Everyday AI Tools: Text, Image, Automation & Workflow

Jon Foster (00:00)
So I know we're talking about some important things, but some other important things you may want to ponder.

Penguins have knees.

Think about it. Do they?

If you drop a bar of soap on the floor, is the bar of soap now dirty or is the floor cleaner?

Sheridan Wendt (00:16)
Welcome to the AI Advantage podcast. I'm your host, Sheridan Wendt, and joining us today is my long time friend, John Foster, as we dive through the topic, everyday AI, tools that you can use right now. John, how you feeling?

Jon Foster (00:30)
feeling pretty

pretty excited to get into these tools that you're mentioning. So I think it's gonna shed some light on how we can actually apply AI in our daily lives.

Sheridan Wendt (00:35)
Yeah?

Yeah, yeah, yeah. Up to this point, we've been talking about what is AI conceptually and practically, and also what is its business value. But now we're going to dive into some actual tools that you can use right now. Anything that you're excited about that you want, you're like, is there a tool for this thing?

Jon Foster (00:57)
I'm just I'm I'm just looking I'll tell let you know when I when I when what pops out, okay I'm just window shopping right now not sold

Sheridan Wendt (01:02)
Yeah, you're window shopping.

All right, let's dive in. So we're going to cover a few different types of tools, different categories, right? So there are writing and language tools, imaging creative tools, automation tools, data analysis and insights type tools, customer service, voice, right? So then we're going to talk a little bit about how you can integrate it into a workflow.

right, a real workflow. So let's get started with the first category of tools, which are language and writing tools.

I know we talked a little bit about LLMs, and we're not going to do a deep dive into how they work on this episode, probably in another episode. But an LLM is a large language model.

that is kind of what is in the background. That's what you're interacting with. It's what's behind DeepSeq and Cloud and ChatGPT is a large language model. And they're used for writing a lot of times for, well, they're used for text, for language, for brainstorming, for replying to emails, for coming up with text. So a lot of times you might hear text to text or text to image or text to video, image to image, video to video.

What people are talking about is what is the input? Text. And what is the output? Text. So, chat.gbt is a text-to-text tool, but they've been adding other features like image and Sora for video. So, that's what you would call a multimodal model. It can do more than one thing.

So.

So the thing that the text to text models are great at is expanding on content, content expansion, if I can talk today, rewriting content, rewording it, remixing it, summarizing it, making it shorter so you can make things longer, make them shorter, improve clarity, or rewrite it all together. It's great.

Jon Foster (02:58)
So

you're saying this is the type of thing that can make a long-winded person like me more concise.

Sheridan Wendt (03:04)
yeah, yeah, concise is key. I would never consider that to be an adjective that describes John Foster.

Jon Foster (03:11)
I see on no never

Sheridan Wendt (03:12)
under normal circumstances. You don't say a lot on the episodes sometimes, but I think it's because you're learning, you're asking questions, it's good stuff.

Jon Foster (03:21)
Yeah man, I'm just sitting back and using more of these than I am this. Normally, normally you have to listen to me. This is, it's nice to take a break.

Sheridan Wendt (03:25)
Yeah.

For the listeners who are not watching, he's pointing at his ears, I think.

Jon Foster (03:33)
Yeah, they're little tucked in, but yeah.

Sheridan Wendt (03:35)
He said he's using his ears more than his mouth. right. So tools, example tools. Many of you probably already know text-to-text tools, ChatGPT, Claude, Gemini, DeepSeek, These are all text-to-text tools that you can use to draft emails, social media posts, proposals, long documents. It can interpret long documents and make a summary, right? You can give it.

you know, take something long and make it short. And then Gemini specifically is integrated into a lot of Google stuff, Gmail, Calendar, the forms, Google Forms, the Google ecosystem. Gemini can interact with a lot of those things if you give it permission to.

So practical things that people are doing with text to text. We just talked about some of those.

