI when I go to Facebook and my simple search bar says, oh, try this new meta AI and everything has to integrate with AI. I'm kind of getting sick of this new buzzword topic. So residential speaking is much different than business. I want to put that first and foremost before we get into our bullet point topic. It seems like every platform that we have out there is trying to put up the tagline so they're capturing the marketing buzz. That's not what we're trying to do in this podcast. We are showing you the active, measurable pieces that we've seen so far that are done in general business.

Absolutely. And so there's a couple of like kind of main tools that you'll hear about out there. There's a couple of really huge players. And I'd like to kind of dive into some of the opportunities, some of the challenges, some of what you should just be cautious of and recognize that it's coming. We need to figure out as businesses how to use it, because if we don't we will at some point be left behind, but at the same time, does it solve all the problems? Well, no, it creates a whole new bunch of problems, but at the same time the opportunity in it is quite large. As long as you're, you're moving forward and aware of it, though. First, I think we're going to cover some of the basic tools. We'll say the biggest players in the market, we'll talk about some of the b's, we'll talk about some of the good, the bad and the ugly. At this point, pretty much everyone's heard of OpenAI and chat GPT. I'm assuming most Americans have at least seen the website, tried using it, chatted with it, and then just to see what it's even like officially been out, I think about two years, maybe a little bit less. I've been personally using it since like February last year.

And if you haven't, stop and give it a go. Just ask it, render questions. There's no wrong answers. It's just a free space to, to try to talk to an inanimate object and see what answers it gives you. It's quite fun.

Exactly. I literally was in a meeting today with a client who's a CPA, just talking through like the challenges of AI. So what I did is I went into the meeting, talked with the guy and was like, hey, let's just try this out. I turned on the chat GPT voice version and I was like, chat? Tell me, like here I'm in a meeting with so and so. Here's the conversation we're having. I need you to give me three ideas that we should talk about during this meeting for AI. And it like, hey, it's really good to meet you, so and so glad that you're here. Here's three ideas. And it basically ripped out. Like we had a conversation with it around everything we talked about. And then I was like, here's something else that I have personally been using chat GPT for, is I'll leave a sales meeting while I have all the content fresh in my mind of the conversations. I'll basically have it draft me a recap email covering all of the key topics and key points. And by the time I get back to and then I also tell it, resummarize this for my CRM. So that way I have all of the information from the meeting and it I'll go back, grab the email, send it, I'll go grab all the content, put it in the CRM, document my conversation, and now I was able to basically treat it like an assistant while driving back from a meeting. So it, to me, it's just been like a really huge, like time saver in that front. And honestly, it blew his mind just having the conversational aspects of it.

Just from the personal assistant standpoint, that alone was enough to change the day to day and consider like, hey, maybe I should try AI.

Well, absolutely. I also use it for shenanigans relatively regularly. If I have a joke that I'm trying to pull, like, I'll have it create funny images. I will do, I'll do all sorts of just random stuff because it's relatively easy to go from nothing to like 95% whatever what I'm trying to do in like under five minutes. And so it's like I was joking with some friends about a playing card around a specific thing, and so I told it to make me a playing card with like these characteristics and these ideas, and it had it made in like under a minute. And the guy, it blew his mind. I'm like, again, the amount of funny and stupid things that I've done with this is really good for my personal outlook on life because it's, I like doing things that are somewhat entertaining to me, but at the same time, it's also a really good way to learn how to use the tool to get it to do what you want. So as we've talked through chat, GPT is kind of like the main player in the space right now for AI, so. And chat shipt is owned by the OpenAI organization. One of the major investors into OpenAI is Microsoft, to the tune of over $10 billion, which is, that's a lot of money. Microsoft's tool is called Copilot. Microsoft, being Microsoft, is as confusing as ever. They have about 15 different copilots, and likely they will have 20 or 30 before the end of the year. Those copilots can do a lot of different things. It happens to be the area that we are the most focused internally because we're a fully Microsoft shop. And then one of the other big players in that is going to be Google. With a Google Bard, like, those are the ones that I'm going to see primarily from like a business use case. You start looking at the, like the unique individual tools. So if you have a line of business app or an ERP, your vendor is probably trying to figure out how to charge you more by baking AI into it. So you can start asking data of it easier or asking information of it easier. Those tools might be really worthwhile. But at the same time, they're going to be a lot, they're probably going to be slower and they're probably going to be less robust than the top three or four players. And actually, Apple just released a really good open source, one that I haven't played with, so I can't speak to it at all. But I think they're really trying to change, change some stuff up with it. And the fact that they open sourced the project is pretty big deal.

