AI Basics for Small Businesses

[00:00:00]

Matt: AI is everywhere, and small businesses are starting to use AI. I was looking up some research, doing my research for this episode, and I found sort of a variety of different surveys, uh, that stated how much businesses are using AI. They're using AI here, there, everywhere, and the usage statistics are also here, there, or everywhere.

Some, some surveys come up with low numbers like 15 or 20%. Other ones say 95%. Now, they're all asking different questions, and AI is obviously pretty new on the scene, so that makes sense. But the whole point is, is that AI can absolutely be used to benefit your small business. It can be used to benefit your business just in general, small to medium businesses, large businesses, et cetera.

And so in this episode, we're gonna be zooming in on the small business side of things and further zooming in on the AI basics side of things. So if this sounds interesting to you and you wanna support the show, you can go and check us out on that Patreon, leave a review or rating in [00:01:00] your podcast app, join us in our Discord server, or share this with your friends.

And if you wanna learn how to code, Scrimba has some courses for you, and they have a Scrimba Pro plan that you can save up to 20% off with using our link. That link will be inside of the show description, the RSS description, if you will, and it will also be in the show notes on htmlallthethings.com, where you will find how the link works as well.

The description for that is there as well. So go check it out. They have a really cool interactive media player code editor that allows you to pause the lesson, play with the code live, the code that the instructor was writing and using. You can play with it, break it, you know, kinda answer your own questions by interacting and experimenting, and then you can press play and the lesson just picks up from where it was before.

Go check that out. Again, those links are where I just said they were. Get up to 20% off. So Mike, do you run a small business?

Mikhail: I [00:02:00] guess technically, yes, we do.

Matt: Yes, we

Mikhail: run a

Matt: t- tech- technically, yeah, you know. Yeah, so Mike and I have a, a small, uh, web dev agency, uh, that we run, and, uh, the reason why I just mentioned that is just it's kind of some of, sort of our credentials. We've been in the small business game for a long time, but we've al- so we've been in the game ourselves, but we've also been helping small businesses for a long time as well.

We've been, you know, developing websites, helping them with their online presence, getting them online, doing all kinds of things. And as people who help small businesses and as people who are, you know, in the game themselves, uh, we have found out a bunch of different, uh, ways that AI can benefit small businesses.

So who is this episode for? I've mentioned small business sort of, sort of to death, so obviously small business owners is part of who this episode is for. But it doesn't just end there. We're also talking to people who are solo business owners or freelancers, people who are more, let's say, self-employed than, like, quote unquote, "running a business."

Um, also people who are [00:03:00] working on small teams where everyone wears multiple hats, places like startups or just, again, small businesses. If you're just a part of one of the teams in the small business, you are likely wearing multiple hats, and AI can help you sort of sort out some of those hats. And also someone who has started experimenting with AI but isn't sure how it actually applies to their business.

Whether you're working for a large business, medium business, or a small business, this is an AI basics episode, so this is kinda perfect for that, uh, subset of person, that, that, for that person where, uh, you're maybe dabbling on the personal side of things. Maybe you're getting AI to look at your personal emails or organize your personal emails by, you know, what's junk, what should you answer, what do you have to answer today, what do you have to answer tomorrow based on severity, something like that on the personal side.

And you're like, "How can this thing be used, uh, in my, in my professional life as well?" So again, an AI business-centric episode. Now, I just wanna briefly mention that we have another AI basics [00:04:00] episode as well, so if you're like, "I have absolutely no context for AI and I'm listening to this," we have an- another AI episode for you called AI Basics: Learn These Tips First.

I believe in the timeline, 'cause we re- we pre-record these, it'll be approximately two weeks ago, but you can check it out. I'll have a link in the show notes, um, back to that episode. Now you don't necessarily need to learn or need to listen to that episode. Uh, you don't have to listen to them in order, you know, that one, then this one or anything like that.

But that episode, that AI basics episode, gives you the fundamentals of AI, um, covering fundamentals of, like, learning how to use AI, what prompting is, uh, verifying answers for how accurate or how good they are, things like that. So just, just an FYI in case you're here thinking, like, "I want AI basics information, but not for small business," there's an AI episode for you that we have in our library.

So let's, let's just kick off this episode then. Let's kick off this episode after all the disclaimers and the, you know, the [00:05:00] whose or what's its and all the rest of it. I want to start off with what I think is a very important note right off the hop, and that is that I don't think you should rush into this.

Now, I know that a lot of small businesses are sort of like, you know, you're wearing multiple hats, things are in a fast-paced environment, and so you're gonna want to rush right into it. You're gonna wanna go, go, go. Let's get this, you know, let's get this AI going over here, let's get this AI going over here, let's get this going, that going, because that's what you've been doing with other systems as well.

"Hey, I need to, I need some financial software to get my taxes done. Which one's the best? Okay, we'll try that as long as it fits my budget. Let's get going. Let's go, let's go, let's go." I'm not saying that you necessarily take this slow, but I don't want you to rush specifically. The reason why I say that is because I don't want you to run rampant with ideas on where AI can and should be used in your business.

Here's the thing. Depending on your specific business, depending on your niche, [00:06:00] all the rest of it, you Might have like 50, quite literally 50 places where AI could be used in your small business. If you rush and try to get all 50 of those set up by tomorrow, you're gonna have a bunch of automations that are gonna half broken.

They're creating more work for you than helping, and it just doesn't, it just doesn't help, right? So the whole thing here is do not rush. You do not need to rush into this. Have it, have your AI, uh, introduction from a professional perspective into your business, have it be more natural, because you're gonna start understanding AI a little bit better.

You're gonna start seeing, okay, like, you know, what can I do? What can't I do? What, what can I improve? Like, can I improve my prompting here to get these different results? And slowly but surely, you're gonna come up with some great automations, some great use cases, but more importantly, because you have a more intimate relationship with the software, you're gonna come up with some great [00:07:00] ideas on where it should go in your business, rather than you just looking up like a listicle of the top 10 or the top 50 uses for AI in small businesses, and then you try to apply them all.

For example, we've seen or we've heard of people that have tried to, you know, really implement it into their tax side. Um, I was actually playing golf with a guy recently, a random stranger, and I'll anonymize some details, but he was getting AI to do some of his investments. And he was like, "Yeah, like, I, I want...

I invested in X company, and it, it just for some reason hasn't reached thirteen forty-two." And I asked him what, like, "What do you, what do you mean by that?" And he goes, "I want the, the stock should be going up to thirteen dollars and fifty-two cents or forty-two cents" or whatever it was. And I was like, "Why?"

And he's like, "Oh, that's just the trigger point that AI had me set, and it should have already happened, but I don't understand why it hasn't happened." And so this person is, and this person was a business owner, was just sort of blindly [00:08:00] following the AI's investment advice, g- going kind of i- in that, in that direction of blindly following the advice.

And he also mentioned some business use cases for his AI as well, and I'm just kind of thinking, okay, if he's blindly following the AI with investment advice, he's probably blindly following it on the business side, and, you know, how much success is he really having here?