Let me give you an example of some of the more advanced ways that you might use a text to text model, right? Language and writing. I'm going to pull up the Advantage AI website, AdvantageLabs.ai, and go down to our agents section where we have various AI agents that do different things. So here's one that is a AI account manager, right? It can take customer inquiries, incoming

emails from customers and triage those and assign them to the right person or if it's basic stuff that's in the knowledge base, can reply back via email to this customer with links, you know, details on instructions but also links to the original article in the knowledge base. We also have an AI ads writer that can generate ad copy for Google, Facebook, LinkedIn.

Any platform you can do marketing on, can generate ad copy for it. I'm just trying to give you some ideas for ways that you can use these text-based systems to do really cool stuff.

Jon Foster (05:20)
So real quick,

for all the people that are not 100 % savvy, when you say ad copy, is that just a formatting that is specifically to these different platforms that allows for advertisement?

Sheridan Wendt (05:34)
Add copy is what you actually write in the ad right so it's the caption the text

Jon Foster (05:42)
Got you okay.

Sheridan Wendt (05:44)
call to action, right? The hook, the pain. They make you feel the pain and they give you a hook, right?

Jon Foster (05:50)
Yeah.

Sheridan Wendt (05:51)
Marketers know what I'm talking about. ⁓ Some more examples of things that you can do with text. ⁓ I'm just going to read some of these off, and then we'll move on. But these are all text-based tools for the most part. Coming up with a cold email campaign to send out to your audience. Coming up with the actual emails for each email in the cold email campaign. The AI cold email writer.

a growth consultant who can help you learn and understand how to grow your business and you tell it what your bottlenecks are, you're stuck at and it'll kind of tell you how to come up with a strategy on making that better. I don't want this to be just 100 % shilling session right now but there are a ton of ways that you can use these ⁓ text to text LLMs to do very cool stuff. right, so.

Jon Foster (06:43)
How many

different agents do you have on there? That's a lot.

Sheridan Wendt (06:46)
Yeah, yeah, there are three, six, nine, twelve...

13.

there's 13.

Jon Foster (06:52)
13.

Oh wow, 13 right out the gate, man. That's impressive.

Sheridan Wendt (06:57)
13 agents, yeah, thanks.

Jon Foster (06:58)
Okay.

Sheridan Wendt (06:59)
So this is a great time for us to take a break and hear a message from our friends and sponsors, and we'll be right back after

So let's get back into the different types of models.

Sheridan Wendt (08:17)
All right, so next up, I wanna talk about image and creative tools, right? Creative tools, images, videos, that type of thing. So image generation, right? Let's talk about that just for a second. It's where, it's just what it sounds like. A model will generate an image. Models can generate them, they can edit them, they can remove backgrounds from them, they can make changes, add things that weren't there, make transformations, put effects on there, right? Graphic enhancements.

They can add branding, can add ⁓ text to an image that wasn't there, lots of things that you can do with image editors. There are some great tools out there that a lot of people use like Dolly, for example, is an older one that I actually prefer not to use Dolly, but it's popular and people know it. It's spelled D-A-L-L-E, Dolly. Mid Journey is another one.

that has been around for a while. And these were tools that were out before more recent giants, like OpenAI's ⁓ new image generation that they have that's built right into ChatGPT. And then Google Gemini, they released something called NanoBanana. Say that. That's really funny to say. Yeah.

Jon Foster (09:28)
Banana, banana. That

is fun to say.

Sheridan Wendt (09:31)
super fun to say, So actually out of all those models, all of those different products that do image generation, I like Google Nano Banana the most. And not because it's fun to say.

Jon Foster (09:41)
You

I uptake my money.

Sheridan Wendt (09:42)
What's

that?

Jon Foster (09:45)
and shut up take my money

Sheridan Wendt (09:47)
Shut up and take

my money. It's called Nano Banana. What else do you need to know?

No, Nano Banana though, it really is legendary. It's crazy. It works way better than a lot of the other models. I used Mid Journey for a long time, probably a year or two, to create different types of art, to generate images from scratch. They had some newer features where you could...

define like a style and generate a bunch of different types of image all in the same style. Problems I used to see with mid-journey were that it didn't get the human anatomy very well, like sometimes people would have extra fingers or... What's that?

