Building AI Automation Workflow, AI Voice Agent, Vector Database, and Model Context Protocol

1. Introduction: Hello, guys, are you interested in building AI automations, AI Voice agent vector database and MCP? If the answer is yes, then this course might be for you. In the first project, we are going to build AI email responder automations using Zapier. In the second project, we are going to build AI customer feedback analyzer automations using Zapier. In the third project, we are going to build AI expense categorizations automations using Make. In the fourth project, we are going to build AI inventory tracker automations using Make. The fifth project, we are going to build AI appointment booking automations using NN. In the sixth project, we are going to build AI content marketing automations using NightN. In the seven project, we are going to build AI product research agent using Relay App. Next, we are going to build AI contract Analysis agent using Gumloop. Next, we are going to build AI receptionist Voice agent using 11 Labs. Then after that, we are going to build AI customer support Voice agent using FAPI and we are also going to build AI sales Voice agent using Voiceflow. Then after that, we are going to build AI lead generation agent using Zapier. We are also going to do a little bit of programming. We are going to build AI Data Analysis agent using Python and Llama. You will be able to upload a CSV file and the AI agent is going to analyze the data and gives you valuable insights. In the next project, we are going to build AI multi agent system for project management. We are going to have three AI agents, and these three AI agents will interact with each other. Next, we are going to build AI foot recipe generator. And we are also going to connect the AI model with Pinecone vector database. So the AI model is going to generate food recipe based on available food ingredients. And at the end of the course, we are going to build MCP server using Zapier. We are also going to connect the MCP server to Google Sheets and Mistral AI. 2. AI Tools & Resources: Holi guys, welcome back to the course. In this video, we are going to get to know more about all tools that we are going to use in this course. The first one is Zapier. Let's click on this link, and you will be redirected to this page. Zapier is one of the most well known platform for building AI automations. You can use Zapier for building AI agents, AI chatbots. You can manage your data here. You can build interface for your clients and still many more. Let's click on products, and let's click on AI Automations, and let's scroll down a little bit. You can build AI automations for many different use cases, for example, marketing, sales, product, IT, and still many more, and it is relatively easy. Just drag and drop. You don't need to know how to code to build AI automations, and don't worry. I'm going to guide you step by step. And let's click on App Integrations. There are more than 8,000 apps that can be connected. Let's scroll them. Look at this. You can connect Zapier to your Google Sheet Gmail slash Google Calendar, Google Drive Notion, Hubspot, Google Form, Facebook link ads, MailChim and still many more. And don't worry, Zapier is completely free to use. Let's click on Pricing. As you guys can see, Zapier offers a free plant. Look at this. Okay, now let's go back to the slide and let's move on to the second tool Make. Let's click on this link and you will be redirected to this page. Make is very similar to Zapier. You can use Make for building AI automations. You can use Make for building AI agents, and let's scroll down a little bit. You can build AI automations for IT operations, marketing sales, finance, customer experience, and people or HR. And let's scroll down. There are more than 3,000 apps that can be connected to Make. Let's click Bros Apps and let's scroll down. You can connect Make with your Google Sheet, open AI, Gmail, Google Drive, telegram, Airtable Pinterest, Notion, Google Docs, slack, and still many more. And it is relatively easy to use, guys, mostly drag and drop, so you don't need to know how to code. And let's click on pricing. As you guys can see, Make is completely free to use. It offers a free plan. Obviously, there are some limitations when it comes to the free plan. For example, you only have 1,000 credits per month. So make sure that you are fully aware of that limitations. Okay, now let's go back to the slide, and let's move on to the third tool n8n. Let's click on this link and you will be redirected to this page. Night N is a little bit different compared to they appear and make because it is self hosted and it gives you more control and flexibility over your AI automation workflow. And you don't need to know how to code because Night N is mostly drag and drop. Let's click on products and let's click on templates. NightN has many templates that you can use and customize. Look at this. So you don't need to start from scratch. You can use one of these templates. Let's click on products, and let's click on Integrations. There are more than 1,200 integrations available. Look at this. Google Sheet, Gmail, open AI, slide, Telegram, Google Gemini, Anthropic and still many more. Okay, let's click on pricing. Unfortunately, Naten doesn't have free planned but don't worry, it offers two weeks free trial. Okay, now let's go back to the slide, and let's move on to the next tool, Flowis AI. Let's click on this link, and you will be redirected to the spades. We are going to use Flowis AI to build AI lead generation agent. FloIS is an open source agent system development platform. It is relatively similar to end at end. So it's going to look like this. And there are so many things that you can build using Flowis. You can build multi agent system. You can build chat assistant. So it's like a chat bot like this. And if your AI automation workflow requires human review, you can add human in the loop, guys. Look at this. And don't worry, it is completely free to use. Let's click on pricing. As you guys can see, FlowIS offers a free plan. Obviously, there are some limitations when it comes to the free plan, for example, you can only have two flows, and the data storage is limited to five MB. So make sure that you are fully aware of these limitations. Okay, let's go back to the slide, and let's move on to the next one. Pinecone. Let's click on this link and you will be redirected to the spades. So Pinecone is a vector database. In one of our projects, we're going to store food inventory data in Pinecone, and additionally, we are going to connect Pinecone with the Lord's Language model. Okay, now let's click on documentations, and let's click on QuickStart. We are going to use Python, so let's click Python. But PyCon also supports other programming languages like JavaScript, Java, Go and CR. Here is the code documentation. You will need to copy and paste this code and you will also need to get your API key. But don't worry, guys. I'm going to guide you step by step. And don't worry, Pinecone is completely free to use. Let's click on pricing. As you guys can see, Pinecone offers a free plant. Look at this. Okay, now, let's go back to the slide, and let's move on to the next tool, 11 Labs. Let's click on this link, and you will be redirected to the spades. We are going to use 11 Labs for building AI voice agent. So we're going to build AI receptionist that is able to talk with your customers and have normal conversations just like human guys. There are so many things that you can do with 11 Labs. For example, you can convert text to spits. You can build AI agent, and that's what we are going to do in this course. You can generate a song. Look at this. Very cool. You can convert spits to text. You can do dubbing, you can clone your voice, and still many more. Don't worry, 11 Labs is completely free to use. Let's click on pricing, as you guys can see 11 Labs offers a free plant. Look at this. Okay, guys, I think that's it. That's all you need to know. I'll see you guys in the next video. 