Mikhail: I, I wanna jump into that right now actually, because this is kind of a little bit of a dangerous path that you can take When, when AI is very useful for you is when you have really good domain knowledge of something. So if you're going down the investing route and you're not an investor, and you give it a generic prompt, it's going to be very close to the median of what it- you can expect. And the problem is, is that you don't know what the median of an investment looks like. A median investor is going to lose money. Like, someone in the baseline is going to lose money. That's how investment works. Like, only the top 10% actually make money on day trading, right? That's about where you [00:09:00] are. And if you don't know that looks like, you're most likely going to get a result that's going to be around the median or below the median.

So when you're doing stuff that could potentially lose you money, be very careful, and you have to, you have to know what to ask it. Now, I'm not saying you can't get a successful result out of AI with investing, but my opinion, really, you need to be kind of a domain expert in the, in the area to then get that extra information out of AI to then put you into that, the, the top 10%.

'Cause you're not asking it, "Hey, invest in the best company right now." You're asking it in specifically like, "Hey, with this EBITDA, with this, you know, with this graph, like, we need to look for companies in this area and this area." So you already have a direction that you're thinking of going, and you're getting AI to execute or do extra research for you on top of that to get in that top 10%, right?

So I w- I, I want you to be very careful when you're going down that route 'cause Matt brings up a really good point there. Um, [00:10:00] on the other si- on the other side of it, when you are a domain expert, that's when you can start to get the most out of the AI that specific space. Or you don't even have to be a domain expert, you just have to know what you're doing. For admin tasks, for example, you, you know, you've done your, you've cleared your email out a million times, right? Like, that's, you've done that a million times. So when you're asking AI to help you, asking it with al- the already known knowledge of what an email, you know, filtering system you would want to look like is. You're not blindly asking it to go in and delete all of your emails or, you know, help you... Y- Your, your question isn't going to be phrased as, "Help me with my emails." It's too general. You're gonna get, you're gonna get garbage. Like, it's just gonna go through and give you garbage output. Your question is gonna be, "Hey, I get, you know, leads from X, Y, and Z.

I need you to sort them out into which leads are the top ones that I should call tomorrow morning from this metric, from these metrics. This is how I've done it." And you give it those specific details that you already gained from your domain [00:11:00] knowledge, and it'll do a pretty damn good job at automating that out for you and save you a bunch of time. So I highly recommend, especially when you're first starting out, lean on your domain knowledge, right? Like you, what you're good at, enhance that with AI rather than going down the route of figuring out how to use AI on something you have no idea about. It's doable. Like, I- the other thing is, like, you can ramp up in stuff a lot faster on that side, but you have to develop trust, and you have to understand how prompting works, and you have to get to the point where you You know what bad output looks like and what good output looks like, and without using AI at all, you're not gonna know that.

So that's something that happens later. That's something you don't tackle on day one. That's something you tackle on, you know, a, a few- after a couple months of using AI

Matt: I, I think the domain knowledge too, like it makes a lot of sense. Like, uh, this is outside of business, but the principle can be applied is I, uh, I have to do a, a, a repair on my vehicle. I have to do, um, a, a clear coat repair. And so I, I wanted to research it. And so, I mean, naturally, it's the age of AI, I could just type in what [00:12:00] process do I do?

What do I do? I didn't do that. What I did is I did the traditional route, and I went and I watched some YouTube videos. There's differing opinions, of course. People have differing opinions of different products and things to use. So I kinda got a, a median amount of information, right? Sort of like, okay, roughly everyone says to sand this way.

Roughly everyone says to use, you know, a product of this type. The brand, you know, everyone has their brand loyalties and things. But, uh, you know, I'm kinda getting a gist of what to do. And then I read some blog posts, read some forums, and then I went to the AI and said, "Okay, I wanna do this." And then I was able to pick apart its answer, and I was able to say, "Hey, I, I was told only do this once."

And it says, "Oh yeah, like, you only... You know, in, in this particular case, maybe you do, would only have to do this once," and things like that, right? Because, I mean, there's nuance to any, uh, prompt. Like, am I doing this repair for showing my car in car shows? No, I'm trying to protect the metal. So if it comes out and suddenly the, the clear coat's a different [00:13:00] color or it's too glossy, I don't care.

I'm trying to protect the metal. I don't care what it looks like. Uh, there, 'cause there's, there's concerns over if the paint has, like, UV problems, if it's been UV'd, or if this is, or that's gonna leave a gray haze. Do not care, only want the metal protected, right? So the idea there is, the idea there is, like, I'm able to pick apart the AI and then make it so that it's not giving me a f- you know, a 57 per- s- step procedure where I'm disassembling the entire vehicle.

It's, it's learning, okay, Matt only wants to do this. My point is is now with that context, I can still pu- poke holes in it, where it'll say, "Use this sandpaper," and I'll say, "Oh, that's interesting. This other thing said to use this." And then we can have a dialogue on what sandpaper's best. And that leans right into what Mike was saying, is that, like, having a bit of domain knowledge, and I'm not, you know, a bodyman or whatever they call themselves now, um, panel beaters I've heard or something.

Who knows? But the [00:14:00] people who work in auto body, the whole, the whole point is, is I'm not in that profession. I'm just doing a little DIY job, l- kind of learning a new skill. You know, if it messes up, it messes up. It's not a big deal. But I would... You know, it's still a couple hundred dollars worth of supplies.

I'd still like to try to get my money's worth or do my due diligence, and wouldn't you want to do the same thing with your business? You know? Um, and I, I think, I think a lot of the rushing in and a lot of the sort of, I'm gonna call it, like, like, the panic around AI is, is FOMO. I don't know if you would agree with that, Mike, but it's like people kind of are like, "Oh my God, I'm missing out on all this AI stuff.

Oh my God. Oh my God," and they're kind of, like, rushing in trying to ensure that they get the most out of their AI, 'cause everyone else is surpassing them and everyone else is passing them, right? That's, that's one of the, one of the concerns.

Mikhail: Yeah, absolutely

Matt: Well, I have some numbers here to maybe, to maybe to quell your, to quell your worries.

So I, these numbers come from s- come right from s- from Stats Canada. This is, uh, from a recent report. [00:15:00] This is actually, uh... It says here, "The release date of this analysis in brief was June 11th, 2026." So it is recent. Of course, the AI world moves quick. But anyway, Statistics Canada found that 19.2%, n- only 19.2%, I wanna iterate, of Canadian businesses used AI to produce goods or deliver services in the previous 12 months in Q2 2026.

Now, that is up. That's up from around 6.1% in Q2 of 2024, so that's roughly tripling in two years, but it's roughly tripling from a very low percentage. So this is, I would say, the time to start researching, understanding, and listening to episodes like this, getting your bearings, you know, getting your under- you know, get your understanding up so that you can either learn that you wanna use AI now, use it later, or just stay informed and you're still watching, but you're an informed watcher on the sidelines.

Among AI, am- among businesses, excuse me, already using AI, [00:16:00] Stats Can, uh, found some of the most common applications were data analytics at 36.6%, text analytics at 34.5%, um, and virtual agent/chatbots at 28.2%. Uh, so that's, uh... You know, we're, we're not talking about, you know, 100% of businesses are doing things, and they're, they're...