Oh, just to interject here, just so people understand what we're talking about with AI, all of these, the co pilot, the bard, the chat, GPT, they're all based on the idea of a personal assistant. So you guys have seen the Iron man movies with Jarvis sitting there talking about, talking to his personal assistant, saying, jarvis, I want you to go, please send this letter to so and so. I want you to turn my temperature to this. And we all have had a taste of the simple functions of Jarvis with like Alexa or the Google nests in our home, those devices to tell simple commands of what to do. What they're doing is they're putting the intelligence behind it so you can communicate with it, free English so you can ask it a complex sentence. I can't go to my alexa and ask if it can pull up information of research papers on something this can, you can absolutely do it. So if you're trying to research something you can pull up. I'm in my hobby outside of work, I'm a fish expert. If I'm wanting to look up something on the fly quickly, the thought of process would be, I could say, hey, Alexa, tell me something about this fish species. And they just say, sorry, here's a link, check your phone. Whereas a copilot, a Google bard, or a chat GPT could actually give me. Here's this, here's the, here's what we know about it, here's a lot more information, and then here's resources to go about it as though you had a personal assistant do that for you. We gave you a couple examples before, but what we're trying to step our toes into is that's how you get into AI, is showing how you, an individual, can use as a personal assistant, but then how you can change these, integrate them into your business tools. So when we're speaking about, you know, incorporating or having some sort of add on to your line of business app of AI, it's, again, most of the time is not near as robust, but there are ways and paths that people can have to connect these Google bards, these copilots and the chat GPTs to them. And that's the forefront of what etop is trying to get on top of at the moment.

First off, Google, you. I just had to. Okay, sorry.

Get it out your system, man. Get out.

Get it out of my system.

I saw when I said Google, you threw like emojis across the screen of pickles. So go you.

Exactly. Exactly. It's Altarco. So to that point, that's part of the value that Microsoft's copilot brings. They have a user based one that is actually pretty, pretty robust. And so basically it's a combination of chat GPT Pro and an integration into your 365 tenant. And so what that means, 365 tenant means a lot to us, but basically to you, it's a kind of like a filing cabinet where all of your Microsoft data lives, for lack of a better word. You might have email, you might have teams messages, you might have like, documents and SharePoint, you might have power automate apps. Like, you might have all, you have all of this different data. You might have a PowerPoint that you made for a presentation. Chat GPT, or copilot can make the PowerPoint for you now. You can feed it information and it can put out like 15 or 20 pages of content for you that you can just go back and edit.

In case the bubble hasn't gone off. I'm going to use the. Use a actual case example. I was using chat GPT to my own discretion. I knew that BJ's birthday was coming up, and I knew that we just went through this wonderful HR training at work that we have to do. It's mandatory, the state of California. And the HR training says that we are not long allowed to call each other boomers anymore. Not that I ever did call BJ a boomer, but let's just say that.

Once you find out, you couldn't.

Yeah.

What?

You can't. You gotta find ways of saying it without saying it now. So I literally went to chat GPT and said I need to find a gift to tell my it boss that he's a boomer without ever saying the word. And it gave me some ideas. And the best idea it gave me is of course, saying that it could find an old piece of technology. And then I said, hey, what ideas for old piece of technology is as a good gag gift. And of course it gave me a laundry list. And of course I picked it, populated an Amazon link to a lovely rollerball mouse, because chat GPT can publicly see Amazon's databases. And it pulled that repository, gave me the link to Amazon and asked me if it wanted to connect my account, order it for me, the whole thing, because it was connected to that. Now what it can't do is it can't create that. Put it on a spreadsheet, put on an order form and a requisition and send it to the requisition department because it's not part of my office 365. It can't create that word document. It can't create that excel file. It can't add it to my expenses because it's not part of that ecosystem. It sure could have had I had the copilot license, had I had a part of it and had I had it integrated with my system, it would have the capability to see inside my internal private secured company files. Had I paid for that license. Well, that's a silly method, but I'm showing you that what it can and can't do based upon what it can.

See, I literally have a, like a 1980s trackball mouse on my desk.

You're welcome.

And memory of right next to my duck call.

I did not call you a boomer. For the record.

You may as well have probably feelings less, but being a millennial, I start out entitled and kind of right.