3. Building AI Email Responder Automation with Zapier: Holly guys, welcome back to the course. In this video, we are going to build AI email responder automations using Zapier. It is a simple automation that reads your email, write reply draft based on past conversations, and it still requires your approval before sending the email. Okay, let's get started firstly. I'm going to log in using my GML account, so let's click Continue with Google. As you guys can see, I have successfully logged in, and here is my dashboard. Okay, now let's click Create and let's click Zaps automated workflow, and you will be redirected to this page. In this case, you have two options. The first option is to use AI to automatically generate Automation's workflow. You just need to write the prom, explain what kind of automations that you want, and in less than a minute, AI is going to generate you the automations workflow. And the second option is to create the workflow manually. And that's what we are going to Okay, let's get started. Firstly, I'm going to click this trigger. And I'm going to select email. So let's click Apps and let's click Gmail. So in this case, we want Zapier to be able to access our email inbox. Let's click Choose an event, and let's select new email. So it will be triggered when a new email appears in the specified mailbox. Okay, now we're going to connect Zapier with our Gmail account. So let's click Sign in and I'm going to select this email account. And let's click Continue. Okay, now we're going to give Zapier access to the email inbox. So make sure that you check these boxes, and let's click Continue. Okay, we have successfully connected Zapier with the Gmail account. So now Zapier will be able to access the Inbox and read our email. And let's click Continue. Let's click this, and let's scroll down a little bit, and let's click Inbox. And let's click Continue. And we are going to test the trigger later. So let's close this for now. And we're going to add the AI actions. So let's click this. Let's click Actions and let's click AI. And let's click right because we want AI to be able to write the email draft, right? For the model, I'm going to use GPD four point oh. That's the best model. But there are many other options. You can use Anthropy. You can use Google Gemini, and there are other options down here. You can use any AI model that you want, it is totally up to you. Next, we are going to write the prom. Okay, I'm going to remove this prom, and actually the prom is going to be pretty long, so I'm going to copy and paste the prom, so you can pause the video and you can rewrite the prom. You are an expert email responder based on the email received below generate a professional and personalized reply that address the sender main concern or equations, maintain a friendly and professional tone matches the style of previous conversations. Be warm and helpful, is concise but complete, includes a clear call to actions if needed. Well, I'm going to remove the date because I don't think it is needed. Okay, so that's my prom. So you can post the video and you can type this prom. And if you want to use AI to improve your prom, you can click this. Next, we are going to generate the output fields from the prom. So let's click this generate from Prom. It will take a few seconds. Okay, here is the output filds, reply body, and let's scroll up a little bit. Now we are going to add the input field. Let's click this select input fields, and we are going to click complete this step. And let's click Test Trigger. Let's click on that button. Okay, now it is testing. Okay, here we go. And let's click Continue with selected record. Okay, next, we are going to click at fill. For the filled name, I'm going to name it. Email from. And let's add the field value. I'm going to select from email. And let's click Save. And let's add another field. I'm going to name it email subject. And let's add the field value. I'm going to select subject. And let's click Save and let's add another fild. I'm going to name it email body. And let's add the field value, and I'm going to select body plant, and let's click Save. Okay, now it is evaluating the prom strength. It will take few seconds. Okay, the prom strength is good and you can click this generate preview. It will take few seconds, maybe five or 10 seconds, then just click Finish. Okay, next, we are going to click actions again and let's click Gmail, and we are going to select the action event. I'm going to create draft reply. And let's click Continue for the thread. We are going to select this one. And let's scroll down a little bit for the body, we are going to add the email body. So let's click this and we are going to keep the rest blank. And let's click Continue. And for now, we're going to skip the test, so let's click Skip Test. And let's click Publish. So we're going to publish the automations. Okay, now it is working, and we are going to test these automations. Okay. Now I'm going to send the email to helpful bottengmail.com, and we're going to test if the automation is working. When we send the email, AI should be able to write the email from our inbox and automatically write the draft. Let's click New message, and I'm going to send the email to helpful boot ten@gmail.com. That's another email. Let's write the email subject. Partnership proposal. Hi, Chris. How are you doing? It has been a while since we worked together in tech consulting team. Currently, I'm building supply chain optimization software. And I was wondering if you are interested to join my team. I believe you have so many things to contribute. I'm looking forward to hear it back from you. Okay, that's going to be the email. And we're going to send this email, guys. Okay, no, it is sending the email. It will take few seconds. Maybe five or 10 seconds. Okay, the message has been sent. Okay, now let's check the inbox. And as you guys can see, we have received this email partnership proposal. And what we're going to do next is to go back to Zapier and we are going to run the automation. So let's click Run. Okay, now the automation is running. Okay, Zap runs successful, and it triggers on one new email. And let's go back to the email inbox again and we are going to refresh the page, and we will be able to see the draft. Okay, here we go. Hi, Bd. I hope you're doing well, too. It is great to hear from you again, and I appreciate you considering me for your supply chain optimization project. It sounds like a fantastic initiative, and I'm definitely interested in discussing how I might contribute to your tip. Could we set up a time to chat further? I'd love to hear more about your vision and see where I can fit in. Looking forward to your personal response, best regard, Chris. Okay, very cool. And if you want to send the email, just click SN. Okay, Fran cool. Okay, guys, I think that's it. That's all you need to know. I'll see you guys in the next video. Bye. 4. Building AI Customer Feedback Analyzer Automation with Zapier: Holly guys, welcome back to the course. In this video, we are going to build AI customer feedback analyzer automations using Zapier. It is a simple automations that will rate customer feedbacks from your Google Sheet file, and it will automatically analyze the sentiment and assign the sentiment label like positive, neutral, or negative. Okay, let's get started. Firstly, let's click Create Zaps and you will be redirected to this page. And let's click Trigger. And let's click on Apps. And let's click Google Sheets. Okay, now let's click Choose event. And I'm going to select New spreadshet Row. And we are going to connect Zapier with Google ****. Let's click Sign Int. And I'm going to select this email, and I'm going to check all these boxes, and let's click Continue. So in this case, I'm giving Zapier access to my Google Sheet. Let's click Continue. Okay, we have successfully connected Zapier to Google ****. Let's click Continue again. Okay, now, I'm going to open my Google Sheets, and I'm going to create a new spreadsheet. So let's click this, and I'm going to name this file customer feedback data. You can name it wherever you want. It is totally up to you. So we are going to store all the customer feedback data in this ****. Okay, now, let's go back to Zapier, and we are going to refresh the page. And we are going to click os value and let's click customer feedback data because we want to connect Zapier to this file. And for the worksheet, I'm going to select **** one. And let's click Continue. Okay, now, I'm going to click Test trigger. Okay, now, it is testing. As you guys can see, we got an error here unable to pull spreadsheet rows. The reason why we got this error because the spreadsheet is still empty. So what I'm going to do next is to use chat GPD to generate random customer feedback. And I'm going to copy and paste this and put it here. So in this case, we have three columns, date, customer name and feedback. And let's go back to Zapier and we are going to click Test trigger again. And this time it should work because the spreadsheet is no longer empty. So I'm going to select this one, and let's click Continue with selected record. Okay, next, we are going to add AI to the automations workflow. So let's click this, and I'm going to select Custom Pm. For the AI model, I'm going to use GPT four, but there are other options like nthropy Google Gemini. Feel free to use any AI model that you want. It is totally up to you. Next, we are going to add the input field. So let's click Add Field. I'm going to name the field feedback. You can name it whatever you want. It is totally up to you. Next, we are going to add the field value. So let's click this. And I'm going to select feedback. And let's click Save. Next, I'm going to write the prompt. Analyze customer feedback, sentiment. Assigned sentiment label like positive, negative, or neutral. Okay, now I'm going to enhance the prom, so I'm going to click this Improve Pm. So AI is