It's every single piece of their business, and it's writing all their emails, and it's writing all their, their sales reports, and it's doing all their quoting, and they're doing all this, and if you don't do it, you know, you're in trouble. That's not, that's not what's happening here. Now, another Canadian, uh, survey found, this is from the Office of the Privacy Commissioner, if you will.

I have links to all this stuff inside of the show notes. But the point is, is another, this, this other Canadian survey found that most common uses were research/document drafting at 45%, uh, marketing at 24%, text/data analysis at 18%, and customer service/chatbots at 15%. So again, some, some decent numbers, but this is, I think, the time to learn.

If you've been [00:17:00] ignoring AI, waiting on the sidelines, this is the time to learn, 'cause people are starting to use it. And I, I would drop the FOMO. Drop the FOMO and start the learning is, is where I think we're at with the- with numbers like this, in my opinion anyway.

Mikhail: Yeah. And, and it's a good, it's a good point to make. The FOMO can make you go down the rabbit hole of using too much right away, and then disregarding it as well. That's the other, th- th- that's the other really dangerous part, is that when you start using it, you'll, you'll notice it can do all those things, and all of a sudden it'll make a mistake. And I mentioned this in the previous episode too, the AI basics episode that we've had. But what I really want you to do is get past the mistakes. 'Cause first of all, verify. We'll, we'll have a, a part, a, we'll make... You know, we're talking about that as well in the episode. But you have to verify the outputs, and when the outputs are wrong, that doesn't mean that the tool is useless, and I think a lot of people bounce out at that route. So it's important to understand that, hey, these things can make mistakes. You, [00:18:00] whether it's your fault by prompting it wrong, whether it's its fault by hallucinating. They still hallucinate sometimes. It's gotten better than it was when it first came out,

Matt: Oh, man.

Mikhail: Yeah, it's

Matt: At le- at least it hallucinates in the same niche. I've had things start talking to me about surfboards, and I was like, "Wait, hold up here. We're talking about Ratchet & Clank the video game. Now we're talking about surfboards?"

Mikhail: Yeah. Yeah. Like it's, it was crazy at the beginning. So again, when you're coming back to it now, it's going to make mistakes, but you have to take those mistakes as they are, they're mistakes in the process, and try to figure out ways to correct them. And a- as you start to figure out ways to, hey, how do you eliminate mistakes, you'll start to figure out what's really important, which is verification. And I'm not saying you have to verify every, every output that the AI comes at you as well. That's important too, because that will make almost more work than what it's saving. What you have to figure out how to do is how to have, like, the right feedback loop. Like, right, how to get AI to almost verify itself. And you can do this in the prompt, you can do this with a separate agent, you can do this w- how- a million different ways. I don't think we're... This, like, the [00:19:00] scope of this episode isn't to give you the entire toolkit how you can do every part of the, of, of the AI workflow that you're gonna implement.

It's that these workflows exist, and you need to start figuring out which parts work for you, which parts don't, and how to make the parts that don't work better,

Matt: Yes

Mikhail: all of it, all of it can be done, but you need to start slow. You need to start deliberately, like, on stuff that you know is gonna benefit you, and you need to iterate on it as you go, right?

Like, you're gonna learn things as you learn, uh, as, as you learn, like, the next step. You're gonna be like, "Oh, I could implement that back into this workflow that I developed before," right? Like, these emails can be actually organized even better because I've learned how to, you know, verify the output a little bit better.

So continually feed that back. You'll make your processes better. Allow mistakes to happen. And definitely don't get dragged into the FOMO of, like, running a million agents at once, 'cause that's gonna generate really crappy output. It's gonna burn you out from having to manage all these things, 'cause you're still [00:20:00] the bottleneck of everything, right? And it's gonna give you a bad taste of AI at the end of the day, which I get it, like, to a certain degree, but also, this is the way forward for most business, in my opinion. So you need to get past that, and you need to learn how it can benefit you and how can it save you time and money

Matt: Yeah, and, and like you were touching on, there's a bunch of different workflows and things like, you know, it self-verifying and whatnot. But that's more at the end of the rainbow, I think, uh, like in terms of your AI journey, and, like, this episode is gonna focus on the beginning to get that ball rolling, and then eventually you will be doing the things that Mike w- Mike is saying.

Because, again, once you start getting these intricate, uh, understanding and ideas and, um, like the use case, like a-applying different use cases 'cause now you understand how these things work, you might be like, "Oh, I can have another thing come in here and just quickly check it." And then maybe there'll be like a re- maybe it'll re-prompt it or something for me, and then I'm getting like a vetted result.

Even though it's not human vetted, it's like now I'll, now I'll [00:21:00] vet a vetted result, so now there's a, you know, a, an 80% chance that it's, that it, it's a, a, a good response to me rather than only a 70% chance or something. Uh, those numbers I just pulled out of thin air, but just as an example. Uh, and I, I, I think a good, I think a good place to start for any small business is not to just, you know, jump in and look at all the consumer, and many of them have business versions, um, of the various bots out there like ChatGPT, Claude, Gemini, Copilot, you know, the list goes on and on.

What I think you actually should start with is the business itself. I think you should be starting to ask yourself, you know, what is it that we do repeatedly? Do we constantly have a backup processing emails from our contact form? Okay. You know, that's something that we do repeatedly, and it is also a problem.

Um, and what do we do repeatedly that isn't a problem, but we do repeatedly, so we're spending two hours a day, maybe a whole day a week on doing this thing? [00:22:00] Okay, that's something to consider. Maybe it... We can hand that to an AI. What takes too long? Is there something that, you know, you just know that you're getting bogged down with?

It's taking you too long. It's taking you away from other things that make more revenue or, or are more important. Okay. You know, list that down. What do we constantly put off for one reason or another? I hate doing my taxes. I don't like doing this. I don't like doing the auditing. I don't like doing the invoicing.

Okay, you know, take a look at that, write that down, add it to the list. What work do we hate doing? And go through this list of questions. I have more questions that'll be in the show notes as well to ask yourself. The idea here, though, is to kind of go in and be like, "Let's start with the business and see what are things that we maybe would want automated."

Because it's gonna be different per business. There's gonna be people, like for example, when we first started this show, uh, we didn't have editors or anything like that back in the day, and like, I was spending a lot of time getting the audio engineering just right. Now, I was still [00:23:00] DIYing it. You know, I'm sure that the, that a professional editor, a professional music person, music mixer, whatever, audio engineer could have come up with a much better result, but I was making s- like it was kind of like my baby, and I was like babying that.

There wasn't AI back then, but back then, if you had asked me, "Hey, an AI could edit the audio," I would've told you to go away. No, I, I edit the audio. What else can the AI do to me? The reason why I mention that is because there's gonna be parts of your business that you baby, that you treat as like part of, like, you, you know, you really wanna learn this scale, or you think that this is your, your moat or your competitive advantage.

You don't want an AI reaching out to the customer after, after a case. Like you, you do something for them, whatever service. Y- you don't want an AI reaching out. You reach out, and you think that that's really valuable that you reach out. If that's the case, then that's off the list. You know? Like, a- and so go through that list, make that list of things where you think, you know, where could we potentially use AI?

And then once you make that list and you're like, "Okay, these are things [00:24:00] that I potentially, uh, would be okay with AI doing or an automated workflow doing or things like that," then you can start going through and saying, "Okay, h- can AI actually be used for these things?" 'Cause some of the things it's gonna be no.

