Why This Boring AI business Idea Makes Money

I'm going to show you what I think is the most practical opportunity in the AI space for freelancers, agencies, and anybody that's looking to build some kind of SAS product. The best part is you don't need to be some kind of AI expert or go allin to build the next big thing. It's easy to learn and it's something that businesses will pay money for. Trust me, you don't want to miss this. So, for some context, I've been in the web development space for over a decade now. I started my career as a freelancer, eventually went on to building my own agency, and I've even built and sold a couple of my own SAS products. So, I know what businesses want and I know how to sell to them. Now, the funny thing is is that in all of this, I never really cared to work on exciting products that revolutionize markets. I always prefer to focus on what people would call the boring stuff because I wanted to focus on things that I knew were already selling and things that businesses were paying money for. Now, I know that doesn't sound exciting, but one of the things that I teach newer developers that are trying to get into freelancing or maybe are looking for their first job, I always tell them that if you want good, consistent money, then boring is good. Now, there's nothing wrong with chasing exciting markets. But the problem with those is that they're unproven and they're often high risk. So, if stability and money is what you're after, then not reinventing the wheel and doing things that have already been proven to make people money is the safest route. So, this is what's kept me profitable for over a decade now. Now, when all this AI stuff became mainstream around November of 2022, I knew this is going to change the market forever. And my strategy with all of this was to pay attention, try everything, but don't chase. And I ended up incorporating a lot of the stuff into my life. So, for example, these AI code editors have significantly improved my productivity. AI surge has changed the way that I do research now. And honestly, I can't imagine going back to not using these. Now, there's a lot of exciting business opportunity that's popping up because of all of this. There's people like Peter Levelvels on Twitter, for example, whom some of you might know. He's building a lot of these AI products. One of those being an AI profile picture generator, and he made great money with that. He's building a bunch of other products, basically everything AI related nowadays. There's also products like site GPT which is basically a chatbot that you can use to integrate into any website and we're starting to see this being used in documentation websites and this just creates a great chat support experience. Now these are cool and a lot of people are going to make money from this but honestly I always prefer the boring stuff so I never really got the edge to build anything like this. In all of these new business opportunities, there was one that did catch my eye. And this is something that I realized was the missing piece to something that businesses were already paying money for in one form or another. And what made this possible is a combination of technology improvements in the AI space that now has made all of this readily available to everyone. So what's changed and what's this thing that businesses are willing to pay money for? Well, in order to understand this, I want to take a step back and see how we got here. So when I first started selling to businesses, I understood a very simple concept early on. And this is that businesses will pay for anything that allows them to make more money. And this can come in the form of increasing revenue through some kind of marketing services like SEO or digital marketing, all things that I've done. Or it can come in the form of automation that allows a business to capitalize on more by doing less. And this way they can stay more competitive in their industry. The automation part of what I and other web developers often sell is usually first the website, right? It's a form of automation because it allows a business to be present in areas where they're not physically there themselves. It allows a customer to gain some insight without any staff intervention in this process and then hopefully they can make some kind of decision like show up at the business or place some kind of order. Now, the other form of automation is going to be like the contact form or hopefully an order form on your website. This allows a customer to make some kind of purchasing decision again without staff intervention. Now, these are of course very simplified examples, but these are the first two that I can just come up with off the top of my head. Now, there are two other forms of automation that I want to point out because they're relevant to this scenario. So, the first form of automation is going to be some kind of robot that a business can set up. So whether you are having some kind of inbound call from a customer and the answer machine picks up and it tries to maybe guide you through some process which usually is terrible, it's limited and none of us like talking to robots over a phone and this can also be some kind of outbound robocall. Now the other one is outsourcing a call center. Now the robot answering machine is a nightmare. None of us like dealing with this and usually it's only for simple edge cases and it never really understands what you're actually trying to say. And I think all of us know this is a quite frustrating process when you have a real problem to deal with. So for me, what I usually do is I just