going to modify the prom and make it better. Next, I'm going to generate output fields from the prom. So let's click this generate from output. Okay, it will take few seconds, maybe five or 10 seconds. So just wait, guys. Hopefully it doesn't take too long. Okay, we have two output fields, customer feedback and sentiment label. Next, I'm going to click Generate Preview. Okay, just wait, guys. It will take maybe ten or 15 seconds. Okay, here is the customer feedback, and here is the sentiment label. And let's click Finish. Next, we are going to add one more step. So let's click this, and let's click Apps, and I'm going to select Google Sheets again. Next, we are going to click Coos and event, and let's type in update spreadsheet row. And let's click Continue. Okay, now, we are going to select the spreadsheet. So let's click this. Obviously, we are going to select customer feedback data. Next, we are going to select the worksheet. So let's click Just value, and I'm going to select **** one. Well, in this case, we only have one ****, right? Look at this. Okay, let's go back to Zapier. Next, we are going to select the row. So let's click on these three dots and let's click Custom. And let's click New Spreadsheeo in Google, and I'm going to select row ID two. Next for the date, let's click this and I'm going to select date. For customer name, I'm going to select customer name. For feedback, I'm going to select feedback. And actually, we need to add one more column, guys. So let's add sentiment label column. And we are going to refresh the page As you guys can see, now we have one more column. The name of the column is Sentiment label. That's the column that we've just added. Let's click this and let's click Analyze and return data and we are going to select Sentiment label. In this case, the sentiment label will be added to this column. Let's go back to Zapier and you can change the color text. You can also make it bold or italic, it is totally up to you guys. You can customize it however you want. Next, we are going to click Continue. And let's click Test Step. Okay, now it is testing the automations. It will take few seconds. And let's click Publish. Okay, now we are publishing the automations. It will take approximately 10 seconds. Okay, we have successfully published the automations, and now the automations is running. And let's go to Google ****, as you guys can see. Now we have the sentiment label, and it's positive. The demand forecasting filter helped us reduce overstock by almost 18%, very impressed. Obviously, that's a positive feedback, right? Okay, very cool. Now we are going to test these automations again, so we're going to add one more row. So let's copy and paste the date. Let's copy and paste the customer name. And let's copy and paste the customer feedback. Well, it should be here. Actually, it is not going to be updated like instantly in 1 second. It will take maybe like five or 10 seconds or you can refresh the page and you will see the sentiment label here, guys. Okay, let's check this again. Well, it only took maybe 5 seconds or even less than 5 seconds. Okay, the sentiment label is negative. UI is confusing at first, and onboarding tutorials are too short. Took us weeks to get used to it. So obviously, this is a negative feedback, right? Okay, very cool. We've just built customer feedback analyzer automations using Zapier. Okay, I think that's it. That's all you need to know. I'll see you guys in the next video. Boy. 5. Building AI Expense Categorization Automation with Make: Hello, guys. Welcome back to the course. In this video, we are going to build a simple AI automations that will categorize your transactions or expenses using make.com. Okay, let's get started. Firstly, let's open Google Sheet, and let's create four columns. The first one is transactions date. The second one is transactions amount. The third one is descriptions, and the fourth one is category. And we are going to connect make.com with Google ****, so the AI model will be able to access your Google ****. Okay, let's log in. I'm going to log in using my Gmail account. As you guys can see, I have successfully logged in. Okay, here is my dashboard, and let's get started. I'm going to click this Make AI tools, and you have two options. You can build the AI automations from scratch, or you can also use one of these templates. There are many templates that you can use and customize. Okay, I'm going to start gilding the AI automations from scratch. Let's click this start Building, and you will be redirected to this page. This is the editing page. Okay, firstly, we're going to add Google ****. So let's click this and let's click Google *****. And I'm going to select What's Rose. Next, we are going to create a connection. So we're going to connect our Googleshit account with make.com. So let's click this. And I'm going to click Sign in with Google. And I'm going to select this account and let's click Continue. Let's check all these boxes. So in this case, I'm giving make.com to access my Google Sheet, and let's click Continue. Okay, it is still loading. It will take a few seconds. Next, we are going to select the spreadsheet. So let's click this. Click here to choose File. And obviously, I'm going to select expense data. That's the name of the file. So let's type in expense data. And for the sheet name, let's like sheet one. Well, in this case, we only have sheet one, right? And let's scroll down. Table contains header, and make sure that you like yes because we have this header, date amount descriptions and category and let's scroll up a little bit. For the search method, there are three options search by PAT, select from A, and enter manually. I'm going to select the first one search by PAT. Okay, let's click Save. And next, we are going to add another module. So let's click this. In this case, we are going to use AI Model provided by make.com. Let's type in AI and let's click this ask anything. This is free AI model provided by make.com, and you don't need API key here, guys. Next, we are going to select the AI model. There are three options GPT OSS 20 B, GPT OSS, 120 B, and GPT five Mini. In this case, I'm going to select the first one. Feel free to select any AI model that you want. It is totally up to you. Next, we are going to write the prompt. Assigned transactions or expense category based on descriptions. So let's click this and let's click Save. Okay, now we are going to add another module. So let's click this and let's click Google Shins. As you guys can see, there are multiple options. Let's, like, update a row. Let's click this for search method, let's select search by path. And let's click here to choose the file. And remember, guys, the name of the file is Expense data. So let's type in expense data. And there is only one sit inside Expense data. So let's click this **** one. For row number, I'm going to select row number. Let's scroll down a little bit. Table contents header. I'm going to put yes because obviously this data has header, right? So the first rope is the header. For date, we're going to select date. For amount, we're going to select amount for descriptions, we're going to select descriptions, and for category, we are going to select answer. Let's scroll down, and let's click Save. And let's open Google ****. As you guys can see, it is still empty. So I'm going to use chat GPD to generate random transactions or expense data, and I'm going to copy and paste this and put it here. Okay, next, we are going to run the AI automations. So let's go back to make.com, and let's click Run Once. Okay, now we are running the AI automations. It will take few seconds, maybe ten or 15 seconds. So just wait, guys. Okay, here we go. As you guys can see here, AI has added the transactions category. So for these transactions, the category is food. Okay, very cool. Actually, you have two options. You can run the automation once, just click on this button, or you can automatically run the automations every 15 minutes. You can activate this, just click this. So if you add more transactions to this Google Sheet, every 15 minutes, AI will update the cell. Okay, guys, I think that's it. That's all you need to know. I'll see you guys in the next video. Boy. 6. Building Inventory Tracking Automation with Make: Hello guys. Welcome back to the course. In this video, we are going to build inventory tracking automations using M. So it is a simple automations that will monitor your inventory data, and it will check if your stock quantity is below the reorder level, then it is going to automatically send you an email. So it is going to be a reminder for you to reorder the products. Okay, let's get started. Let's click this create scenario. And you will be redirected to the editing page. Okay, it is still loading. It will take few seconds. Okay, now let's click this, and we are going to choose Google Sheets. Okay, let's click this, and we are going to select what's new rose. So it is going to be triggered when a new row is