You need to take it down a peg.

Hurt on the inside, but yeah. Okay, so, sorry, way distracted here. But again, so I was referencing the conversation I had with our CPA, or one of our, one of the team members at our CPA earlier today, who's both a client and our CPA. We had that conversation with him, but I showed him in copilot. I was like, let me go find a document here. So I went into the copilot dot Microsoft.com, and I was like, tell me what I paid in taxes in 2023. And it went and found that my partner and I are married. And so we happen to have a sharepoint that's tied to only the two of us that we call the executive site, and it keeps our information there. Should we keep our personal taxes there? Probably not. But, like, we own the company. We're married. Whatever. It's, it's our, it's our thing. It pulled it up. It was able to summarize how much we paid in taxes last year from a simple prompt.

And how long did that take?

20 seconds?

And that would have normally been a request that had been like, hey, can you find that out a couple days later? Been like, here's what I think it is. We're still compiling it to be sure it would.

It would have taken me at least 20, probably ten minutes, 15 minutes to go find it in the folder, much less. Yeah, it would have just taken me forever. So you want to upload a PDF into it and then ask it questions? Sure.

Now. And equate that to sales calls. You know, I'm looking for all of the so and so's that I've sold in the last three years in this county. Boom, it's there. So it's suddenly I would have to have an intern look for four days to audit all of that, to bring that record to me so I can go have the sales people work on it. You know, insert that anywhere. Now, it's only as good as your data, but I guarantee you it's going to find that data way faster.

Well, and so that's exactly it. Well, so these are ways that you can use it to improve customer experience. These are ways that you can use it to leverage, like, data analysis. These are ways that you can leverage it to prevent the need of repetitive tasks. So, like, if you're one of his questions was, can I use it to write an email that I would send to all of my, all of the, all of our clients about needing this, this, and this? And I was like, sure, why write that yourself? You literally go in with a prompt. I need to write this email to these clients, and we need this information back. Blah, blah, blah, blah, blah. And I basically gave it that information, and it wrote out a perfectly formatted, well written email. He probably would have had to make, like, one tweak to it instead of spending an hour writing an email. It took 20 seconds.

Boom.

Right? Like, that's. That's huge to me. I show him. I show. So, like, one of the biggest things that I use copilot for is honestly, my, all of my meetings, I'm in freaking teams meetings all the time. So as long as I have recording and transcript on, I'll get to the end of the meeting and I'll be like, give me a recap of our discussion and give me list of the key action points that we need to follow up on. I'll get to the end of the meeting. I have all the information. I can pop it into an email and send it. I didn't have to take notes. I'll typically take some offline notes just because it helps my mind remember stuff. But, like, I'll get to the end of the meeting and be like, okay, now this is an accurate summary and representation of what the meeting was about. Here's the things that we decided on, and here's here's the key takeaways. And all of that was done in, like, 30 seconds.

What was that hilarious deal that took notes, but it also got, like, our side banter conversation at the end of the meeting for the last group meeting. It made my day so, so great.

We're like, please do pilot. Oh, yeah. Lord have mercy.

Please give a recap. And then it. It went through in articulate detail all of the actionable items of the meeting, but then it grabbed, like, the one actual items of, like, it wasn't something that super negative, like, go jump off a bridge, but it was basically like, rob me fat. That's what it was. And then Rob's made notes about how you called me.

It was my fat soul. That's what it was.

Yes, about BJ's delightful fat soul. And then that was, like, the underlining last bullet point of the actual nice professional meeting. So, yeah, it picks up on everything because it really capsulates each part and then even flags which ones were thought as just a water cooler conversation. So it does a very nice job of categorizing and making it a bit less robotic than you'd assume.

Well, and that's it comes down to, it's like, is it going to immediately plug into and help you get access to all of your tools? No, it absolutely isn't. I wish it. I wish it did. And are there ways to make that happen? Yes, but they're usually going to be pretty expensive. It's like, how can you leverage it in all of the other areas of your life to save you time, to make you sound more professional, to help you iterate through ideas faster. Like, that's one of the biggest things I use it for, is to iterate through ideas. Like, I'm trying to build a website for this, this, and this. I need my content to sound kind of like that. And so for me, it really prevents the blank slate situation where, like, I'm trying to come up with something. Like, I usually know what I want to say, but I don't know how to say it. So I will come up with, like, my half a dozen prompts for it and let it go, and then I'll go through and read it and be like, okay, this fits these three things really well. Like, and then I'll just start piecing it together. And so now I'm editing versus making, which, for me, works really well. Like, that fits my personality and my, like, how I like to work really well.