press zero a bunch of times if the robot doesn't get what I'm trying to say. And I just try to brute force my way into getting a human agent and hopefully that'll solve my issues. And this system is used quite sparingly at companies. Usually it's a company that maybe can afford to lose some customers. I noticed airlines often have it. Banks have it because you have no other option and it's the best solution they have. Now the other thing is the call center and this is extremely expensive and my issue with this is that it also lowers the quality of service you can provide for your business because you're giving the service over to people that don't actually work at your company. Now I know this firsthand because I was tasked with solving this issue at my company and I had to come up with a solution for some kind of call center system. And the problem we had was that we ran national ad campaigns across the US. And we were a West Coast company. And the issue was that in our campaign, we would advertise all over the US. And a lot of our customers would call very early on from the East Coast. So 9:00 a.m. on the East Coast is 6:00 a.m. on the West Coast. And by the time our staff would come into the office and we can actually call back or maybe respond to some kind of lead, that customer, there was always a chance that they found someone else and went with another provider. So, we spent money and then we often left money on the table because we didn't get there in time. Now, I eventually was able to solve this problem, but the issue was that it cost us $30,000 a year, plus or minus a little bit, probably on the higher end, actually, and we basically just had to staff the office with somebody showing up very early on. And that wasn't the ideal solution, and it was very expensive. So, we know that robocalls suck, and call centers are expensive and usually not practical for most businesses. But something has changed over the last couple of years with the advancements of AI and other technologies that we have that really provides an alternative to both of these issues and solves most of these problems. Okay, so let's break this down. So first we had these large language models that were released and these really changed the world forever, but they still had their limitations, right? So, if you ask a chatbot what the current weather is today or what happened yesterday in London, uh it doesn't know that information because it's only trained and has its current knowledge set and that's it. So, it wasn't until we started plugging these into our own custom data sources, feeding it some kind of PDF or letting it make some kind of API request to our own database where we can actually ask at the weather, have the LLM make some kind of external request. This is what really made them powerful and made platforms like site GPT which allow you to add a chatbot into your own website and a lot of these other tools. It's when we mix them together. So this was amazing. But the other part of this equation is the advancements in the voice AI space. And honestly, it's incredible. It's getting so good and so customizable. It's often hard to tell whether you're hearing some kind of AI or an actual human voice. You can customize the inflection points, how it carries a conversation and mix that up with the knowledge base, it's getting really difficult. So with this combination with the LLMs and the custom data and voice AI, we now have the power to create amazing bots and amazing agents that can actually carry on an intelligent conversation and can actually provide information and even take in information from a user and perform certain actions like maybe place a reservation, offer a refund, and so much more. I want to take a moment to show you how powerful these can be with a quick example. So, I'm going to play you a clip from a video posted by Ibrahim Kosov, who builds and sells these voice AI agents as a service. And in this example, Ibraim demonstrates how one of these can sound. And the thing I want you to pay attention to is how the conversation flows. So, the conversation goes in various directions, and the agent is smart enough to be able to take in various types of input and still carry on the conversation, which would normally make an algorithmic bot just freeze up. So, let's take a quick look. Thank you for calling Wolf Gang Puck. This is Sophie. How can I assist you today? Hey, Sophie. I wanted to go ahead and do a reservation. Can you help me out? Of course. I'd be happy to help with your reservation. Could you please provide the date, time, and number of guests for your booking? Yeah. So, it's going to be four guests, and we want to come in some like tomorrow, 400 p.m. Got it. May I have your name and email address to confirm the reservation? Yeah, of course. So, my name is going to be Ibrahim Kashiff and my full email address is going to be IBR Ibrahim Kashif. Thank you, Ibraim. Let me quickly check availability for tomorrow at 4 p.m. for four guests. One moment, please. Great news, Ibrahim. We have availability for tomorrow at 4 p.m. for four guests. Uh, go ahead and reserve the table for you. One moment, please. You'll receive a confirmation email shortly. Is there anything else I can assist you with? So, the first thing that I noticed about this was the agent's tone and ability to actually listen and not force him into a certain structure in the conversation, but instead actually taking that input. So, that means that regardless of how the question was asked, the agent can actually make sense of it as long as there is a legitimate question in there. Something