Jon Foster (10:19)
I was about to say, yeah.

I've seen a lot of the AI images where they have extra fingers and stuff like that,

Sheridan Wendt (10:27)
Yeah, people that have extra fingers. I have one where there's a girl who's like has her hands up like this, you know, but her one of her hands is facing the other way even though her wrist is facing this way. It's like it's super weird. She looks crazy. So, you know, the the models are getting better over time. That was probably I'd say six months ago at this point. Maybe no that was

Almost a year ago now. So that was mid-journey. So I'm not trying to bash on mid-journey, but everybody's been getting better a little by little over time. But Nano Banana is just on another level. So if you haven't tried it yet, try it out. It's free to try. I was really impressed with its editing.

giving it an image and saying add this text, right? Text was also another thing that's been really hard for image generators is they'll generate an image but adding text on it for whatever reason, just, it couldn't do it. It'd be misspelled, it'd be in another language. It'd be in a language that doesn't exist. It's weird.

Jon Foster (11:21)
like Klingon?

Sheridan Wendt (11:24)
Klingon?

It would just be random characters that didn't make sense. Anyways, so, you know, I'm not getting paid to promote Google Nano Banana, but it's nice. It's very nice. I like it. If you want to really make your content stand out, go try it out.

Canva has an AI tool as well, it's kind of worth mentioning, but I personally have not used it that much. I've heard other people use it and found it difficult to use.

Jon Foster (11:50)
Is it a photo editing one?

Sheridan Wendt (11:51)
So

on Canva, I hear they have like some AI suite now. It's relatively new, but yeah. Have you made any images?

Jon Foster (12:00)
⁓ I have not. not. Like I told you, asking any questions and like, you know what mean? Like just using, for example, if you just use Google, if you use Google now and use the recorder button, technically that's AI. I mean, so my extent is pretty much asking Google 40 million questions a day. So.

Sheridan Wendt (12:20)
You gotta

try the image ones, I'm surprised, man. With all the stuff you do, I'm surprised you haven't made like some cat ones or some dog ones or something.

Jon Foster (12:28)
I mean I love animals and everything like that but you know probably other things but no honestly you know it's there's certain certain things that should not be like given to certain people and I mean like you wouldn't give your son a pair of sharp scissors right?

Sheridan Wendt (12:41)
no.

Jon Foster (12:42)
Yeah, giving me an AI tool that can do a lot of different things that deals with pictures, maybe not the safest. I'm joking. Joking.

Sheridan Wendt (12:49)
All right, we are going to, we're just gonna try it out real quick so everybody can take a look at how amazing it is. I'm gonna do chat GPT versus Google Gemini and we're gonna see which image we like better. We're gonna use the exact same prompt for both images and let's see how it comes out, right? So what, give me a prompt, John. Give me something crazy.

Jon Foster (12:53)
Okay, let's do it.

Sheridan Wendt (13:12)
Let me share my screen.

Jon Foster (13:13)
Yeah, let me see your screen. Yeah, let me get some context. Um.

Alright, super random, super random. ⁓

Or we can ask it to depict the stereotypical Florida man.

Sheridan Wendt (13:25)
my gosh. All right, John is from Florida for anybody who doesn't know. you got to tell me the prompt, right? Because I would come up with some crazy prompts to put in here, but exactly what do you want me to type?

Jon Foster (13:36)
Show me...

a picture of

societal.

view of Florida man.

Sheridan Wendt (13:43)
All right, so John gave me this prompt. He said, show me a picture of societal, well, let's do this, society's view.

Jon Foster (13:51)
society's,

Florida school system, y'all.

Sheridan Wendt (13:54)
Ha ha ha.

All right, so right now we're doing a battle between two different models, right? These are multimodal models. Gemini and Chad GBT, which one has better image generation? And John Foster here said, this is the prompt. Show me a picture of society's view of Florida, man. So here we go. Gemini and Chad GBT, off to the races. Chad GBT's taking a while, that's because.