And let's be real. If you're a professional and you're doing different presentations, you're trying to do some different education, or you're just doing documentation you've already made, you're already an expert in that field and that details of what you're doing, you're just trying to get through the busy work of getting it formatted in the first place.

There's so much to it. So that's part of what I like about copilot over chat GPT is. Here's one of the main concerns about chat GPT. If you're using the free model, it's absolutely using your data to build its model, which can leak out like, that can other people can come up, ask similar questions to what you've, the data you've put in, and it can take it and it will return your data to them. The chat GPT Pro, I think, has more walls on it where it still will use your data to train it, but it kind of privacy policies it off a little bit. What I like about Copilot is it keeps all of your information inside of your file cabinet. And that's a really huge data privacy and security issue that I'm like, I'll run things through Copilot that I won't run through chat GPT simply because it's like, chat GPT is a little more creative and a little bit, it's got the more current models than Microsoft, but, like, I won't run client data through it because, like, there's no guardrails and, like, I need guardrails when it comes to security. So that's a big part of why I like Copilot so much, because I know, I know the data that's in it is mine or, or the company's, if you will.

The safety rule of thumb is, well, how you would use and what, which one you would use is, would you talk to a third party that isn't signed or doing business with that doesn't have an NDA and just willy nilly, someone off the street and start bouncing internal, you know, company functions off of them. No, they might be really good at it, but you certainly like to have paperwork done with these people or relationship built with these people beforehand, understanding the security risk with them. We don't expect any different from the tool that you're asking those exact same questions from. Why would you use a free tool to do it?

Fortunately, yeah, expect less. Yeah, we know chat GPT is going to be using your training to feed the model because that's part of how you grow the model. But at the same time, it can be really scary to think about, right? So like, the challenge I'm trying to solve internally is how do we start using, how do we start feeding AI all of our databases? Because like that sounds scary, kind of. But at the same time, like, think about this from my perspective. If I could feed it our ticketing system, if I could feed it our documentation system, if I could feed it our client documentation systems, like if I could feed it all of these different systems and somebody emails in, I'm having a problem with quickbooks with this error message. I want it to be able to go through and see, okay, we've seen this error before three times for this client. Here are the links to the direct tickets. Here's a summary of what was done in the last those three tickets to fix the problem. Here's a link to documentation both on the web and in our documentation platform on fixes, unknown fixes for this problem. Talk about empowering. And here's the computer that they did it from. And here's the server that it's on and have direct links to everything.

And here's the last three tickets that you fixed and showing the results on it.

Bingo. You know, like this, this is, this is where it's going. It will happen someday soon. I don't know how long it will take. So it's, again, it's making sure that there's enough guardrails, making sure you know what's happening to and where the data is going. But I see this happening for us probably within under a year. I'm hoping faster than that. But like I, it's, it's very, it's a lot of development time. Like I'm going to be putting somebody very smart, like spending a lot of their time to make that happen. Probably several somebody's, to be perfectly honest. And so it's like I need to make sure that the ROI of it's worth it.

And I think that gets to the menial task portion. So we've talked about how it can be the personal assistant. It can definitely populate data if you're looking for some sort of targeted audience. We've mentioned sales before. This can be done for all your different reporting issues that you need for business to grab direction. But repetitive tasks are huge. So in ours, tickets, we're it. So help desk tickets are the repetitive nature. We're going to get those in you mentioned quickbooks and that shows the repetitive for us. What's the repetitive task in your business? You know, is it going to be the ingest of work? Is it going to be the orders that come in and automated the bundling some sort of order system? Is it going to be inventory and trying to find ways to automate the ins outs of inventory? What are the things that you can free up your employees from to go do the customer service elements that the AI can't do to go make the sales of the AI can't do to go build the business that you already are struggling to find employees for? How can you get more time out of the existing people you have without hiring anybody else?