that typically a robot would have difficulty with because they're programmed for only certain keywords and maybe certain key clicks on the phone. So, the agent, if needed, can ask for clarification. It can ask for a rephrasing. This is all things that a normal robocall would have issues with. The other part is that the agent actually has access to certain endpoints. So, it's actually able to update a database. It's able to make the reservation and even send out a custom email to the email that was provided. Now, the rest of this conversation I want to demo because Abraham actually asks it a series of questions, things that might not have been by default programmed into this bot, but it does have access to this information. So, he's going to ask it some questions, and I just want to I want to make sure you pay attention to the flow of this conversation and the type of data that it actually pulls up and how it responds to questions that again would throw off a normal robot. Yeah. So, actually, I had a quick question. Can I request a specific table or seating area? Yes, absolutely. You can request a specific table or seating area. Would you like me to note down a request for you? Yeah. So, I actually had a quick question actually. Um, do you guys have any vegetarian options? Any halal options? Yes, we do offer vegetarian options such as our seasonal vegetable risoto and roasted cauliflower steak. As for halal options, while we don't have a dedicated halal menu, we can certainly accommodate dietary preferences where possible. Yeah. Do you guys do private uh dining options? Is that an you know thing that you guys do? Okay. So when I saw this example and others like this, I realized this is going to be something big and I can absolutely see businesses paying for something like this. I mean, if IM approached me and my company when we were facing our issue, I would have absolutely paid for something like this. And I'm also pretty sure it would have solved most of our issues. I also think that a lot of our customers would probably have no idea they talked to a bot. And this could totally solve that 3-hour window that we had and probably a lot of other calls that we handled. Now, you know how I like boring businesses. I like going for the safe bet. Well, I did some research and I found dozens of examples of people actually selling this kind of service and making money with it. So, I want to look at a few of these and I want to show you that this isn't some kind of theory. This isn't some new market that no one knows. It's new, but it's also been battle tested. And I like that idea. I like the idea of letting others test it and then coming into play once things have already been proven. So let's take a look at these real examples. For our first example, we have an outbound solution and this is an agency called Ingratagus run by Giannis Moore. And what they do is they provide AI lead nurturing, qualification, and reactivation services. And they specifically don't do any cold calling. So anybody that they're going to call for you has to have some kind of relationship with your business already. Now, the way this could work is if your business has some kind of CRM or a database and maybe you have a bunch of leads from your website or customers that maybe order a product once a year, this service here, Giannis can set this up where they can go down this list and call thousands of potential customers for you and try to reestablish that relationship, try to get them to maybe make some kind of decision on another purchasing order. And Giannis sells this for about $3 to $8,000 in most cases and has sold it for as much as $23,000 to a single business. Now, for an example for inbound solutions, we have an AI digital receptionist for property management companies created by Brendan Joit and this company is called Inflate AI. So, what this system does is it handles maintenance requests, general inquiries, and even transfers emergency calls. So, for property management companies that maybe manage hundreds, if not thousands of properties, this can be very useful. And in one case, he's even sold this service for $12,500 and a monthly retainer. So, I can go all day with these examples. I found quite a bit of them. And it's definitely a proven industry that people are already making money with. So, the facts add up. Now, if you made it this far, you're probably curious about the tools people are using to build systems like this, and you maybe want to get a sneak peek into how this works. So, in this next step, what we're going to do is we're going to get our hands dirty and we're going to build our own voice AI agent and we're going to add data from this phone call that the agent's going to take and we're going to add it into a database. Now, we're going to build a very simplified application. We're going to try to replicate like a pizza shop. So, we're going to call this pizza shop. This agent will take our order. They'll take our customer information and they're going to add this into an Excel file. Now, the actual database that you add this to, that's up to you. you'll see how to customize this, but we're just going to make this work and by the end of this video, we're going to have this working product. So, let's go ahead and just dive right into it and see how this works. All right, so let's build out our voice AI assistant. Now, the platform we're going to use for this is a tool called VPY. And this is one of the tools that I kept seeing come up in all of these examples. So, clearly, a lot of people like this and it's been working for people. So, that's what we're