edited. So let's click this in this example, we are going to use this inventory data. We have five columns, product name, SKU, stock quantity, reorder level, and locations. So basically, the AI automation is going to check if the stock quantity is below the reorder level. If that's the case, then it is going to automatically send an email. It is going to remind you to order the products. Okay, let's go back to make.com, and we are going to click this and we are going to connect Make with the inventory data. So let's type in the file name inventory data. And as you guys can see, there is only one **** here, St one. So let's click this and let's choose St one. And let's scroll down for the limit, we are going to set it to ten. And let's click Save. And let's choose all and let's click Save again. Next, we are going to click at another module. And let's type in AI. So we are going to use AI model provided by Make, so it is a free AI model and you don't need to have any API key. And let's click as Anythink. And we are going to filter the data first, so let's click this. And I'm going to name the label quantity check. You can name it wherever you want. It is totally up to you, and we're going to create a conditional statement. So we're going to choose stock quantity. And for the text operator, it will be less dent. And let's choose reorder level. Here is the conditional statement. If the stock quantity is less than the reorder level, then we are going to send the notifications. Let's click Save and we are going to write the email content using AI. Let's click this. For the AI model, I'm going to select GPT five. I think this is the best model, and now we are going to write the prom. Write a brief reminder email to reorder the product if the stock quantity is less than reorder level. Use these columns as reference. And I'm going to add three columns, product names, stock quantity, and reorder level. And let's click Save. And let's add another module. Let's click this. And let's type in GML. And let's click send an email. Okay, firstly, we need to connect our Gmail account with Make. So let's click Create Connections. Okay, now let's click this sign in with Google. And you will be redirected to this page. And I'm going to click this. This is the email account that I want to use, and let's click Continue and make sure that you check all boxes. So you are giving Make access to your inbox, and let's click Continue. Okay, we have successfully connected Make with GML. Okay, firstly, we are going to add recipient. So I'm going to add my other email account, Chris aldica@gmail.com. So the reminder email will be sent to this email. Next, I'm going to enter the subject LowSokRminder. So that's going to be the subject of the email. Next, we are going to add the content. So let's click this collections of contents. Let's scroll down a little bit. And for text, let's add answer. So this is going to be the AI generated email content. And let's click Save. Okay, guys, we are done with these automations. Now we are going to run the automation. So let's click this run once. So it will check if the stock quantity is less than the reorder level. If it founds one, then it will send an email. It is going to be a reminder email, guys. Okay, let's click this now we are running the auto missions. It will take a few seconds, maybe ten or 15 seconds. So just wait, guys. Hopefully, it doesn't take too long. Well, it didn't take too long, guys. It only took 10 seconds or even less than 10 seconds. Okay, here is the email from helpful Bd. That's my other email address. Subject reorder needed aluminum plates. As you guys can see here, for this product, aluminum plates, the stock quantity is only one, but the reorder level is 15. So obviously, the stock quantity is less than the reorder level. That's the reason why AI sent this email to me. Okay, guys, we have successfully built AI inventory tracking automations using make. And if you want these AI automations to keep running, you can click this run every 15 minutes. So if there is an update, then it will send you another email. Okay, guys, I think that's it. That's all you need to know. I'll see you guys in the next video. Bye 7. Building AI Appointment Booking Automation with n8n: Hello, guys. Welcome back to the course. In this video, we are going to build AI appointment booking automations using NightN. So this is a simple automations that will access your Google calendar book appointment, and send you an email. Okay, let's get started. Firstly, you will need to create a new account and you will get 14 days free trial. And let's click this and I'm going to log in using my Gmail account, and let's click Open Instance. And I will get redirected to my jazz board. Okay, here we go. And let's click Create Workflow. As you guys can see, we get redirected to the editing page. Okay, guys, you have three options. The first option is to build the automations from scratch. The second option is to build the automations using AI, and the third option is to build the automations from a template. There are many free templates that you can use and customize. In this case, we are going to build the automations using AI. So let's click this book with AI, and I'm going to write the prompt. Okay, here is the prompt. Create a workflow that starts with an appointment booking form, uses a manual workflow configurations note. Then store the appointment data in a sit, parse the time and prepare the data, create a calendar event, generate the email content, and finally send a confirmation email. I'm going to make this problem more specific. Appointment booking form, asking my client to input their names, email, and preferred time. And let's click Generate. Okay, now AI is creating the automations workflow. It will take a few minutes, maybe two or 3 minutes. So just wait, guys. Hopefully, it doesn't take too long, and I'm so excited to see the result. Well, it didn't take too long. It only took a minute or even less than a minute. Okay, here is the automations workflow. Firstly, we have appointment booking form, so our client will enter their name, email address, and also preferred time here. Next, we have Workflow configurations. So let's click this. And for the email, I'm going to replace the placeholder value with my own email. Okay, here we go. And I'm also replace the Sender email with my email. Next, we will need to add credentials for Google Sheets. So let's click this and let's click Create New credentials. And let's click Sign in with Google. And I'm going to select this account. And let's click Continue. So in this case, we are giving eight and access to Google ****. Okay, now we have successfully connected Google **** with eaten. Next, we are going to add credentials for Google Calendar. So let's click this. And let's click Create New credentials. And let's click Sign in with Google. I'm going to use this account, and let's click Continue. So in this case, we are giving eight and access to Google Calendar. Okay, we have successfully connected eight and with Google Calendar. And lastly, we are going to do the same thing for GML. So let's click this and let's click Create New credentials, and I'm going to sign in using the same Gmail account. So in this case, we are giving NAN access to or email inbox. Let's click Continue Okay, connection is successful. We have successfully connected GML with eight N. Okay, very cool. Now we are going to refresh the page. And let's click Reload. Next, let's open your Google Sheet, and let's create a new file. I'm going to name the file appointment Booking data. And this is going to be an empty Google Sheet. And let's go back to eight end, and we are going to connect eight n with this Google Sheet file. So let's click this store appointment data, and we're going to click From List and let's type in appointment booking data. And in this case, there is only one hit **** one. So let's add **** one here. For the operations, make sure you select a pen or update row. For the resource, make sure that you select **** with document. And make sure that you've already connected Google Sheets with NN that's what we did before, right? Okay, now let's click Save. Okay, now let's click Execute Workflow, and you will be redirected to the spade and let's enter the full name. I'm going to enter my name Chris Aldika let's enter the email address. I'm going to use my other email account. And let's enter the preferred date and time today at 5:00 P.M. And let's click Book appointment. AziGus can see the client data has been added to my Google **** automatically. Okay, Chris Altica, that's my name. That's my email address, and the preferred data and time is today at 5:00 P.M. Okay, very cool. Okay, now let's open the Gmail eight box, and we got a new email. Hi, Chris. This is a confirmations of your appointment on Tuesday, December 9, 2025 at 5:00 P.M. Click here to add it to your calendar. Looking forward to our meeting Best Regards. Okay, very cool. That's the confirmations email. And if