Do we have humanoid robots at the, you know, the price point that a small business can afford and is readily available? No. Okay. So this, this, and this have to go. Get that out of here, right? But it's like, okay, you know, these five things on my list, maybe, uh, my social media posts. I have a very, you know, a very generic social media post that I do once a month that quite literally just lists the statistics in some report, and I post it every month It's not personable, it's not an editorial thing, it's very robotic.

AI can maybe able to tackle that. Now, that's one small thing, but the point is, is you're starting to get the ball rolling. You're starting to understand where things go. And you can come up with multiple of these too. Like I said, make a list of these questions and make a list of the answers. Like, you know, what do we do repeatedly?

What do we do [00:25:00] this and that, this and that? But don't just say like, "Okay, now we have a list of 50 things, and we know that 50-- that f- out of 50, 45 of them could be AI could do. Let's go do all 45." No, we're still taking this slowly. We're choosing one or two, right? And we're going through. What I like to say to do for small businesses is I like to say automate high-frequency, low-risk work first.

You are learning how to use the AI, you're also seeing how the AI is going to work in your particular niche. And so what I like to do is I like to take a look at the things that you've listed originally, and you say, "Okay, I do this task regularly, right? It consumes meaningful time. I've determined that the AI can do it and is reasonably good at it through my own research," and not asking the AI if it's good at it.

Doing your own research, looking up some YouTube videos, reading some blog posts, looking at some forums, looking at Reddit, things like that. You can also ask the AI, but that should be part of it. That shouldn't be all of your research. Ask the AI if it's good at something, of course, it thinks it's gonna be good at it.

Of [00:26:00] course, right? Make it something that you can easily check the result, and make it something where a mistake that the AI does, and hallucination, as Mike stated, and hallucination is not catastrophic. So, for example, maybe you thi- maybe you think, "Okay, I'm getting 50 responses via a contact form a day.

It's a lot. So what do I, what, what do I have to do? Well, I have to answer all 50 of these responses. I do this regularly, it takes a lot of time, but a lot of my responses are very, like, cookie cutter. Like, I'm sending people reports. I'm, I'm answering, uh, low-hanging fruit. I'm, I'm answer-- I'm, I'm saying, you know, "Well, yeah, we'll, um..."

You know, like, "Yeah, like, yeah, we'll post that on our, on our website, you know, thanks for telling us that there's a typo there," or, like, whatever, right? Like, these are kind of low-key responses. So that's an example of something that you do all the time. So what, what, what, what might you have AI do? Read those emails and draft, here's the key, [00:27:00] draft up responses to them.

Then you go in and in this first step, you, you review them. "Yep, you're right." Like, you know, like, that's a good response, send. That's a good response, send. That's a good response, send. Wait, that one's not good. Now, don't just quit your AI usage, like Mike said, just hang on. Go in, fix it, potentially tell the AI, "Hey, don't answer like that.

Answer like this." Now continue along, continue along w- with what you're doing, right? Next, send, send, send, et cetera. Other things that are really common is gonna be something like meeting summaries. There's people that will, you know, sit in on meetings or maybe they're the people that are in the meetings themselves, and they'll sit there and they'll just take notes, like, kinda like a court secretary where they'll, like, take notes on, like, what's going on, right?

Kinda writing down, like, w- what, who said what, and why we, why we're, why we were meeting about certain things and what, what went off the rails and why, what needs to have another meeting made for it, things like that. There's tons of meeting note takers out there. Notion has one, Zoom has [00:28:00] one. There's a ton.

I believe Copilot has one. It's all over the place. You can absolutely... And every time I j- join a meeting with a, with a stranger for, like, sales or what have you, a different note taker shows up. There's a ton of these note takers, note takers here. The point of the matter is, if you're, if it's, if it's a kind of a run-of-the-mill meeting, it's not a high, uh, brow meeting that you absolutely need every bit of information from, and if you accidentally lost the notes it wouldn't be catastrophic to your business, try this AI note taker, right?

Try to automate these high-frequency, low-risk task first to learn how it works, to learn how whether you like it, dislike it, et cetera

Mikhail: Yep

Matt: I think what you, at that point, should start doing is start expanding out. So what you should, so what you should start doing is let's say you start with the customer emails, and then you move to the meeting summaries. Again, get stable with it, understand what's going on, right? Start doing this.

Then maybe you can do other things. Oh, [00:29:00] we do a lot of research for people, and it's just like sort of this person is interested in our business, but they're not a guaranteed lead. There's no money changing hands yet. Maybe we do a little bit of research for them. Let's get the AI to do the little bit of the research.

We'll go in, we'll just, you know, confirm it. Like, tell the, tell the AI, like, this is a key thing. Like, you might, you might tell the AI to do research and you realize, oh shoot, there's no, uh, citations here, so I can't confirm anything the AI is saying. That's like a little lesson where you're like, "Okay, I'm gonna force it to tell me its sources."

Now, when it gives me a research report that I'm gonna use for the customer, you know, for this customer, this, uh, potential lead, w- now you can go into the sources and quickly check and make sure that the statistics and the information that the AI pulled are, was accurate. The report's already written, and now you can send it to the, the staffer or the customer that you need to send it to.

The point is you're slowly starting to understand and slowly starting to put things in motion I think a key thing that you shouldn't let get away from you though, as a small business, is you should not let the AI [00:30:00] create your entire brand. There's a lot of startups or there's a lot of like, you know, people have a business idea and they, they will run rampant, and if they're, if they're versed in AI on the personal side, they're not a business owner right now.

They're using AI in their personal life for this, that, and the other thing. They're excited about AI. Fantastic. They decide they're gonna start a business in whatever industry. "I'm gonna be a small business owner. I'm gonna do this." The AI will absolutely spit out a bunch of stuff for you. It'll create a logo, it'll create a brand identity, it'll create a whole thing.

But like Mike mentioned earlier, a lot of the results that you're getting are middle of the road. It's just looking at it and going like, "SEO is good for welders like this," or, "SEO is good for plumbers like this," assuming that's the small business that you're gonna actually open. And now you have an issue because, like, now you're just middle of the road.

There's no personable part or anything. Now, people will argue with me and say, "Well, you can get the AI to, you know, add that personal touch and things like this." Yes, you can. You can start getting into the weeds like that. But I think that the AI should [00:31:00] be, you know, your assistant, at least in the early days as you start to learn this stuff, and especially when you're setting up the core of your business, especially a small business.

A lot of the time a small business does get sales from being personable. The owner of the business is only a phone call away or a text away. Even if there's 10 or 15 people working there, it's not, there's not eight managers and two directors and then two executives before the CEO. You know, the CEO is, again, a text or phone call away.

It's a personable business, and it's part of what a small business moat... Some, some small businesses use the access to their, their brass, if you will, to the higher-ups as part of their moat. There's gonna be other businesses, especially if you're in B2B sales, that prefer having a personable kind of small business that they work with, and they have instant access to the, the person at the top.

They have a major issue, they're not talking to a manager, they're talking to the person at the top that then goes to the team and tells them what to do, and that's part of a small business's moat, the local [00:32:00] angle, the small business angle, that type of thing. And that's very personable. That's very sort of, uh, human.