going to stick to. I've used it myself for a little bit and I really enjoy the interface. It's super easy to set up your assistant, set up the prompts, create a custom phone number, and so on. So, you're going to see how simple this actually is. Now, if you want to follow along, go ahead and use the coupon code linked up in the video description. This way, you get 100 minutes free. You can get started and follow along and even have minutes left over to build whatever you want after this for a little bit. So, make sure to use it. No reason not to use a free coupon code. So, this is our VPY interface and we're going to go through this pretty quick and we're going to start by building out an assistant. Now, there's a few ways to do this. We can use the new workflows, which VPY is building on top of still, but workflows allow us to build these more complex workflows. So, if your assistant needs to handle different kind of scenarios, this is one way to do it, and they even have some examples. Now, we're going to go with the simple route. We're just going to go with the assistant here, and we're just going to start building this out. So, we're going to create a new assistant. We'll go into blank template, and we're going to call this Jane Pizza Shop. So, it looks like it already filled it in for me because I was testing this out before. And we're going to create the assistant. Now, this is our interface to create the assistant. I just want to recap a few things real quick. So, here, this is where we can set up our provider. We're going to go with Open AI as a default. We can configure our model here. And this is where we set up our prompt. So, we have the first message mode. So, does your assistant speak first or do you want the user who's calling to talk first? Then we also have the first message. The system prompts. This is where we can add in all the information to the assistant and then actually let it know how to handle certain cases and so on and then any files that we want to upload like any knowledge sources, menus for our restaurant, whatever we're doing here. Now for the temperature, this is where we can configure how much do we want our assistant to follow our flow. Do we want to deviate from that and maybe use its own language or do we want to stick to the script here? So we're just going to set this and leave it at 0.5 here. And then we have our voice configuration here. So, our provider is by default set to Voppy. And our voice for this, I'm just going to go with let's see, we'll go with Kylie here. So, we'll use a female voice. And for the additional configuration, we have a restaurant. We can upload some kind of background noise. I'm not going to worry about that, but this just gives you an idea of how customizable this really is. Now, we're not going to go over everything, but one thing I did want to quickly show you here is the tools here. So when our assistant is actually updating a database or maybe querying some data, this is where we create a custom tool and we add it to this specific assistant. So we're going to do this in the next step, but really quickly, let's just go back to our prompt here. And what I like to do is paste in some layout here for the assistant. Now, you can set up whatever structure you want. There's really no predefined rules, but I find that this system right here works pretty well. So first, we tell our assistant its role. We give it a task. We give it some specifics. We provide context and then we tell it about the business, what the business does, provide examples, and then some kind of notes here. So, let's go ahead and just add in each one here. So, first for the welcome message. So, first message, I'm going to paste in, hello, welcome to John Doe's Pizza Shop. How can I help you? I think that's a good starter message. For the role, we're just going to go ahead and say you are Jane, a receptionist from John Doe's Pizza Shop. You are here to take orders and assist customers with any questions. And by the way, we're going to be doing this in markdown format. So this way it's able to understand the structure a little bit better. For the task, we're going to tell it your task is to help take orders. Add items into the system database while collecting customer and order details. For specifics, we're going to say refer to the menu information provided when answering questions about food options. Gather all order information with specifics. Now, for context, we're just going to add a little bit of information about the business. So this is John Doe's Pizza Shop. Authentic New York pizza serving the Upper East Side of New York for 10 years. You can add in store hours, whatever extra context you want to add, go ahead and do that. We're just going to keep this very simple. The what we do, we're just going to say we offer dine in and takeout orders. And then one of the things that's really good to do here is to provide an example of a conversation. Now, it doesn't mean that this is exactly how every conversation is going to flow, but it's really good to help an agent actually understand this flow. So, if we provide an example, it's more likely to follow it. So, for this example, I'm just going to provide an example conversation. So, we're going to say for the agent, you're going to welcome the customer. Customer is going to try to place an order. We're going to try to get their name and phone number. And we're just going to provide it some sample conversations that could happen. And this is really going to help here. So, with that, the additional notes that we want to add here, I'm just going to paste in a menu