Well, and that's so much of it. It's like, how do you creatively and thoughtfully tie this in? So something we've been preaching on our, on the new company side that we do, that's process automation, is you can't automate something that isn't, you don't have a process for. And you can automate a bad, convert a bad process, but can you make it a good process and then make it automatable? Like, so the more you start thinking about, like the, the process of what you're trying to automate the outcomes of what you're trying to leverage AI for, think about the end goal as much as possible. Like, what would you like in your head? Like, what's the grand vision that you're trying to solve with AI? And it's like, and then let's scale it down to, like, what's something that, what's something that covers the cost of using it? So, okay, if it costs you $500 a year to use AI, like, what's, like, what's the first thing that you can use it for to save you that $500 in time or make you $500 and like, what's that, what's that leverage point? And try to, try to get to that as fast as possible. For me, it was meetings. Like, the amount of time and money I save, like, not having to recap meetings, is easily worth the $400 a year.

Is that the, I was going to ask what the price points are because that's going to be a question for the audience. So what does copilot sit at? Because you said there's 20 skus. What's the average skew? Is it $400 a year per person?

The main sku that we see is going to be the copilot per person. And it's dollar 360 per user per year. They don't have a monthly sku for it yet. I anticipate it'll be $30 a month per user.

So that's pretty easy to write an ROI off, that's for sure. At least for those intro use cases.

Well, if you could save a couple of hours a month on meetings like that starts being a really valid use case for a lot of your users, especially sales folks. I mean, if you can come out with a clear, you know, next steps, clear recordings and all of that, that starts being a really big deal.

The alter ego of that is people say, oh, it's saving time, but it's also a loss of jobs. And there's very few scenarios where I've ever had someone say, okay, show me an example of a loss of job that AI is going to kill. You know, the next couple years. It will happen. But show me in your small business right now of what enabling something like this would do for to kill a job. That person should be ready and able to handle something else in your company that they already were doing. Partly you're just freeing them up for real responsibilities is the real low hanging fruit. And I see that time and time again, oh, we're going to eliminate jobs. I feel like you're just clicking on the news outlets term of what AI is going to do in 30, 40 years. Right now, all we're trying to do is trying to figure out how to make this business float and grow. And right now we're having a employee crisis. In most businesses, it's hard to find good people to fit in a lot of different positions, but inevitably there's going to be a handful. And I know that in my neck of the woods in the midwest, graphic designers are some of the ones that have been hit. We were having a conversation before the podcast on this.

Two people, two people, like literally the one PhD graphic designer who's terrible got replaced by AI. Like you said, he got replaced by your goldfish.

So again, he was replaced, was the perfect example because there wasn't something else that he wanted to do. There are capable people in your company that if they're that one, one menial task of them having to generate the reports, of them having to pull up the meeting notes for you, of them essentially being the assistant to you, those people are still very valuable, probably more valuable doing something else with more oomph and brainpower.

What I see most of the business owners I talk to wanting to utilize AI for is not to replace people, but it's to minimize needing to hire new people going forward. It's like, how can we help our existing team be a little bit more effective? Like, can we, can we go 5% more effective or efficient by giving them some extra powerful tools, can we make them maybe 15% more effective? Because it's part of why, like, as a managed service provider, we buy so much software because it keeps us from needing to hire bodies to. To service. Like, if we had. So 25 years ago, being a managed service provider generally meant that you put somebody in somebody's office. Like, if somebody had a problem, you were rolling a truck every single time. Do you know how often we roll trucks to client sites?

Not often.

Not often. So, like, those tools allow my, my team being all over. Like we have people. Like, we're primarily southern California based and Robbie's in Minnesota. One of our teams in Austin. We had somebody in Phoenix. They're now in Redlands. Oddly enough. I work from client offices. I work from my office, I work from home. I work when I'm traveling. Like, there's nowhere that we can't do our job from. Literally the only places in the desert when there's no cell reception. But then my freaking brother in law took his starlink along and we were able to do work in the middle of the darn desert. It was wild that it's not AI, but like, how do you use tools to empower your people and optimize your business? Like, that's. I don't see jobs, I don't see most jobs going away, at least not in the short term. But I think we, I think it's, I think the biggest concern is not trying to figure out how to use it.

The other concerns just to go ahead, go on this because we're talking about devil's advocate. So we'll stay on the subject for a minute, is biasness and fairness that we have here in the list, BJ. So when, what was it, Microsoft's original AI on Twitter that they released a long time ago? That was just an experiment to see if it could emulate stuff that it found online to see if it could generate conversations. What was that called again? I trying to remember something not that old. Definitely not that old. So they put basically it was just like a chat bot. One more time.

I know they've been using Cortana.