It's not AI, is the reason why I mention it. It's not AI. And so I'm okay with you brainstorming with, like, a brand identity and this type of thing with an AI. It should not be the thing that creates your entire brand. I'm very much against that. And pre-AI, I saw people doing this where they would log into, like, a, a, a tool online where it'd be like logo maker and, you know, website maker and this and that, and it would be just like fill in a form or whatever.

And they would get a cookie cutter website, a cookie cutter logo that's, like, exactly the same as another person's logo in their niche, but it's, like, a slightly different color. And I'm sorry to say this, you can tell. But the scrappiness, the part, like, the, the, the part of your website that, you know, maybe is a little incomplete or whatever 'cause you just didn't have time to do it, sometimes that's your moat.

It's a little bit of per- like, personality where it's, okay, this is a small business, and it's, it's legitimately a small business. Like, it's not like there's [00:33:00] garbage everywhere and you're leaving everything untouched, but you don't have 10,000 blog posts like some enterprise sites would to maximize SEO.

You have 5 or 10 or something like that

Mikhail: I would take this a l- a step further and say that, like, you know your differentiators. If you're a really good car mechanic, like if you have really good, you know, the entire car mechanic pipeline is amazing, like you have the best employees and stuff like that, that's your, that's your differentiator, right? That's cool, and you can put that in y- on the website. You can put- you can do, you can do your best. The problem is, is that unless you hire a really good team, it's still, even though it's gonna be personable, it's not gonna be, quote-unquote, "very professional." You, you wanna take your personality, and if you're not gonna hire a very high-end or like a, a decent web developer or a web design agency, shout out Digital Dynasty Design, uh,

Matt: oh

Mikhail: but if you're not gonna hire them, then, and you, and you are gonna go with AI, then take your [00:34:00] brand, take what you're good at, feed that into AI as best you can, look at your competitors, feed them into AI.

Like, there's ways to get a more personal experience from an AI without, again, just being a generic, "Hey, make me a website for a car mechanic," right? Like, you can get a little bit more out of it, and you can make your existing branding look a little bit more professional, still keeping your person- personality in it, right?

So AI does have the capability of leveling you up, as long as you're willing to iterate, as long as you're willing to ask it the right questions and do the, u- use the right models, like, use the right tools, um, to get there. Now, it's not easy to do that. That's not something that you're gonna learn day one of automating AI. And that's another thing that I wanna kind of give you a recommendation on, you might, if you're a small business and you're, you know, you don't have time for leveling up on AI, you might want to reach out to a [00:35:00] agency to help you. I don't know if you have any locally. Like, ideally, you want someone to actually come in to your store and help you figure out these areas that M- Matt was talking about, like the, the lower-hanging fruit, asking the right questions on what can be automated because you're, you might not even know that something can be automated due to the fact that you don't have the idea of what AI can, is capable of. So depending on where you're at, right? If you're, if you're a technical person, wow, like you're, you're ahead of the game right here, right? But if you're not a technical person, and if you're just getting into the idea of technology, and you've really just been, you know, using a computer as like an Excel, Excel spreadsheet thing, then a- You might wanna reach out for some help on this because you can get quite a bit, um, of a boost if someone knowledgeable goes in, looks at your business, and figures out the right workflows to, to, to, to implement, also does it securely as well.

I don't think we're talking about, like, data security here a lot. [00:36:00] I'll argue that it's not particularly important for a small business unless you have some really secretive competitive data that you don't want anyone to, to know about. As long as you're using kind of like a, a paid plan on any of these services, their terms of services state that they do not retain or train on your data. Um, so that can be ex- extended to your customer's data as well,

Matt: I was actually gonna, I was actually just gonna say though, like I would look out for data that isn't yours. 'Cause like many fellow businesses will offer services and it's sometimes like I d- I wouldn't want... If I wouldn't bring up the information at a party or at a conference I was at, then sometimes I, I, I will at the very least ask the question from my, of myself whether I'd want it typed into AI.

And, and even for us, if I, if I mention, uh, [00:37:00] like tax information, I'm not gonna give it like my social, like my social insurance number, social security number, whatever. I'm not gonna do stuff like that. But I'm also... But there's also other things that I wouldn't bring up either. Like, I just wouldn't, I wouldn't be like, "Okay, this is, uh..."

Maybe, maybe I would now, like I've gotten more trustworthy, but like a couple years ago, for example, I wouldn't say, "This is our revenue number." I wouldn't, I, I just wouldn't be giving out that information. Now, I'm comfortable with giving out that information to an AI now about me. But if I'm, but if I'm, if I have a client who is in a highly competitive environment, I wouldn't wanna say, "Oh yeah, this person's doing 125 grand a year."

That might sound great when all their competitors are doing 75. Now all their competitors, if that somehow gets leaked or whatever, are now looking at them and watching what they're doing. Like I, I just wouldn't... It, it, it's definitely sort of like at a paranoia, like a paranoid level, but it's li- it's like a, it's a bit of security through obscurity.

You don't wanna be [00:38:00] dumping information into this thing that you wouldn't say at a party, in my opinion

Mikhail: I, I think that i- i- it is good to have this conversation right now because there is this kind of like gray area, and it's called... Actually, there, there's a term for it. It's called shadow AI, where people in companies will use their own personal subscriptions for doing different work, and

Matt: Of course

Mikhail: data, they'll upload per company data into, into ChatGPT thinking that it's all good to go, and, you know, that's technically not Like, you're not allowed to do that kind of thing.

Matt: Yeah

Mikhail: of people still do, um, and if you're listening to this and you do that, maybe think about it a little bit, right? Like, this is a, this is proprietary information that you do not have the authority to make that decision on, let me be clear. Um, having said that, if you are using a company account or if you, if you're your own small business owner, which is probably where you're at right now listening to [00:39:00] this b- like listening to this podcast, if you purchase a subscription, not a free one.

Again, don't upload anything personal. Don't upload any customer data. Don't upload anything to a free subscription because on a free subscription, they have rights to do with your data as they please. You have no control over that. If you're on a paid subscription, like you pay for ChatGPT, uh, especially if you're on a small business one, that's an even higher tier of security, you have a little bit more flexibility there.

Now, customer data I'd still be very careful of because as we know, anything that we upload on the internet is, uh, susceptible to being hacked. Uh, definitely would not upload social security numbers. Like, that's a good one. Like, that's a very, very, careful thing, right?

Matt: Passwords, API keys, you know, similar things, especially in the dev space, which our listeners are in, a lot of them are in

Mikhail: Absolutely, yeah. So just be very careful with those things. Now, other things like maybe [00:40:00] slightly anonymized customer, know, customer revenue information, as long as you slightly anonymize it to a certain degree, I wouldn't worry too much about that. Um, you might wanna put something into your contract agreement going forward that allows you to do some of that. Uh, again, respecting the services', uh, s- d- data policies, right? So these services, most of them have the data policy of no retained, no retained data. They can't look at it, they can't train on it. They're, you know, ISO compliant, SOC 2 compliant. Like, you, you get their compliance. You can actually look that up for any, you know, ChatGPT, Anthropic, OpenAI, whatever.