here. So, I'm going to go ahead and grab this. I created a quick menu and we're just going to throw this in here. And you could provide this menu in an alternative file. So you could go ahead and select a file, upload something here, and you can even make a query to an external database. For now, we're just going to throw everything in here just to make it work. So let's go ahead and paste in that menu. And here, I'm just going to go ahead and expand this. We're just going to see the menu itself. So we have the signature pizza menus. So we'll have pepperoni, meat lovers, Mediterranean, spicy Italian sausage. We have the price, some extra information about it, and then we have the price for gluten-free options, certain sizes, and then extra toppings. So, this should be enough for our agent to understand our business. Now, it's going to go off of this, and this is how we're able to price details and so on. Now, let's go ahead and hit update. We are going to publish this agent, and I could try to talk to it this way by just clicking talk to assistant, but instead, what I'm going to do is I'm going to go with a phone number, and we'll add this right here. So, let's go ahead and use a free VPY phone number. And for the area code, uh, let's just do 941. Let it create one. And this is our phone number. So, I'm going to dial this. I need to tune this to this specific assistant. So, Jane Pizza Shop, that's our assistant. And once this is ready, I'm going to go ahead and make this call. Okay. So, it looks like the phone number is ready. I'm going to go ahead and make a call. Put this on speaker. Hopefully, we can hear it. Hello. Welcome to John Doe's Pizza Shop. How can I help you? Hi, I'd like to place an order for takeout. Sure. May I please have your name and phone number to start the order? Yeah, my name is Dennis and the phone number is going to be 5552637418. Just to confirm, your phone number is 55526 [Music] 37418. Yeah, that sounds right. Great. Thank you, Dennis. What would you like to order today? Um, I'd like a large pepperoni pizza. Okay, that's one large pepperoni pizza. Is there anything else you would like to add to your order? Uh, no. But what's the total price to that? The price for a large pepperoni pizza is $16.95. So, I'll go ahead and place the order. All right. Wonderful. Thank you. In about So, you get the idea. The agent can definitely use some finetuning. There was a little bit of interruption there. Uh, this is something you can do through the Vappy interface. That's all highly customizable. Now, one thing that we didn't do is we didn't actually place an order. So, the agent's going to try to do something, but it doesn't really know what to do. The way that we can do this is we can create custom tools. And these are basically custom actions that we can have our agent perform. So we can query some kind of data source. We can transfer a call, perform an API request, integrate with Slack, Google Sheets, and so on. But what I'm going to do here is I'm going to create a custom function. Now, this function is going to be called place order. and we're going to go ahead and just send a web hook where call a certain endpoint, take the order summary like the size of the pizza, the name of the customer, phone number, and we're going to send this off there. So now this assistant can actually call this function, and the function will take care of the rest. So let's go ahead and add a description, and we're going to connect this function or tool to our agent. So for the description, I'm going to say send order details to URL. And then for the options here, we're going to leave that as is, but I am going to enforce strict parameters here. So I only want specific parameters. And this is where we're going to add them. So first, we want the size of the pizza. We'll leave that as a string. We'll say size of pizza. And we're just going to add in some details just in case, just so the agent knows more. Go ahead and hit apply. And then we'll just add in the rest quick here. We'll add a property for pizza type. For the description, we'll say type of pizza. We'll add that one. Then we'll add in customer name. Leave that as a string. Save that. And for the last property, this is where we want extra toppings. So extra toppings. And we want a customer to be able to add multiple. So we'll just set this to an array. For the description, we'll just say list of extra toppings. Customer added. And then for the type, let's just set that to a string. We'll hit apply and that should be all the parameters. Now I do want to make sure that some of these are required. So the size is required. Type of pizza is very important. Uh customer phone number that's going to be required. And then the others I can leave. So this is just ensuring that we have these. Now I'm going to go ahead and save this. For the actual URL, we're going to add this in a second. So we need an external tool to actually use this. Now to connect this to our assistant, we want to go to our assistant. Then we'll go into tools here and we just want to make sure that place order is selected so that way we can actually use the tool. And then I'm going to update specifics here. I'm going to add in a comma and I'm going to say once we get the specifics of the order then call the place order function to add the order to the database. Do not end the call without actually adding the order to the database or before adding it. So I just want to make sure to publish this and that's all updated. So this is how we can actually call this tool. Now in the last step here, what we want to do is we want to take in this