Yeah, it was, it was a flavor of Cortana that they released on a Twitter. And it was to emulate conversation where people could like tweet it and they could tweet back. It was kind of a fun deal. They didn't put any restrictions on it to see what type of experiment went. And due to the data accumulation and the amount of trolls feeding it bad information on purpose, of course, it only emulates what it learns from. So the amount of extreme bias and dirty posts that emulated was, was not good. And Microsoft had to take it offline very quickly because a bunch of trolls ruined it. For the rest of us, it probably.

Went offline in about six minutes, let's be honest.

It was, it was a lot longer than that, but it would definitely not. Again, it did enough damage, let's put it that way.

Most of the big AI's are putting a lot of guardrails on stuff like you can't do offensive things. The biggest concern I have is like, it sometimes is limiting things that are not particularly offensive if you're like in the middle of life. But it's like, it feels like there's a lot of conversation controlling going on. And so I try to use it for things that are going to be the most impactful to my life, which tend to be very business oriented and cause very little. Like, it doesn't, it doesn't put guardrails on me.

Right.

But you start having it deal with anything political. Unfortunately, you can make Chad GPT a nazi or, you know, a fascist, or, you know, you can have it, you can very quickly like skew the conversations towards like what you think, like your political view. And they've tried really hard to like put guardrails on it to keep that from happening. But my concern is they've maybe tightened it up a little too much.

But even outside of political biases and other, you know, general biases, this is business. Business in itself can be biased. If you have a person saying you need to invest or you need to conserve, AI is going to pick up on that as well. It's just going to be biased. So you have to know that when you put something in, it's going to base it upon your data, the data that your company has seen in the past, the reports that you've given out, and it's going to build its intelligence off of the data that it shows from your business. This is outside of politics. This is just, you know, should I do something or should I not and give business recommendations. So know that it will show you the data. It is built to try to be as unbiased as possible, show you the data. But if you're asking for the AI to build recommendations upon the data, know that it can be biased based on the information that it's pulling from. That is, that is always going to be an asterisk. No matter what AI platform you're doing is, it's going to only be as good as the data.

It's pulling from, well, to some degree, welcome to everything. Well, computer related. It's literally that the quality of your data is going to very directly impact the quality of everything you feed it crap. So part of the challenge that I face with like Copilot and chat GPT and all this stuff is a lot of their training material comes from Reddit. This is like, there's good stuff there. Don't get me wrong, like, there is a lot of really good stuff, but it's also kind of the toilet of the Internet in some cases. And so, like, you're training AI on kind of like a. Yeah, I don't really have a positive thing to say there. Like, I'm like, that's good enough training. We should be training it on classical works, not on Reddit. Yeah, you're, you're kind of.

Training is from a Internet forum. What do you expect? You're. I mean, it's going to be the open Internet. You're going to get the open Internet. I think that does it for at least the major topics of devil's advocate that we would like to go over. We did talk about the personal assistant automating repetitive tasks, enhanced data analysis, and more importantly, targeted data queries, which is by far my favorite to use in business environments. But I don't think we talk most about the improved customer experience, which I think on the surface, on paper, potentially angers customers the most, which most of the time they didn't ever even knew they were talking to an AI helper.

So actually funny, we'll do a funny AI story about customer service and then we can kind of wrap up. One of my buddies is a total, like, hacker, Ciso. So CiSOs, chief information security officer, nerd. Like, the guy is a hilarious, love him to death shout out. He listens to this podcast a lot.

Of times, can't give you name because you're a security guy.

Yeah, he's a security guy. So I won't, I won't operationally secure. I won't, I won't expose him. He literally got a call from a mortgage like a bank. And the second line was, this is an AI assistant. And he's like, it's an AI assistant. Let's see if I can teach it something. So he literally put it up and started teaching it how to do automation with a, a coding language called Jinja. By the time he got done with a call in an hour. So, like, he spent over an hour on the phone with this chat bot or voice sales call, he had it looping back data and helping him solve problems. It was. It literally was an hour long phone call that he recorded on how to. He hacked their chatbot, basically. I mean, he probably cost him, like, a $100 in, like, voice time because he just kept that thing on the phone so long. But what's happening is it's keeping the entire conversation in its memory, and so he can actually. He's actually teaching it coding on the phone. It was. It was one of the wildest use cases I'd ever seen for, like, a funny sales call AI bot thing.

Did he do anything with the coding? He was teaching it.

Yeah, he had it looping back lists and forms, and he taught it how to give back, like, help him finish the coding for the program that he was building.