I'm on, well, I'm using the free tier on both of these. So shouldn't be a factor, but hey, Gemini and I actually just didn't generate a new image. It found an image from Board Panda.

and it didn't even show me the number one Florida man image. Maybe it's this. that's an advertisement. So it gave me an image.

Jon Foster (14:46)
No, that's good enough.

Or we should have said generate.

Sheridan Wendt (14:48)
Who?

Yeah, we should say generate next time, we use the same prompt in both of them. ChatGPT understood that it needed to generate a new image. Here we go. Here's the shirt. This is what it was trying to show us. Google Gemini is trying to show us this picture right here of Florida man. Who needs drugs? No, seriously, I have drugs. All right, Florida man. So, ChatGPT.

Jon Foster (15:11)
That's accurate.

Sheridan Wendt (15:17)
Still thinking over here. Let's do this. Generate. We'll do the same prompt and we'll change one word and we'll change it to generate.

generate picture of society's view of Florida man

I'm curious if Gemini is... I can create an image... Okay, it still didn't do it. got a voice or a text reply. I can create an image. Okay.

Jon Foster (15:46)
⁓ chat gbt's attacking me

Sheridan Wendt (15:48)
Maybe it doesn't like the word picture. Generate an image.

Jon Foster (15:53)
Hold on.

Sheridan Wendt (15:54)
Florida man. It totally looks like you! John's... John walked away. ⁓

Jon Foster (15:58)
Hold on.

Hold on.

Hold up. There we go.

Sheridan Wendt (16:06)
Are you putting on a Hawaiian shirt?

match this guy?

Dude, it's totally you. The beard, the hat, the sunglasses. You got the drink in your hand, except you're drinking green tea, I know.

Jon Foster (16:20)
Yeah, yeah. I'm a little healthier.

Sheridan Wendt (16:21)
So these are

the two different images we got. Now, and I, the image quality is better, but it took us two prompts, right? We had to reword the prompt a couple times. And chat GBT was very slow, but I only had to give it one prompt.

Jon Foster (16:26)
Okay, Gemini wins.

Well, I think I think we were more specific in our verbiage, ⁓ which I think isn't a bad thing. You know what mean? It's it. Depending upon what you ask, you get different things. Like if I if I asked generate or the same thing with Chat GPT I could have got the same photo. You know what mean?

Sheridan Wendt (16:44)
Yeah.

Right. So let's talk about for people who are just listening and not watching the Google Gemini image is a guy in swim like swim trunks flip flops. He's on a like a boat, a fan boat. He's got an alligator on his shoulder, a raccoon on his other shoulder. There's a tornado. Yeah.

Jon Foster (17:13)
Fanboat.

It's flooded. It's

a neighborhood street that's flooded. There's a tornado in the background and a shopping cart's on fire. Hold my beer.

Sheridan Wendt (17:29)
Yeah, he's thinking, hold my beer. He's got like a thought bubble and the caption on the bottom says Florida man, the legends are true. And he has a, what are these like hot fries, hot Cheetos or something in a bag in his hand?

Jon Foster (17:42)
Yeah, something like that. you know, he is in his Florida best, you know what mean? He's in flip-flops and swim trunks with no shirt, you know? That's how we go to funerals and weddings.

Sheridan Wendt (17:48)
Yep.

He even has a little bitty beer belly.

Jon Foster (17:53)
You mean?

Sheridan Wendt (17:53)
Alright, so

then the chat GPT version also this one looks more like John actually. He's got, ⁓ yeah, he's missing some teeth. He's got a Hawaiian shirt on, he's holding a beer, presumably is a beer. ⁓ the state of Florida is in the background. The dude has a beard and sunglasses and a hat on with his missing teeth and it just says Florida man at the top. So.

Jon Foster (18:01)
I have all my teeth, thank you.

Dude, so Chat GBT probably has access to our camera and is copied this man after me. Like honestly, it is, it is very similar. And this is the thing I also want to point out. And this is very accurate. He has no undershirt underneath that Hawaiian shirt. That is a fashion faux pas everywhere else except Florida. Don't it out there. It's accurate.