You can look up their compliance stuff. So as long as you're comfortable with those compliances, you can start to be a little bit more free with what you use these services for. Because the more free you, the, the more, like, open you are to using these services, like we talked about emails, right? Like organizing emails. If they're reading your [00:41:00] emails, there's a lot of customer data in there. Let's be very

Matt: It's a very, it's a very, very, very good point

Mikhail: Yeah, it, it, it is a, that is a dangerous offering. So when you're doing that, make sure you're using the best service that you're willing to trust with that information. And I think you're going to have to trust it as, like, security conscious as I am, have-- I, I haven't actually opened up my emails, like personal emails to it, honestly, other than obviously my personal emails on Google, so they already have all the information that they could possibly need. But my bu- on the business side, I haven't opened it up, uh, to emails 'cause I'm still a little bit cautious. But having said that, I am in the process of probably gonna... I'm, I'm probably gonna do it because I just see the benefit of it, of it being able to control my emails and filter out the, the junk to a certain degree. Even, like, look in my spam folder and, like, see if there's anything that, you know, the spam folder accidentally put in there, 'cause I've had that happen fairly often [00:42:00] as

Matt: Oh, it happens a lot now, 'cause just 'cause there's such a volume of spam

Mikhail: Exactly. I've missed leads. I've missed important documents that have been sent over my way. I've missed, like, login information.

Like, I've missed so much 'cause I just, I check my spam once a, a couple, every couple weeks, uh, just because I am missing stuff. But if AI is able to look at it... Anyway, I, I see the benefits over, um, starting to creep up to be, like, worth it as long as I'm using the right subscription with the right data policies.

You know what I mean? So you have to consciously understand that you are making that kind of compromise when you're uploading personal data to a certain degree, at some point it starts to become worth it. But definitely never do it on a free subscription, right? Don't do that ever. Don't give it access to your emails on a free subscription.

Only do that on paid subscriptions, and just look... Before you do it, just take a look at what they, uh, the terms of service. You can e- you, you can even ask a different AI if you don't trust the AI, that AI, to read the terms of service, and then you can ask it different questions on,

Matt: Well, do, do your research too. [00:43:00] You can

Mikhail: Yeah. So there is this really delicate balance of, like, giving it information, but also protecting the information of, like, example, your customers and maybe yourself that you don't want leaked. Um, not an easy, not an easy solution. I don't have a concrete, like, do this and do that. This is a very personal thing, but I'm starting to lean towards giving more information on, like, higher subscription tiers especially.

Matt: Well, it, that's like a generally, like, good rule of thumb. A- and the thing is, at the end of the day, like if you wanna be the most precautious person, you should do exactly what you said, Mike. Research the terms of service, look at what the terms of service says, look at what the company's policy is, uh, the AI company's policy is.

Look what their privacy policy is. Like, do your due diligence for security. The free tier, like, like you're, they're, you're probably right. Like the free tier, they're making money on it somehow, they're getting information from it somehow. If you're concerned, look into the free tier, look into the paid tier.

Like do your due diligence across both the free and the [00:44:00] paid tier. Look across multiple companies if you're look, if you're interested in multiple companies. Not that you... I mean, this episode is kind of about how like, you know, you should be introducing and not buying every single freaking service out there, 'cause especially as a small business, we have budgets.

We don't have, you know, $6,000 in tokens here, 12,000 to- tokens there, 34,000 here, $200 a month plan here. Like we don't have those luxuries. That's, that's absurd, especially if you're just a freelancer, like you're just alone. That's absurd. But you know, if you, if you've chosen ChatGPT, for example, do the due diligence.

Which plan should I get? What security measures do I have? That includes what account, what account, uh, security measures you should have too. 'Cause they've offered me... ChatGPT came out and said, a- as like popped up and said like, "Hey, we think that you should have the enhanced security plan." And there's like n- not a plan as in to purchase.

There's like an enha- an, uh, an enhanced security account, like mode or feature that I can turn on, and then it, there's a bunch of things that it does. It logs me out of every single instance [00:45:00] I've done, and it a- adds like another layer of security. Then I can log in everywhere and things like this. There's all these different ways to do it.

Like for example, I just enabled remote access to my computer from the ChatGPT app. I just enabled that That's a choice that I made, and I'm doing it slowly. I still get ChatGPT to, uh, you know, I, I still have to approve a lot of the commands. Slowly that leash will be let out. I'm not saying, like, when, when we say, you know, jump in and get used to AI, it doesn't mean just open the freaking door from a security perspective and just, you know, let everything, let everything fly.

Like, you, you know, definitely do security due diligence. You're 100% right, Mike. It's a good conversation to have, especially for small businesses. Like, if you're a small business that handles people's medical information, you don't wanna be like, "Oh, yeah, that Matt has, like, this terrible rash on his butt," you know?

And then, like, then that leaks, and that's a whole freaking thing, you know? Like, what the heck, you know? That'd be a whole mess. I literally have a point here in the, in the show notes with a, with a quote, uh, here. For example, you can [00:46:00] anonymize details. We say it on the podcast a bunch where we'll, we'll, like, I'll say, "Here's a customer case," and then I'll anonymize details on the fly.

I'll change the numbers. I'll change the person's name. I'll change the industry they're in. But, you know, the, the principle's getting across. So that's an example here. John Smith at 123 Main Street owes us $4,218. Here's his entire customer history. How do I, based on this customer history, you know, get him to pay it?

Well, if that ever leaks, now you know John's in debt. You know where he lives. Like, that's a mess. You don't wanna say that. You might wanna just say this. This is the prompt that, you know, this is an anonymized prompt that you might send out. A customer is 60 days overdue on a $4,200 invoice. Help me draft a polite payment reminder.

M- Same, same principle. Everything's been anonymized. A customer, it's not the exact amount. It's, there's no address in there. That's just an example of ways to do that. And if you s- And if you practice that, then you'll just get good at it. Like, M-Mike and I are fully practiced in [00:47:00] it because of the, 'cause of the podcast and because of, you know, I-IT security back in the past

Mikhail: And th- there's other things you can do on bulk data as well. So like I've done this before where I have like an Excel spreadsheet with a bunch of customer data in it where they have their emails, their addresses, their full names, and a bunch of responses and like, you know, it's like a, it's like an export out of a form, right? And I don't really like uploading that to a ChatGPT because there's just so much customer data in there. So what I'll do is I'll either like using, you know, Excel macros, I'll go in and create a, uh, a reversible approach where like I know if I, if I need to go back to having the customer data there, I'll, I'll be able to understand it.

But essentially it just changes everyone from having their names to customer one, two, three, four, five, six, seven, eight, nine, 10, you know, all the way down the board, deletes their addresses 'cause that's not relevant a lot of times to a, uh, uh, uh, whatever you're asking. If it's relevant, then just remove their like [00:48:00] home address and just keep their, you know, city or whatever. Um, and then removes like their phone numbers and email addresses if they're not relevant. So like I'll, I'll go in and sanitize the data. I'll anonymize the data. And you can write a script to do that. You can ask ChatGPT to write you a script to do that, right? Like there's different ways that you can handle that to make it faster. But I do think that it's important to take these steps a lot of times because, again, uploading something like that to a ChatGPT is first of all, like illegal i- in some st- in some cases. If you're in Europe or something like that, like there's probably some legality that you have to be careful with.