you open your Google calendar, AZEGus can see AI has booked the appointment automatically. Okay, very cool. I think that's it. That's all you need to know. I'll see you guys in the next video. Bye. 8. Building AI Content Marketing Automation with n8n: Hi. Holly guys. Welcome back to the course. In this video, we are going to build AI Content Marketing automations using Naten. Okay, let's get started. Firstly, I'm going to click this Build with AI, and I'm going to write the prom. Build AI Content Marketing automations. That takes two inputs from the user. Product name and product descriptions. Based on those inputs AI will draft code email. Write Newsletter article. And store it in my Google Docs. Generate image and post it on Instagram. Create short form video using FiO. That's one of the best model for creating video guys. Then post that video on YouTube and Tik Tok. AI will also write Tweet and LncnPase and automatically pause them. For the AI model, use Mistral AI. You can use any large language model that you want, it is totally up to you. Okay, let's click Generate. Now AI is going to create the automations workflow. It will take a few minutes, maybe two or 3 minutes. So just wait, guys. Hopefully, it doesn't take too long. Well, it didn't take too long guys. It only took 2 minutes or even less than 2 minutes. Okay, here is the AI automation workflow for content marketing. Okay. Firstly, we have product input form. Let's click this. As you guys can see, we have two input. The first one is product name and the second one is product description. So the user will need to enter those values. Next, we have workflow configurations. So the data from the form will be stored her product name and product descriptions. Next, AI will write the cold email generator using Mistral AI, so that's the large language model that is being used here, and it will create the draft email. So it will be ready in your Gmail inbox. All you need to do is to click SN. Next, AI will write newsletter article, and it will automatically save the article in your Google Docs. Then after that, AI is going to use Gemini image to create an image and automatically post that image to Instagram. Then after that, AI is also going to generate video using VO. That's the AI bottle that we use, and it will automatically post the video on YouTube and also TikTok. So this is like a short form content, guys. Next, AI is going to write a tweet, and it will automatically post the twit. And lastly, AI is also going to write the Lincoln pose, and it will post that to Linkn. Okay, now let's scroll up a little bit and let's click on the Gemini image. And you will need to choose the model. There are many options. I'm going to select Gemini 2.5 flash image preview. This is nano Banana Model. Let's click Save and let's scroll down a little bit, and let's click on VO. You will also need to choose the model. There are many options. I'm going to select Model VO 3.1, generate preview. This is going to be a very short video, just like 15 seconds video guys. Let's click Save. Next, you will need to add credentials for your Google Docs. So let's click this, and let's click Create New credentials, and you will need to enter your client ID and client secret and just click Safe. Next, you will also need to add credentials for your Facebook. So you will need to connect your Facebook account with your Nightn. So let's click this. Let's click Create New credentials, and you will need to enter your accessToken. Next, you will need to add credentials for your YouTube account, so you will need to connect your Night N with YouTube. So just click this and you will need to add new credentials, and you will need to enter your client ID and client secret. And you will also need to do the same thing for Twitter and Linknt. Make sure that you add your credentials and connect your Twitter and Linknt to an Aten. Okay, guys, I think that's it. That's all you need to know. I'll see you guys in the next video. Bye. 9. Building AI Product Research Agent with Relay App: Holi guys. Welcome back to the course. In this video, we are going to build AI product research agent using Relay App. This AI agent will help you to conduct comprehensive product research and also summarize the research report. Okay, let's get started. Firstly, I'm going to log in using my Gmail account. You can also log in using your Microsoft account. It is totally up to you. As you guys can see, I have successfully logged in, and here is my dashboard. Okay, now let's click this U Work flow, and you will be redirected to the editing page. And let's click at trigger. And let's click Manual. Let's click at input. And there are many different data types like text, number, yes or no, single multi select email address, date and time, and still many more. I'm going to select text, and I'm going to name the perimeter. Product name. You can also add default value, and let's click DT. Okay, now let's click this at Step, and let's click AI, and let's click Research. Okay, now we are going to write the prom. I'm going to use this prom. AI agent will conduct product research for at product name, ter pricing, competitors, demand signals, review sentiment, and sourcing options from major marketplaces. Then it will generate a comprehensive product research report with viability score. Okay, so that's the prom. Next, I'm going to select the AI model. I'm going to use perplexity. That's one of the best options for product research. And there are other options like Open AI, Anthropy, Google Gemini, and X AI. Feel free to use any AI model that you want. It is totally up to you. And let's click this. Use this model. And let's click DT. And let's click this again and let's click AI, and let's click summarize. And obviously, in this case, we are going to summarize this research with AI. So let's click next for summarizing the product research report, I'm going to use Gemini 2.5 FAS, but there are other options like GPD 5.1 or CloudSnt 4.5. Feel free to use any AI model that you want. It is totally up to you. And let's click DT and let's click this again. And let's type in Google Docs. So we are going to save the product research report in Google Dogs. Let's click this and let's click Create Document. And let's click Connect. Okay, now we are going to connect our Google Docs with Relay App. Let's click on this email account. So in this case, we are giving Relay App access to Google Dogs. Let's click Continue. And I'm going to check all these boxes and let's click Continue again. Okay, we have successfully connected Relay App with Google Docs. Next, you will need to specify your target folder. I'm going to select my drive. Next, you will need to add the title. Product research. Next, you will need to add the content. In this case, we are going to have two contents. The first one is product research report. Let's click this research with AI, and let's click AI output. And the second one is product research Summary. At. Let's click this summarize with AI, AI output. And let's click Done. Okay, guys, we are done. Now we are going to run this AI agent. We are going to test this AI agent to make sure it is working. Okay, let's click this. Start runt. And I'm going to enter the product name. In this case, I'm going to conduct product research for memory foam plow. And let's click Start Run. Okay, now, the AI agent is running. It will take a few minutes, maybe one or 2 minutes. So just wait, guys. Hopefully, it doesn't take too long, and I'm super excited to see the result. Well, it didn't take too long, guys. It only took a minute or even less than a minute. Okay, let's open Google Docs, and here is the file product research. As you guys can see, AI has generated the product research report and also product research summary. Look at this. We have executive summary, market size and growth, pricing analysis, competitive landscape, demand signals, product innovation trends. What else? Regional analysis, challenges, viability assessment. So the viability score is 825 out of ten. We have the SWOT analysis, strength, opportunities, we have recommendations and conclusions. Okay, very cool, guys. Very insightful research report. I think that's it. That's all you need to know. I'll see you guys in the next video. Bye. 10. Building AI Supply Chain Contract Analysis Agent with Gumloop: Holigays, welcome back to the course. In this video, we are going to build AI supply chain contract Analyzer agent using Gumloop. This AI agent will be able to rate your contract assess if the contract is fair and analyze all potential risks. Okay, let's get started. Let's click this, and I'm going to log in using my GML account. As you guys can see, I have successfully logged in, and here is my dashboard. Okay, now, let's click this Create flow, and you will be redirected to the editing page. In this example, I'm going to use this potato supply chain agreement contract. So I use hat GPT to generate random contract. And we're going to download this as PDF. Okay, now let's