information that we collect and we want to pass this along over to some kind of backend. So we can actually save this to a database. So all this order information like the size, type, customer name, and phone number. What we're going to do is we're going to add in some kind of URL here and we're going to pass along a payload to that URL. Now the way that you do this can be any way you want. You can build your own custom backend, create some kind of API endpoint, and then send it there. But what we're going to do here is we're going to use a tool called N8N. And we're going to let that handle everything from this point on. Now, if you don't know what N8N is, this is basically an automation flow builder. And it's great for building these processes here. So on the website, the first thing I see for an example here is an automation flow that we can piece together in these steps. So there's always some kind of trigger action here. In this case, it's whenever a user submits a form. Then we can take this to the next step, pass the data along to an agent, add this to a database, and then move on to the next step. And then we can decide what action we want to take from here, like add this message to a channel or update a profile depending on what data was passed along to it and what actions we're performing. Now, these processes can get very dynamic here. So, we can do quite a bit here and it's a great tool for that. So, let's go ahead and actually build out a very simple flow. And our first action is going to be that web hook URL that we're going to trigger. And then we're going to add this to a database. So you can go ahead and sign up for free. You get 14 days free with nan. Go ahead and create an account. And we're going to go ahead and start from scratch here. And we're going to start our workflow with our first action. So I can go ahead and click plus right here or in the first step. We're just going to select on web hook call. And this is going to give us a URL. So what I want to do is I want to send a post request. So from voppy by default this is going to be a post request. So we're going to select that as the HTTP method. We're going to leave this in test URL mode and we're going to copy this URL. So that's our endpoint that we're going to send data to. Now from here what I'm going to do is go ahead and leave authentication as none. And as an option I want to set cores here and I'm going to leave this as a wild card. So we're just going to leave an asterisk here. This just means that anybody can send a request to this endpoint and we don't need to specify anything. Now, if we want to test this out, let's go ahead and click on listen for test event. So, it's currently listening. We can go ahead and go into VPY and I'm going to add in the URL right here. So, this is the endpoint that we're going to send data to. And we're going to click on test here. And what I'm going to do is go ahead and just paste in an example of an API call that we can make. So, this is going to be JSON data. If you don't know what that is, this is just a key value pair. So we have a name and then a value and then we have a type and then a value. So key and value. This is the way that we're going to send data to this endpoint. So I'm going to save this and I actually want to call this and we're going to hit send request. And if it goes through we should see a status code of 200 message workflow started. And if we go into n here we can actually see the data that was sent over. So if we go into schema here we can actually see the body. So we'll go down here and then we can actually see the information that was sent. So we see the name, type, phone number, size, toppings and so on. So we just pass the data along here and I can see this in different formats here. So if I look at this in JSON format, this is the way that we sent the data. So that's the first step. This is how we trigger an action. Now in the next step, what we're going to do is go ahead and pass this information along to a Google sheet. So I'm going to go ahead and select Google sheet here. and I want to select append and update row sheet. So I just created a very simple Google sheet and what we need to do here is go ahead and actually connect to our Google Drive account to a specific sheet. So on Google Drive I went ahead and created this simple sheet with these headers here. So we have name, size, type, extra toppings and phone number. Then in here I'm going to go ahead and create credentials and I'm going to sign in with Google and I'm going to use the account that I created my document with. So I'm selecting that specific account here. And then once that is authenticated, I can try to connect to this actual sheet. So I'm going to go ahead and close this out. And within here, I'm going to go ahead and select a sheet within a document. For the operation, I want to select append and update row. And then we need to find the actual document. So mine is called pizza orders. So that's my Google sheet. And then for the actual sheet name, I just have this named as sheet one. So we'll select sheet one. And automatically once we select the sheet, what's going to happen here is it's going to take all these headers and it's automatically going to map these values here. So now we need to actually figure out how do we add this data that's being passed over from the web hook into these values and this is what actually adds it to the Google sheet. So if we go into our web hook, so basically when we trigger the web hook, the next step is to go to the Google sheet and it passes that data along. So from Voppy to the web hook to the