Sheridan Wendt (18:43)
Hahaha

All right, all right, so there we have it. Quick live demo, competition between the two, Google Gemini 1, at least this time.

All right. That was fun, yeah. So there we have some image generation that we just did. There are actually tools out there that you can call an image generation API. So you can get real fancy. You can create a workflow where you say, let's say you want to create a social media content engine. And you want to either have a, here's something that I'm doing with Advantage AI.

Jon Foster (18:55)
See that was fun.

Sheridan Wendt (19:22)
We take these podcasts, put them into a system, like feed it in as input and have it write a blog post based on this episode, right? And then take the blog post and the episode, use them as inputs for a workflow that takes the transcript of these and generates content from them for social media and then generates an image that is relevant to that post, to that caption.

Right, so a thumbnail, yeah. It could be a thumbnail, could be like, we could totally, you know, all day we can use it to make the thumbnail for this podcast episode. But what I'm talking about is like, just drop, for example, this episode, the video of it in a Google Drive folder, and that's it. And then you have like 10 pieces of content that get posted out on social media because it went through a workflow, generated a thumbnail, took the most important parts of this interaction.

Jon Foster (19:49)
Kinda like a thumbnail.

Sheridan Wendt (20:17)
and maybe made like a a model like a post called model wars and it shows Chad GBT versus Gemini and that Gemini one right and so it generates an image of that and then it has the caption and it posts it for you and all of that happens all you had to do is drop the video in a folder

Jon Foster (20:32)
We don't really even need AI for like the caption, you know, or the title of this. It's AI in Florida, man.

Sheridan Wendt (20:40)
You don't need AI for sure, but you could use AI to make five different versions of the post and pick the best one. ⁓ Or take this video, this episode, that's going to be however long it is and make a bunch of posts based off of it so that we didn't have to name all of them.

Jon Foster (20:48)
Yeah, right.

Sheridan Wendt (20:58)
All right, so we're going to move on to the next section talking about automation tools. There are a bunch of tools out there that you can use to automate things. There are so many ways you can do it, but there are some very user friendly, no code or low code ways like Zapier, which we talked about before, and make.com and n8n.io, I believe. They also have n8n.cloud for hosted instances.

And all of those are what I would consider like a workflow engine that can be connected to a bunch of other apps Google Gmail Google Drive, right? I'm talking all about Google stuff here, but also custom API's web hooks your own applications All kinds of things I couldn't begin to most of them have over a thousand different integrations different apps that they can integrate with So

crazy, crazy workflow stuff. you might be thinking, where does AI come in if you're just making a workflow? You can use AI as one or two of the steps in the workflow to maybe to give feedback, to iterate, to create different versions, pick the best version, to generate new text or images as part of the workflow. So it'd just be one step in the workflow. And also, do you remember our definition in episode one?

the different categories, different types of AI.

Jon Foster (22:20)
Yeah, I remember them, yeah.

Sheridan Wendt (22:21)
You do? What are they? I'm curious.

Jon Foster (22:24)
Generative. Generative. ⁓ Protocol.

Sheridan Wendt (22:24)
put them on the Put them on the spot. Generative, yeah.

rule-based, rule-based automations.

Jon Foster (22:31)
rule-based.

I'm doing it backwards doing all over the place because that I did three in one and then there was a Generative is the second one and then rule based is the first one and then the second one was transcribed

Sheridan Wendt (22:38)
Yeah, you are.

So the second one was

machine learning, like data analysis and machine learning. All right. So what we just talked about were automation tools. Automation technically is a form of AI, the early primitive form, I would say. But you can connect machine learning tools or generative AI tools to these automation workflows, make them even more

Jon Foster (22:54)
Machine learning, okay.

Sheridan Wendt (23:14)
Speaking of which, next up, the next category we're going to talk about is data analysis and insight tools. So the way you can think about that is maybe you have a spreadsheet and you want something to happen based on that spreadsheet, you can connect AI, like you can connect a model, an LLM or something to your workflow and have it interpret the data that's in the Excel spreadsheet and write you a summary or report.

of what happened.