go back to Gumloop, and let's click this start with a trigger. Okay, let's get started guys. Let's type in PDF, and I'm going to select PDF reader. Okay, now we are going to set the reading mode to standard. But if the contract has some images, you can also add OCR or optical character recognition, so it will be able to extract text from the image. Next, we are going to connect this PDF reader with AI agent. So let's type in AI agent. And let's drag this. And let's click Select Options. Let's click Create agent. And for the AI model, I'm going to select Cloud 4.5 Okus. You can feel free to use any model that you want. It is totally up to you, let's click Save. And let's close this. Okay, now we are going to connect the PDF reader with the AI agent. And we are going to start writing the prong. Analyze this supply chain contract. Let's drag the PDF contents and put it here. Assess the fairness and identify all potential risks. Okay, so that's the prom, a very simple prompt. Next, we are going to type in Google Docs. And let's dreg this Google Docs writer. Firstly, we are going to connect Gumloop with Google Doc. Let's click this authenticate and let's click Link, and I'm going to log in using this account and let's click Continue, and we are going to give Gumloop access to Google Docs. Let's click Continue again. Okay, now we have successfully connected Google Docs with Gumloop. Next, we are going to connect the AI agent with Google Docs. All the analysis will be added to Google Docs. Okay, here we go. I'm going to direct this to the right, and let's add the title. Contract Analysis. Result. Next, I'm going to add the content. So let's add the AI responses. And for the content format, you have three options Plain text, DML, and Markdown. I'm going to select plaintext. Okay, now we are going to test this AI agent. We are going to run this AI agent, and we're going to make sure if the agent is working as expected. Okay, let's click this. We're going to pick a file to upload, and we are going to upload this file, supply chain contract PDF. This is the same file that I showed you previously. And let's click this supply chain Contract PDF, and we are going to click Run. Okay, now we are running the AI agent. The flow has been started. It will take few seconds, maybe 20 or 30 seconds. So just wait, guys. Hopefully, it doesn't take too long, and I'm so excited to see the result. Well, it didn't take too long. It only took a minute. Okay, now let's open Google Docs, and let's open this file. Contract analysis result. This document was created by AI. Okay, here is the title supply chain contract Analysis. We have overall fairness assessment, and these are critical risks identified. There are many sections like scope of supply, price and payment, delivery and logistics, quality, standards. What else? Inspections, durations and terminations, confidentiality, liability, and still many more. And down here is the summary. The fairness is only one out of five, so it is severally one sided. The buyer protection is also one out of five. The legal soundness is two out of five, and here is the recommendations, reject and renegotiate. Okay, guys, I think that's it. That's all you need to know. I'll see you guys in the next video. 11. Building AI Receptionist Voice Agent with Eleven Labs: Hoy guys. Welcome back to the course. In this video, we are going to build AI receptionist voice agent using 11 lamps. So this AI voice agent will be able to talk with your customers and answer their questions. Okay, let's get started firstly. I'm going to log in using my Gmail account. As you guys can see, I have successfully logged in, and here is my dakbard. Okay, now let's click on agents, and you will be redirected to this page. You have two options. You can build the AI agent from Scratch, or you can use one of these templates and you can customize it. Okay, now let's click this start from blank. So we are going to build the AI voice agent from scratch. Okay, firstly, you will need to write the system prom. Okay, here is my prom. You can post the video and copy my prompt, or you can also write your own prom. It is totally up to you. Okay, let's read the prom. You are an AI receptionist for a barbershop, responsible for answering calls, treating customers warmly and understanding their appointment needs. Your job is to schedule booking, provide service information and ensure every color has a smooth and friendly experience. Okay, that's my prom, a very simple prompt. Next, I'm going to add the first message. This is the first message that the agent will say. Hello. How can I help you today? That's the first message that the AI voice agent will say to your customers. Okay, now let's click Save. Next, we are going to select the voice. Just click this and just click Voice. There are many options. In this case, I'm going to use Eric's voice. If you want to test the voice, just click on the Play button. His complaint Next, we are going to select the language. I'm going to select English, but feel free to use any language that you want. It is totally up to you. Next, we are going to select the AI model. There are many options like GLM, Quint, GPT, Gemini. I'm going to select Gemini 2.5, feel free to use any AI model that you want. It is totally up to you. Next, we are going to click this knowledge base. Let's click at document, and let's click Create Text. Okay. Okay, now we are going to provide general knowledge for the AI voice agent. Berber schedule. The Barber is opened from Monday to Saturday from 8:00 A.M. To 6:00 P.M. It's haircut sent is 30 minutes. The price is $25. And the customer have additional options for saving beard and head massage which will cost $10.15 dollar respectively. Currently, we have three hairdressers. Jon, James and Mark. John is available on Monday and Tuesday. James is available on Wednesday and Friday. Meanwhile, Mark is only available on Thursday. And let's click this Create text. So the AI voice agent is going to answer the customer questions based on this knowledge base. Okay, next, we are going to test the AI voice agent. So let's click this preview agent, and you will be redirected to this page. And I'm going to increase my volume a little bit, and I'm going to click on this button. Call AI agent. Hello. How can I help you today? Hey, I want to get a haircut. Certainly, I can help you with that. To schedule your haircut, could you please tell me which day and time you would prefer? Maybe Tuesday or Wetness Day. Who is available on Tuesday and who is available on Wetness Day? On Tuesday, John is available. On Wednesday, James is available. Do you have a preference between them, or would you like to know more about their availability? Well, actually, I'm going to do the haircut on Tuesday. So can you book the schedule for me? Yes, I can book that for you. So you'd like to schedule a haircut with John on Tuesday. Right. What time would be best for your appointment? We are open from 8:00 A.M. To 6:00 P.M. Maybe around four or 5:00 P.M. Is that okay? Yes, both 4:00 P.M. And 5:00 P.M. A available for a haircut with John on Tuesday. Which time would you prefer? 5:00 P.M. Alright, I'll book you for a haircut with John on Tuesday at 5:00 P.M. Thank you so much. Okay, guys, that's how AI Voice agent works. It was really cool. It feels like I was having conversations with Human. Okay, I think that's it. That's all you need to know. I'll see you guys in the next video. Bye. 12. Building AI Customer Support Voice Agent with Vapi: Holi gays, welcome back to the course. In this video, we are going to build AI Customer Support voice agent using FAPI. This AI agent will be able to talk with your customers and answer all questions that your customers have. Okay, let's get started. Let's click this OpenDSWord. And you will need to sign up. You can sign up using your GML account. You can sign up using your Github account. You can also sign up using your Discord account. In this case, I'm going to sign up using my GML account. As you guys can see, I have successfully logged in. And here is my daskbd. Okay, now, let's click on Assistance and let's click Create Assistant. And I'm going to name the assistant AI Customer Support agent. You can name it wherever you want. It is totally up to you. In this case, you have two options. You can build the AI agent from scratch. Just click this blank template or you can use and customize one of these templates. There are several templates like customer support specialist, lead qualification specialist, appointment scheduler, and still many more. In this case, we are going to use the blank template. Let's click Create Assistant. Okay, here we go. And let's scroll down a little bit. Firstly, you will need to select the AI model that you want to use. There are several options like open AI, Azure anthropic, Google, Gemini, Grock, Dipsik, Mistral, and still many more. I'm going to select Open AI, and I'm going to select GPD four cluster. Feel free to use any AI model that you want. It is