Google sheet and here we can actually see that body right here. So we have the name, the type, the phone number. This is that test that we just ran. And what we can do is we can take these values and simply drag them in here. So here it goes ahead and adds that in. So the next request that we make, it's going to take the name value and add it into this field which will add it into the Google sheet. Now this is going to look a little bit different when the actual agent makes this request. So right now we're just testing this through our test tool. So what I'm going to do is I'm quickly going to make a phone call. We're going to see the new data type here. So we're going to take that data and then we're going to append it to the sheet. So let's quickly just make this call and then I'm going to fast forward through it and then we'll actually analyze the data. So I'm going to go ahead and save that. And in order to actually test this, we want to click on execute flow. And we want to wait for this event to trigger in test mode. So I'm going to go ahead and make that phone call. Hello, welcome to John Doe's Pizza Shop. How can I help you? Hi, I'd like to place an order. Thank you for your order. Dennis, you have a large pesto pepperoni pizza with pickles and it should be okay. So, I just completed that call. We can see that flow trigger through, but it didn't add anything to this Google sheet. And this is because we still need to map all of this to the sheet. So, let's go ahead and go into open here. And now that the agent sent this request, the data is going to look a little bit different. So if we go into body now we see that we have message and then we see tool calls and then function arguments and the actual data now sits within arguments. So it looks a little bit different from that test call. So now what I want to do is I just want to go ahead and map all of these to our actual Google sheet. So we're going to start with the customer name. We can put that in here and we'll just make sure that's saved. Then the actual size, we're going to throw that in here. For the pizza type, we'll throw that into type. And then toppings looks a little bit different because it's an array. So we'll throw that in here. It's going to map it in its own way here. And then last, we want the phone number. Now, we added all of our values, but one thing that we need to do is we need a column to match on. And what this means is that if we have a customer, let's say with a phone number, and if that customer calls again and they want to update their order, we're going to find this one unique value to match on. So it's kind of like an ID that we identify that customer with. So from there, a customer can actually modify their order. If they don't have a number, it's just going to add in a new order into the database, and we don't want that. So we want some kind of unique value here. So here, we're just going to select phone. And then this way, we can actually modify certain orders and so on. So at this point, this should be good enough. So I'm going to go ahead and close out here. And I should be able to re-execute this. So again, I'm going to go ahead and click on execute workflow, and we're going to test this out. Okay. So I put these side by side so we can see this entire flow. Let's go ahead and give it a call and see how this works. How can I help you? Yeah, I'd like to place an order. Great. Can I start by giving your name and phone number? Yeah, my name is Dennis and my phone number is 5552361234. That's correct. Perfect. Thank you, Dennis. What can I get started for you today? Um, I'll have a large 16inch pepperoni pizza, please. Just to confirm, that's one large 16-inch pepperoni pizza. Is there anything else you would like to add to your order? Um, I want to add onions to that. I think that's going to be good for me then. Got it. So, that's one large 16-inch pepperoni pizza with one extra topping of onions. I'll go ahead and place this order for you now. Just a sec. Actually, you know what? I want to make a change to that. Can I remove the onions? No problem. I will remove the onions from your order. So now it will just be one large 16 in pepperoni without any extra toppings. Is that correct? Yeah, that's correct. This will just take a sec. Awesome. Thank you. I placed your order for one. So just cut it off there. But there you go. We were able to add the order. It went through the flow and it added it to our Google sheet here. And you notice how it updated the topping. Well, the way that it was able to do that was because of this unique value with the phone number. So, because we were still using the phone number, it actually knew what sheet to update. If we didn't have a unique value, it would basically just add a new row to the database. So, we don't want that. We were actually able to modify by having this value. So, that's just a quick preview of one of the ways you can build something like this and the tools you can use. I know our example was super simplified, but even with only a few minutes of building and tuning this, we were able to make it sound pretty good. I know it had some issues with it, but that also was with us not really doing much with it. So, if you're able to actually provide it the right data, tune the model, really train it on what you need it to do, you can build out something truly amazing. I hope this sparks some ideas in you. If you've ever built something like this or you have some ideas, leave them down in the comment section. I would love to hear from you. So, that's going to be it for now. Until next time, I'll see you all in the next