No ideas.

Nothing to say.

Jon Foster (23:42)
So

I want to redo the whole entire one because I thought I had it. I thought I had it, but I didn't. protocol is pretty much a rule. It's a command. You know what mean? It was just, yeah. I get half points.

Sheridan Wendt (23:50)
That's okay, we'll make it shorter.

Yeah, you were there. I gave you that one.

It's pretty close man, it's pretty close. So...

So anyway, data analysis and insight tools, definitely some things that could come in handy. One of the things that I'm working on for a client is they have a lot of data that's stuck somewhere that they can't, you know, they can do reports based on it, but the reports are really ugly and they're not very insightful. They're kind of confusing and really long, like 30 pages long. Like who wants to read a 30 page report? That's basically a spread, like a giant 30 page spreadsheet, right? And so, ⁓

Jon Foster (24:26)
No.

Sheridan Wendt (24:32)
What I'm trying to help them do is take that data, put it through one of these workflow engines, interpret it with a model like a generative AI model, and the model will look at the data and then generate a summary. Generate like a summary and some charts and then continue, pass that information on in the workflow and stitch it all together into like a PDF report and an email summary and then send that over to...

their client, my client's clients, right? That way they don't have to put all this stuff together themselves and go and explain it and answer a bunch of questions about a 30 page report, right? So the way, if you combine, I guess one of the key takeaways here is if you combine generative AI with automation and data analysis, you combine those three different types together into a single workflow, you can really do some cool stuff.

Jon Foster (25:24)
Yeah, it seems like it would save like a lot a lot of time and make things a lot more clear You know that that doesn't just save you an hour that saves you days weeks

Sheridan Wendt (25:34)
It's gonna save, they have, this particular client I'm talking about has a lot of clients, so many that they're trying to get all this information to them and answer these questions and they, in a month they can't even do it. And then the next month it happens all over again, right? So, it's gonna save them a lot of time.

Jon Foster (25:52)
That's amazing.

Sheridan Wendt (25:54)
All right, so another thing that you can use AI for right now today, customer service, right? Probably everybody at this point who's listened to this has seen before a chat bot.

Right? You go to a website, you click on support, there's a live chat. Well, it says live chat, but it's not a live person. Right?

Jon Foster (26:15)
I mean you're chatting live, just not with human.

Sheridan Wendt (26:17)
You are chatting a lot.

Just not with a live person. So, right, right, right. So there are, and when I say customer service, that's one of the things, that's probably the first thing most people think about is like a chat bot that you interact with. But there are also AI assistants that help support representatives while they're actually interacting with the customer.

Jon Foster (26:21)
It says live, not alive.

Sheridan Wendt (26:41)
There's also suggested replies, right? Somebody might get an email and AI might suggest a reply and the support representative can choose to use it or start with that reply and modify it somehow and then send it. And then there's also instant answers from knowledge bases, right? You type your question in a chat and it says, ⁓ I think what you're asking for is this, here are the steps and it'll actually spit the steps out but also give you a link to...

the article that it got those steps from so you can read it in detail if you want to.

you seen anything like that?

Jon Foster (27:12)
I don't think I've seen that yet, no.

Sheridan Wendt (27:15)
Yeah. So Zendesk AI is a product. It's a feature of Zendesk, something new that they're releasing. Intercom Fin is one. There's also a stack that I want to say is probably part of GitHub or Git pages where you basically put your knowledge base there and there's an AI component so that when people search,

It's not just like a normal text-based search and it returns all the results. I mean, it will return all the results, but also there'll be some sort of AI component where it tries to make the best guess about which one of the articles is the right one that you're looking for.

Another.

Jon Foster (27:57)
neat

very neat

Sheridan Wendt (27:59)
Yeah, yeah, yeah, yeah. Have you interacted with any of those before? Have you seen any of those? You probably are not going into developer docs very often.

Jon Foster (28:06)
I'm not, I'm not going into developer docs. I'd being a developer. I, I respect a lot, but I think it takes a certain person, somebody that was a lot of patients. And I mean, to go through that many lines of coding and stuff like that and not go crazy. So, but I I have some experience, definitely, definitely in college and like being able to spot like vulnerabilities and script and stuff like that.