totally up to you. For the first message mode, I'm going to select Assistance Speaks First. And the first message would be, hello. How can I help you? Next, we are going to enter the system prong. I'm going to use this prong. You are an AI customer support voice agent for accounting SAS platform, responsible for helping users with onboarding troubleshooting errors and explaining key features in a clear and friendly way. You should use the knowledge base to answer questions about invoicing, expense tracking, perl setup, subscriptions plans, integrations, data export and common error resolutions. Always confirm the user's issues, provide accurate step by step guidance, and escalate only when the problem cannot be solved through standard procedures. Next, I'm going to upload one page knowledge based document for my SS product. So this is accounting size knowledge base. We have three main features in voicing, expense tracking, and also payroll. So the AI agent is going to answer the customer questions based on this knowledge base document. Okay, let's click this. Let's click at New File, and we are going to upload this PDF file. And let's check this box. Okay, now let's click on FOIE it there are many different voices. I'm going to select Tara. You can use any voice that you want. It is totally up to you. Okay, next, we are going to click this transcriber. For the provider, I'm going to select DipGrem and for the language, I'm going to select English. Okay, now, we are going to test this AI voice agent. We want to make sure if this agent is working. Okay, next, we are going to click Publish. So we're going to publish the assistant. Let's click Next and let's click Publish. Okay, it has been published successfully. Next, we are going to try this AI voice agent. Let's click this Talk to Assistant. Okay, now it is connecting to the server. It will take few seconds, maybe ten or 20 seconds. It is still connecting with the server? Hello. How can I help you? Hi. I have questions about the software. How many feeders that you have? I'm here to help with specific features related to invoicing, Expense tracking, payroll setup, subscription plans, integrations, data export, and common error resolutions. If you have questions about any of these areas or need assistance with something specific, feel free to ask. Okay, I'd like to ask about expense tracking fitter. Can you explain a little bit more? Thanks. The expense tracking feature allows you to manage and monitor expenses efficiently. Here's a brief overview, receipt management. You can upload receipts, and the system uses OCR, optical character recognition to automatically extract details like fender Look, you guys, that was really cool. That was beyond my expectations, and it feels like I was having conversations with real human. I think that's it. That's all you need to know. I'll see you guys in the next video. Bye. 13. Building AI Sales Voice Agent with Voiceflow: Hello guys. Welcome back to the course. In this video, we are going to build AISLs Voice agent using Voiceflow. This AI agent will be able to talk with your potential customers or potential clients and answer all questions that they have. Okay, let's get started. Firstly, I'm going to log in using my GML account. Let's click this continue with Google. As you guys can see, I have successfully logged in, and here is my dust board. To build the AI foie agent, you have two options. The first option is to write the prom, and the second option is to use one of these templates so you can use and customize these templates. In this case, I'm going to write the prom, and in this example, we are going to build AI Sales voice agent for solar panel company. Okay, let's read the prom. You are an AI sales voice agent for a solar panel installation company. Your job is to qualify leads, explain the benefits of switching to solar and guide colors toward looking a free on site assessment. Use clear friendly language, ask targeted questions about of type, electricity bills, and locations and handle objections with confidence while staying focused on converting the lead. Okay, now let's click Generate. It will take few minutes, maybe two or 3 minutes. Just wait guys. Hopefully it doesn't take too long. Okay, here is the AI agent workflow. Firstly, we have lead qualifications routing. Next, the AI agent will clarify the colors intent or eligibility. It will perform lead qualifications, and it will also handle uninterested or ineligible color. Next, the AI agent will explain solar benefits and offer assessment. It will also handle unqualified leads. It will handle objections and book on site assessment. Okay, very cool. Now we are going to test this AI voice agent. So let's click this test your agent. Okay, it is still loading. It will take a few seconds. Hello. Thank you for calling. My name is Sarah, and I'm calling from Solo Max Solutions, a leading solar panel installation company. We're reaching out to homeowners in your area who may be interested in learning about how solar panels can help reduce their electricity bills and increase their home value. I'd like to ask you a few quick questions to see if you might be a good fit for our free solar assessment program. One, first, are you a home owner at this address? Two. Are you currently paying an electric bill of 75 or more per month? Three. Would you be interested in learning how solar panels could potentially? Okay, very cool, guys. You can also call the AI agent, just click this, but you will need to enter your phone number. And after testing this AI agent and making sure it is working, you can publish this AI Voice agent. Just click this publish. Look, guys, I think that's it. That's all you need to know. I'll see you guys in the next video. Bye. 14. Building AI Lead Generation Agent with Zapier: Holly guys, welcome back to the course. In this video, we are going to build AI lead generation agent using Zapier. So the AI agent will be able to analyze every single lead. It will be able to assess if the lead is qualified or not, and it will also follow up the leads and try to close that lead. Okay, let's get started. Firstly, you will need to log in using your Gmail account. Okay, now let's click Create Agent. In this case, you have three options. You can build the agent from scratch. You can use one of these templates. As you guys can see, there are many free templates that you can use and customize. And the third option is to use AI to build the agent. You just need to write the prom, describe your workflow, and in less than a minute, AI is going to generate you the AI agent workflow. In this case, we are going to build the agent from scratch. So let's click this start from scratch. And you will be redirected to the editing page. Firstly, I'm going to name the agent lead generations agent. You can name it wherever you want. It is totally up to you. Next, we are going to add the trigger. Let's click this and let's type in Google ****. You will need to create a Google Sheet. As you guys can see, I have five columns, name, email, company message, and budget. So in this case, the AI agent is going to analyze all leads from these Google Sheet. Okay, now, let's go back to Zapier and we are going to add a trigger to start the AI agent. So let's click this and let's choose the Google Sheets, and let's type an update, and we are going to select this one new or updated spreadsheet row. And let's type in the name of the file, which is agency ds data. And there is only one ****, **** one. So let's click **** one, and let's click Save. So in this case, the AI agent is going to extract data from this Google ****. So make sure that you are connecting Zapier with the correct file. Okay, now we are going to write the prompt. Okay, here is my prompt. Let's at the prompt. You are an AI lead generation agent for a video editing agency. Your job is to analyze every new lead from my Google ****. Assess whether they are qualified based on budget, project scope, and business potential, and label them accordingly. For each qualified lead, draft a personalized code email in my Gmail that introduced our services highlights relevant benefits and invites them to book a call. Next, we are going to add two additional tools. The first one is Google Sheet Update. Let's click Add tool and let's click Google ****, and let's type in Update. And we are going to select this one, Update spreadsheet throw. Then after that, we're going to add GML, and let's type in draft. Okay, next, we are going to scroll down a little bit. And if you have knowledge based document, you can add the file here. Just click at Koletsurce and you can upload the file, or maybe if you keep it in your Google Docs, you can connect your AI agent with Google Docs. Okay, now we are going to click this Agent Preview. And before running the AI agent, you need to make sure that you have already connected your Gmail with Zapier, and let's click Test agent. Okay, now we are running the agent. It will take few minutes, maybe one or 2 