And then depending on the terms of service that you've signed with the customer already, you're probably violating that as well potentially by doing it. So just be careful. Um, make sure that you're, you know, due diligent on that stuff, but also to find ways around it because if you're just gonna be like, "Oh, I'm not gonna use this at all because of that," then again, you're at that point where you're disregarding it and you're gonna be left behind because [00:49:00] your competitor's gonna use it, and they might use it nefariously.

Like they... Not, not nefariously, but they might bypass all the s- the stuff that I'm talking about, and they'll be way faster than you. So you need to find a way to have that balance of being, you know, security conscious, but also getting as much information and as, and as much use case out of these systems

Matt: It, yeah, it's a, it, this is a, this is gonna be a sticking point, I think it is what you're, like is, is what we're talking about. This is gonna be a sticking point for some people. Some people are gonna get that, like get scared and say, "Okay, I'm not gonna use this thing." But, you know, just being smart with your own and your customer's data is extremely important.

But it's also smart to have AI on your side because it's gonna make you, if used properly, more productive, make you more revenue, uh, give you more free time, whatever. Like, there are benefits to it, and so don't just kinda toss the baby out with the bathwater. Just, again, like you said, Mike, find ways to work around it, [00:50:00] because I know for sure there's gonna be...

Like give it, give, you know, give it a small leash. Do these things that I do. Like, it'll, do your due diligence, but don't be like super afraid of it all the time. Like, I feel like some people are gonna get super scared, like, "Oh my God, there's a security concern? Okay, I'm out." Well, I mean, there's a security concern with everything.

Like, what if I shared something to OneDrive and then OneDrive didn't, you know, then there's a whole list of, whole list of nonsense. We live in an interconnected world. Just navigate it as you... Navigate it responsibly is what it, is what they're

Mikhail: could get hacked tomorrow and all of the information that it stores could be out there. So like it's already, you know, like you can't protect the information completely. It's not

Matt: Correct, correct. Do your, just do your due diligence.

Mikhail: the risk. That's it

Matt: And, and on that same thing, I, I have a note here in the show notes about, I think that small businesses should have a basic AI policy if you employ people. If you are... If you employ people, you absolutely should have a basic AI policy. Uh, and the, uh, the complexity of that AI [00:51:00] policy is, you know, starts at basic because you're a small business, and then grows based upon your use, uh, of AI and how many employees you have and things like that, of course, naturally.

I think it should be a live document that changes as your business, the AI itself, 'cause AI is constantly changing, and your use of AI evolves. It should be th- this, this living document. But what it really should do is it should tell or inform your employees what tools are acceptable, what or which ones are not.

Can they use their personal accounts? Like, can they use their personal Copilot account? Uh, can they, uh, can they upload images? Can they upload this? Can they upload that? Can they talk about this? Can they talk about that? I think that that's really important. It's in the same vein, not to get stuck on this point, but I think it is important, especially for staffers, because, I mean, I've...

W- we know people that, you know, work for whatever company, and they're just starting to put in AI policies now, and there's been violations, like left, right, and center before that policy was put in place. It was like, "Hey, [00:52:00] everybody. We must use Copilot, the Copilot that comes with our Microsoft 365 plans."

And it's like, uh-oh, you know, said person has been using ChatGPT on the free plan for, like, nine months, and God knows how much information has gone in and out of that ChatGPT, you know, account. So AI policy, another, another just smart move to have, I think, uh, in this sort of, you know, day and age

I think as a sort of a last sort of message out, I think that you should move your AI usages into workflows. So move off of one-off AI uses and get it into repeatable workflows that you do all the time. So something like if your policy is someone go- you know, gets to the website, fills in the form, you receive an email.

That email, you know, then someone looks at the, the, the DL, the [00:53:00] distributed list. They, they, they g- they go to the place where the, the inbox where the email has been received, and they decide at that point, "Okay, what I wanna do is I need to, you know, send these two emails to finance, these two emails to the manager, these two emails to directors," you know, the list goes on.

That's a repeatable workflow. I would say, like, try to get an AI in there. Try to get an AI in there to, like, do that part, do the filtering, whether or not you need to go in and approve that. But if it's just simply, you know, forwarding emails and everyone's NDA'd, and, like, if someone gets the wrong email, they can just quickly forward it off, and, you know, you're n- you now are, you know, saving somebody two hours a day, an hour a day, 30 minutes a day.

That's just one small example But I think the idea here is, is like... And I, I did, I did write up a, a step-by-step procedure for this. So step one would be, you know, do it manually with AI, right? So, you know, paste the information in, ask the AI for help, and review the results. So in this case, it's like, "Hey, where should this email go?[00:54:00]

Should this email go to finance? Should this go to the directors? Where should it go," right? And then you repeat it. So you do this several times. So maybe for a week or a month, you choose however long you're, that you're comfortable with. You decide, "Okay, I'm gonna do, uh, I'm gonna paste the emails from the email form in to this AI manually every day for 30 days," or something like that, right?

You're gonna slowly start to learn things. You're gonna learn what context it needs, where it starts making m- m- making mistakes, right? Are there some hard line points where you're like, "This always must have the CEO CC'd on it when you forward it. This, like these type of requests must have the CEO. These type of requests must have finance, et cetera."

'Cause sometimes there'll be some niche use cases where it's like this needs to go to the engineer, but it's on a, it's on a, a client that makes so much money that finance has to be read in on it no matter what. You can make those kind of hard lines. You can maybe change your prompting a bit to figure out [00:55:00] where it fails and try to like work around its, you know, its failures.

Slowly but surely, you're gonna work yourself to step three, which is your standardizing. You're starting to standardize this workflow, right? Uh, you are creating reusable instructions. You're creating templates. You're creating prompts. Maybe you just have a list of prompts. I've seen some people on YouTube that will just have these list of prompts that they share out with their subscriber base.

Um, so like that's, that's something that like will absolutely work. You can have... You can, uh, give it, you know, business context and things like this. So you've given it like, like the, the framework of like we answer these emails in a friendly tone. We answer these emails in a more utilitarian tone. These guys are the mechanics and so they only care about the stats and so like, you know, we, we, we answer these with just stats.

We don't have like all the pleasantries and things like this, right? You, you, you, you start standardizing the procedure and now every, now everyone, like if you handed, like if someone else was gonna take over your job, you could hand them the, the set of rules, this set of prompts, these sets of [00:56:00] things, and you could hand...

You could potentially hand it off to somebody, right? Now what you start doing is now you go on and you just pass it to AI. So you, you, you figure out, you know, w- what do I need here? Do I need an agent to do this for me? Do I... Does my CRM have like some sort of endpoint for my AI? And like can my CRM with ChatGPT or with whatever AI it comes with, maybe Notion AI if Notion AI plugs into it or whatever, can I use integrations or can I use whatever?

And then can I, quote unquote, "pass my standardized procedure to this AI to work with my software and to use this thing to filter through my emails," right? And then- The last thing is to sort of set that procedure free. Can I set this task to run every day? Do I just have an agent check it every five minutes?