Sheridan Wendt (28:23)
Yeah.

Jon Foster (28:33)
I have some experience with that, overall, no, I'm very far from a developer, but definitely going to hop on this AI train to try to, you know, kind of compensate one of my weaknesses, if you will.

Sheridan Wendt (28:44)
Yeah, yeah for sure. All right, still on customer service tools, voice agents, right? There are a bunch of platforms out there that do voice agents. I'm gonna pull up a few and we can, I'll just kind of name some of them and then we'll try one out if we have some time.

here are some websites where you can go and create your own voice agent if you wanted to. One of them is called AutoCall, autocalls.ai. They integrate with some of the voice models that people like from like Eleven Labs, for example, but you can also use some of their default stuff.

⁓ There's also VAPI.ai, which I don't know for sure, but to me sounds like what they're going for was voice API. ⁓

Jon Foster (29:34)
Real

quick, you say voice agent, you're talking about like when I'm talking to a bot for like customer service and that's the voice I hear, right? Okay. Got you, okay.

Sheridan Wendt (29:42)
Yeah, it's an AI agent that's voice enabled.

Yes sir. Here, I'm gonna do a quick, I'm gonna just call one. I won't tell everybody.

which one I'm gonna try out just to make it fair for competitors and whatnot. But I'm going to put my number in here and have it give me a call. We're gonna do a test phone call.

so people can see how good this stuff is. This is something that people ask us for a lot, because some people have like 50,000 people they're trying to make phone calls to, and they just don't have time. And you can automate it with this. So here we go. Here's an example.

Hi Sheridan, I'm Archie from Watercooled, how are you? Hey Archie, I'm doing well, how are you doing?

good i'm doing great thanks for asking i'm here to showcase what what to call can do we offer a i'd have a ton of the agents what they can come in coming in outgoing anything specific you're interested in learning about our platform what kind of features do you have or school for the variety of features including support for forty languages

Jon Foster (30:45)
Bye.

Sheridan Wendt (30:46)
directly

from the platform. SIP trunking for integration with any local VIP or PVX provider. Over 300 tools available for integration in our internal no-code automation platform, similar to Zapier and Make. Additionally, we recently launched a white label option which allows you to resell our fully featured platform to your customers with full branding capabilities. Is there a particular feature you're interested in or a specific challenge you're facing with your current setup?

Could you help me answer calls at night when my business is closed? I understand. Absolutely. Autocalls can help automate answering calls during off hours, ensuring that your customers receive timely responses even when your business is closed. Our AI-powered agents can handle inquiries, take messages, and even schedule appointments, providing a seamless experience for your customers. Would you like to explore how this could be set up for your business?

Yeah, can I schedule a demo? If I implemented this

in my business would it actually work if ⁓ I wanted somebody to schedule an appointment now on the phone?

What's the next available appointment? Okay. Let me check the next available appointment slots for you. Please hold on for a moment. It looks like there are no available slots for tomorrow.

November 19th. Let me check for the next available day. I'm looking for available slots for you now The next available appointment slots are on Thursday, November 20th at 3 a.m. And 10 a.m. But either of these times work for you 10 a.m. I Understand great choice. I'll go ahead and schedule the appointment for you at 10 a.m. On Thursday, November 20th Please remember that this booking and you'll need to confirm it using the link you receive via whatsapp after this cool

Would you like me to proceed with scheduling it now? Yeah, let's do it. Noted.

Your appointment has been successfully scheduled for Thursday, November 20th at 10 a.m. Remember to confirm it using the link you'll receive via WhatsApp after this call. If you have any other questions or need further assistance, feel free to reach out. Have a great day, Sheridan.

All right, man, what'd think?

Jon Foster (33:24)
is crazy.

So okay, one, they get the whole entire accent in this and the pronunciation of the words, like schedule in English. English. Yeah, one one one that bothers me internally because I'm American. But, you know, yeah, the second the second thing that I really liked is that it's programmed to utilize psychology. Great choice.