Oh, that. That coding. Okay.

Oh, yeah, yeah, yeah, yeah. No, it was. I mean, like, and this is a mortgage chat bot.

They called, like, mortgage chat bot. Okay. That gives us.

He's like. And the best part is, it's really gonna break the next person this freaking company calls, because it's like, let me tell you about Python and Jinja. And they're like, all I wanted to know was, like, is my mortgage too high?

What are the rates right now? Have you heard about your lord and savior, Python?

Yeah, exactly. Have you considered automating your house? What?

Welcome to the forum.

Welcome to the future.

But anyways, the overall customer service aspect, especially for tickets, is huge tickets. They already expect a not necessarily automated response, but an emailed response. An emailed response comes from a team. You don't see that necessarily, that Rob's emailed you back. You saw that etump emailed you back. It's very human, very convincing. Not saying that we're using that in our email right now, but I'm giving you an example of a team email that would seem human. So you say, hey, I see that you're dealing with Quickbooks. And this particular error, this particular error we've seen a lot of different times. If you go here, click this, and click update, it should fix your issue. Please email us back if you need any further support. Hey, that worked. Thanks, guys. You did a great job. All taken care of by a chat. Uh, the essential AI chat. Uh, that did all that in an email.

Well, and the funner side of this, like, long term, what I would like to do is start being able to classify our clients as, like, formal response and, like, maybe fun. Like, people that we know have a good. Would be willing to have a little bit of fun with us. And I would love to be able to, like, custom write, you know, we'll tag, we'll classify them inside our system with, you know, different categories or tags or whatever. But, like, whenever we finished a ticket, I would love for it to find a funny meme on the Internet and send it to them. This ticket's done. Okay. And just, like, you send a thumbs up or, you know, send funny, like, send a funny recap email that's like, hey, we did such a good job. Would you fill out our survey? I mean, I don't know, like, come up with. I would love to have funny responses. Being able to be live changed. So every time they got a ticket close from us, it was something unique. Like, I want people to actually want to get our ticket closures because it's, what are they going to send me today? And then there's some people that are just really stick in the mud and they're never getting anything more. Thanks for contacting Etop. Have a nice day.

So fun conversation. I had a very short conversation with someone on the phone, one of our customers, they said that they could not open email attachments. And I said, hi, this is Robbie from Etop Technologies. I'm here to help you with your computer. See if you can't open attachments. Can I jump on your computer? Yes, jump on. Okay, I see you can't. I'm gonna send you an email. I'm gonna call you right back, and we'll see. See if that does it. Cause she was busy at the moment. So I get off the phone, send her an email, and I send her an email. This is a person I knew that was close to you, thank goodness. And I sent a picture of just like, you know, BJ in a goofy face, just like, you know. And of course she sees it and she says, I'm not gonna lie to you. I thought that you were a robot. Your voice was uber professional on the phone. Your responses were extremely, extremely canned. Your professionalism was overdone. You sent the prompt email back and text. But the moment I saw BJ's face, goofy face on my email and I could open up the attachment, I knew my issue was, one, resolved, and two, that you were a human all along, and I was just blown away. I'm like, oh, man, oh, man.

If we don't have goofy, what's the point?

Yeah, so send goofy messages. That helps.

Exactly.

But don't be afraid to at least take, ingest or follow up through with AI. It can save a, save a lot. There's plenty of different tools that you can use, and I know that copilot definitely has those add ons as well, and making sure that they stay secure to your intranet of data.

With all that said, we appreciate you listening if you made it to the end. If you have any specific questions about AI, there's one we might want to do a podcast on. Like actual, like actually integrating into your databases, anything like things outside of Microsoft. That's a lot more technical of a conversation. So we may not actually cover that here just because it's like, if you think this was technical, like, this is like super nerd stuff. And so like, I don't know if that's a good, a good conversation, but if you think it would be valuable, email us podcastoptechnology.com. we'd love to. We'd love to talk to you. Like subscribe, leave us a review, give us a high five, tell us we suck, whatever. Here we are.

And don't be afraid to drop off some Del taco at the office there in Redlands, California.

If you walk in the front door with Del Taco, I'll talk to you for however long you want.

Well, after the first hour, you'll have to subscribe to our AI.

Exactly. Exactly. Like and subscribe.

Like and subscribe for more. Until next week.

Next. Have a great one.

We're doing in one next week, aren't we?

I sure hope so. We better, we better.