You, you, you choose that thing. So that would be step five, is you just consider, okay, do I... [00:57:00] Is this more of like an IFTTT where it runs every day at noon and there's still an AI doing it, but it's just, you know, like clockwork, the actual sendoff or the, the trigger is just a new day at noon? Or is it quite literally the bot is a little bit intelligent, you give it an AI agent and the bot realizes, oh, more emails are coming in than normal, maybe I'll check it twice a day instead of once a day, right?

And you kind of give it some autonomy there. But that's your kind of final thing where you kind of give it a push and you're like, "Okay, let's see if my standardizing and my research and all that stuff has, uh, come to fruition more or less." And then away you go at the end of the day. And so that, that's kind of the, the sort of rough framework that I've kind of laid out there for allowing you to go from just kind of like prompting ChatGPT or prompting things one-off all the time to actually getting it into a workflow.

Because you can see how this would extrapolate. If you get this working, now no one's checking the emails and that person's now dealing with, you know, some revenue-generating things. There's gonna be multiple procedures [00:58:00] inside of your business, multiple procedures inside of your business that are gonna naturally, are gonna naturally be potentially replaced with AI.

Oh, we have a 10-step process here. We'll put AI in step two and step eight, right? Now step two and step eight, that's two people. We don't need to do those things. We can put them over and have them do this now, have them be more customer-facing, have them generate more revenue, et cetera, et cetera

Mikhail: So I, I just wanna kinda end it with, uh, some suggestions for tools, 'cause I know we've talked very, like, theoretically about this, and we've mentioned ChatGPT and stuff here. Um, just to look at... And again, if you're listening to this a couple years from now, I would disregard whatever I'm saying, 'cause I'm guaranteeing you the tools that have been developed in the last two years probably are gonna be better, so just look, you know, do your own research on this.

But if you're listening to this now, you're a small business owner. Some of the kinda consumerized tools that have just come out, like in the last couple months, that make some of the stuff easier stuff like Grokbot. So the, this, uh, these are [00:59:00] tools that kinda give a agent its own computer, and then you sign into all the things that it needs access to, and it does things automatically for you.

And you can ask it to automate things for you. That's the biggest difference here, where as before you would have to kinda, like, go to ChatGPT, get it to write you a script, then implement that script somewhere, and have it, like, take you through how to do that. It was a lot more technical. These new tools like Grokbot, Metamuse, um, even ChatGPT Work, right?

Like, that one is a little bit l- less, you know, its own computer agenty, but still can do automated tasks. Uh, those kinds of tools are able to drive the implementation. So if you figure out places that you want automated, you can get those tools to implement those automations for you, right, on a basic level. I would still be a little bit wary of that if you're not technical at all. And if you're not technical at all, what I will say is that, like, we're in the space of potentially helping you if you need help. [01:00:00] DigitalDynastyDesign.ca, uh, you can go there, find our contact information, give us, send us an email, uh, and we can potentially set up, you know, a, an introductory call and figure out what we can help you with on automation.

But if you are slight-

Matt: Uh, d- sorry, digitaldynasty.ca

Mikhail: Sorry, digitaldynasty.ca, my bad. Uh, if you go there, and it'll be... The link will be in the show notes as well. We can potentially help you out. You know, the first call will obviously be free and all that, and, uh, the, you know, the s- the standard stuff. But, uh, if you're at all technical, you wanna try out one of these tools, again, Grok Bot, Muse, uh, Meta Muse, ChatGPT, Work Edition, all those tools can kinda almost guide you through the implementation. Uh, a- when you have a question that you would ask your IT guy on how to set something up, ask it. That's kind of how I've been phrasing it to my friends, to my parents, on, like, what you ask them, because they're like, "Well, what can I ask them?" I'm like, "Ask them that. Ask what you can ask them," and then they'll give you a list of things to ask.

So, like, you can have [01:01:00] it handhold you through a lot of this stuff, and get, get to a point where s- starting to help you like a coworker, right? Starting to... You're starting to think of them as a, a virtual assistant, essentially. Like, right? I- if you're, if you're gonna hire someone to do that, you can almost outsource some of that to, to these bots now.

And they're getting better and better by the day, so again, if you're listening to this even six months from now, I bet you these tools are now kind of outdated, and there's new tools, or these tools have grown so much that some of this advice might not apply. So, highly recommend you to do your own research.

But this stuff is really important to get your, get the handle on, because I can almost guarantee you that for most businesses, this will be the future

Matt: Yeah, I think AI is gonna definitely, definitely... Like, it's, it's already infiltrated the sort of small business scene. We've seen it, like, with those statistics that where the... It's, it's tripled in recent years, the usage, and I wouldn't be surprised if that, if that growth curve, you know, either continues at the same rate, or at the very least, it's gonna grow in [01:02:00] the amount of, uh, use cases and in the amount of businesses that actually use, use AI.

And when it comes to small businesses, you, you commonly wear so many hats that, like, you may as well... Like, you may as well get, like, a, a leg up wherever you can, especially if you're one of those small businesses where you're, you are kind of running rampant and barely kind of finishing things. Like, run here, barely meet the deadline, run here, barely meet the deadline, run here, barely meet the deadline.

Uh, definitely, you know, look into this. And, um, like Mike said, we can always help if you want, digitaldynasty.ca. Uh, we have our, we have our new website there. But yeah, go and, uh, maybe, you know, check that out if you're interested, if you're like, "I don't have time to, to figure out some AI stuff," you know, we can absolutely, uh, help you with that.

But I think that's it. That is the episode. Um, I would like to say that there's a, a quite a lengthy bit in the show notes, uh, talking about common AI use cases for small businesses. Again, I threw that in the show notes. [01:03:00] That'll be on htmlallthethings.com. If you're, uh, interested, uh, i-in those show notes and you want to see some of them, some of them are...

you know, inc- uh, include, like, a use case would be communication and document writing, uh, s- some, some stuff like customer service, uh, research. The list goes on and on, things we've touched on, but there's just a, a fair bit more detail inside of the show notes, so please go and, uh, check that out, uh, on htmlallthethings.com.

But that's it, unless you have anything else to add, Mike. I'm gonna run the old conclusion.

Mikhail: Runner up

Matt: Well, remember that we are on Patreon. That's patreon.com/htmlallthethings. Check out the tiers and give that a go. And many thanks to our $3 tier patrons, Tim from the Web Hacker on thewebhacker.com, Jason from Geek Life Radio via geekliferadio.com, Garrett Segall, Level Up Financial Planning via www.levelupfinancialplanning.com, Magnus from Yes Web via yesweb.se, Syntaxify from the HTML All the Things Discord server, and Stacy Mosler from the website screenworthydesigns.com.[01:04:00]

And remember, we have a contributing author. That is Michael La Rocca, and he is the, he is the author, excuse me, of Self-Taught: The X Generation Blog at selftaughttxg.com. And remember that we have a Scrimba discount link. If you want to get up to 20% off a Scrimba Pro plan, please go and check out that link.

It'll be in the RSS description as well as in the show notes on htmlallthethings.com with full details on how the link works inside of those show notes. Feel free to leave a comment or review on the platform you're listening to this on, and we are signing off