Sheridan Wendt (33:35)
schedule. The British accent.

The British accent got you.

Yeah?

Jon Foster (33:53)
congratulating them on their decision, making it a, which when you're dealing with something that is not personal, you know what I mean? It's high tech, not high touch. Seeing that incorporation of the psychology when it comes to language is huge. That's, that's, I love that. That was great.

Sheridan Wendt (34:13)
Crazy, Now, there were some glitches. Don't get me wrong. It wasn't perfect. There was a couple times where I was still talking and it was like, up, right? But if you've got to call 50,000 people and you can deploy a tool like that, I think it's pretty helpful, right?

Jon Foster (34:14)
Mm-hmm.

Yeah, I think so, too. It's a lot better than I was expecting, for sure.

Sheridan Wendt (34:33)
Yeah, yeah, very nice. All right, let's see. What else did we want to dive into? ⁓ How to integrate this into a real workflow? I would say, let's not try to reinvent the wheel here. Pick something that you already do every day and figure out how can AI help me with that thing. I would recommend probably going back to, I want to say, episode

It's one of the other earlier episodes in the series where we actually talk about workflows. we do kind of a, here's how you decide which workflow AI would be great for.

All right, so we kind of dived into how you can use AI for writing, for creative stuff, for business stuff. We are going to... We had some actual examples, right? Let's see what else we got in here.

could probably do a whole episode on what not to put in AI.

Jon Foster (35:19)
⁓ yeah. I would run that one.

Sheridan Wendt (35:23)
So real quick, before we go, want to touch briefly on AI hygiene, right? Prompts, privacies, policies. We're going to do probably a deep dive episode on this later, but just a few things to note not to put into AI. Don't put sensitive stuff in there. Don't put PII, right? Personally Identifiable Information, your birthday, your social security number. Don't put your client's information like that in there.

Don't put sensitive or internal only information in there that you don't want anyone else to have, right? So just don't put anything in there that you wouldn't tell a stranger, right?

Jon Foster (35:57)
question when it comes to inputting information in there. Having the ability to pull up and having a system that is ran by AI. It's kind of hard to get around not having some personal information in there. Question. Can you use ⁓ masking and other techniques like that inside of AI to kind of keep the information as private as possible?

Sheridan Wendt (36:22)
I mean, you could, yeah. I mean, it sounds like what you're saying is maybe for a social security number, like XXXX and then only the last four, right? I wouldn't even put the last four in there. Like, why do you need it in there? You know?

Jon Foster (36:29)
Yeah.

Yes.

I wouldn't either.

mean, it depends on what's going on. Some account numbers are referencing it, normally having an account number for that individual client, then yeah, that's probably a better route. yeah, keeping all that stuff off.

Sheridan Wendt (36:48)
So I would just.

If you can, I mean, I would just put like SSN 1, SSN 2, unless you're processing just a crazy amount of data. The other thing you could do is keep it all on premise and not use some sort of cloud-based AI. So don't put it in ChatGPT, but get a large language model. Bring it on premise wherever your data is at. Have it interpret the data there and not share it outside of your firewalled, sectioned off environment.

Jon Foster (36:53)
Yeah.

Sheridan Wendt (37:15)
Yeah. Then it doesn't matter what data it has access to because it's contained, right?

Jon Foster (37:22)
sure. Yeah, because honestly, the thing that I really want to learn right now is how to incorporate into security and not having it as a vulnerability, but to, you know, improve workflow, but having that safety. And I think that's going to be a lot of other people's concerns and wants to. So I think, yeah, having a whole entire topic of what not to put in the AI is probably going to be the most informative, I believe.

Sheridan Wendt (37:46)
All right, man, this was fun. This was really fun.

Jon Foster (37:49)
Yeah man, thank you for taking the time to part some knowledge onto me,

Sheridan Wendt (37:53)
any time, man, any time. So thanks, everybody, for listening. This has been the AI Advantage podcast. I'm your host, Sheridan, signing off. I'll catch you next time.