minutes. So just wait guys. Hopefully it doesn't take too long. Well, the AI agent was running for 3 minutes, and here is the result. So firstly, the AI agent extracts the data from the Google Sheets and it picks the leads Serakim. The company name is Bright Content. Let's scroll down. Then after that, the AI agent created the email draft to Serakim at bredcontent.co. So that's the leads email. And here is the email body. Look at this. Very cool, right? And here is the lead analysis. Lead assessment. And actions taken. Okay, very cool. Now we are going to enable this AI late generation agent. So let's click this. Okay, now, let's open the GML Inbox, and let's click DRF and let's click this. As you guys can see, the AI agent has created the email Draf for Sara. Hi, Sara, I noticed Brad Content Co is looking for a weekly video editing support for YouTube and shorts content as a video editing agency specializing in consistent high quality content productions. I'd love to help you streamline this process. Okay, very cool. And you just need to click SN. Okay, guys, I think that's it. That's all you need to know. I'll see you guys in the next video. Bye. 15. Building AI Data Analysis Agent with Python Llama: How you guys. Welcome back to the course. In this video, we are going to build AI Data Analysis agent using Python and Llama. The IDE that we are going to use is Google Colab, but if you want to use different IDE, maybe VS code or PyCharm, it is totally up to you. Okay, let's get started. Firstly, let's create a new file. Just click File and click New Notebook. Okay, now we are going to import several Python libraries. Let's import requests. We are going to use this library for handling API requests. Next, we are going to import JSOT or JavaScript object notations. We are going to use this library for JSON encoding. Then after that, we are going to import Pandas. We are going to use this library for data handling. And lastly, we are going to import Google Colab files from google dot Colab import files. We are going to use this library for enabling users to upload datasets to Google Colab. Okay, now, let's compile the code, and let's add another block of code. Next, we are going to get the API key from Open Router. For those of you who are not familiar with Open Router, it is a unified interface for Lord's language model. And don't worry it is completely free to use. Okay, now let's log in. You can log in using your Github account. You can also log in using your Gmail account. In this case, I'm going to log in using my Gmail account. It will take few seconds. As you guys can see, I have successfully logged in, and here is my dashboard. Okay, now let's click on your profile picture, and let's click Keys. And you will be redirected to this page. To create a new API key, just click this create APIKey. You will need to name the APIKey and just click Create. You can also set the expiration date, but it is optional. Okay, now let's go back to Google Cap. And let's create a new variable. I'm going to name it API key. Okay, this is my API key. So I'm going to copy and paste this and put it here. Please do not use my API key guys, because I will deactivate this API key once I'm done with this project. So make sure to use your own API key. Next, we are going to define the large language model that we want to use. Let's create a new variable. I'm going to name it model. And in this case, we are going to use Llama. So let's go to Open Router and let's type in Llama. We are going to use this model, Meta Llama 3.3 70 B instruct three. So let's click that. Okay, now we are going to open the code documentation. So let's click Quick Start, and let's scroll down a little bit. And we are going to copy and paste this. And let's put it here. So this is the Lord's language model that will perform the data analysis. Next, we are going to create a functions to generate a readable preview of the data frame without overloading the prom. I'm going to name the functions safe preview. You can name it wherever you want. It is totally up to you, and these functions has two parameters. The first one is the F, which is the data frame and the second one is maximum characters. And I'm going to set this to 3,000. Okay, next, we are going to convert the entire data frame into a clean text string without row indices. So let's create a new variable. I'm going to name it TXT. We are going to use the F two string and we are going to set the index to falls. Next, we are going to return full text if short. Otherwise, it is going to return the truncated preview. Returned TXT, if the length of the text is less than or equal to maximum characters. Remember, guys, previously, we've already defined the maximum characters to be 3,000. Es TXT, let's create a break it, and let's put max characters inside the breakePlus let's N dot, dot, dot, truncated. Okay, guys, we are done with these functions. Next, we are going to create another functions that will send the data frame to the AI model for analysis. I'm going to name the functions. Analyze CSV. You can name it wherever you want. It is totally up to you, and this function has two parameters. The first one is CSV path, so this is the file pad and the second one is goal. Okay, firstly, we are going to load the uploaded CSV file. Let's create a new variable. I'm going to name it DF, and to load the datasets, we are going to use pd dot D CSV. And let's copy and paste CSV path, and let's put it inside the parenthesis. Next, we are going to create a trim preview of the datasets. Let's create a new variable. I'm going to name it preview, and we are going to call these functions safe preview. And we are going to pass down the parameter, which is the data framed. So let's copy and paste DF and put it inside the parenthesis. Next, we are going to construct the instructions prom given to the AI model. Let's create a new variable. I'm going to name it prom. Okay, now let's write the prompt. You are an autonomous data analysis EI agent. And data analysts. Okay. Firstly, we want to specify the AI role. And we are also going to define the user goal. Okay, now let's copy and paste this variable goal and put it here. Next, we are going to define the rules. The first rule is to think step by step before acting. The second rule is to decide which analysis to run. And we are going to give the examples, summary, patterns analysis, anomalies analysis, correlations, suggestions. So in this case, the AI agent will be fully autonomous. It will be able to decide which type of analysis to run. Rule number three, explain your reasoning. Then give results and actionable recommendations. Step number four is keep answer concise and business focus. And we are also going to include the dataset preview. Possibly truncated. And let's copy and paste preview, and let's put that variable here. And we are going to close this. Next, we are going to send a post request to open Rotter chat completions and point. So let's create a new variable. I'm going to name it response. And we are going to utilize request dot ps. Okay, now we are going to add the API endpoint for chat based model completions. At TTP. Double opened Rotert AI, API one HAT, slash Completions. Okay, so that's the API endpoint. Next, we are going to define the headers. Okay, now we are going to include the API key for secure access. Authorizations. Brer. And we are going to copy and paste the API key and put it here. Next, we are going to specify that the request body is Jacint. Content type applications JasonT. And let's add Coma Let's add comma again. Next, we are going to convert Python dictionary into Gson payout. Data JsontTms Okay, now we are going to specify which AI models you'll process the prom. We are going to set the model to Llama. So let's scroll up and let's copy and paste this and put it here. Next, we are going to provide the user message containing analysis instructions. Messages. And let's create a break it. We are going to set the role to user. Next, we are going to set the content to prompt. Let's scroll up a little bit, and let's copy and paste Sproum and put it here. And let's add coma. Next, we are going to set the maximum time to wait before request is canceled. We are going to set the time out to 120. Then after that, we are going to throw exceptions if the API request failed. Response raise for status. Next, we are going to extract AI generated analysis and returned with data frame. Return response dot Jason and let's create a break it and let's put choices inside the break it. Let's put zero inside the second break it. Let's put message inside the third breaket and let's put content inside the fourth breaketTF DF is the data frame guys. Next, we are going to display instructions prompting user to upload the CSV file. Print upload a CSV file. After uploading, the AI agent will run. Okay, so that's the instructions for the user. Next, we are going to create a new Variable. I'm going to name it uploaded. And to enable user to upload file to Google Colab, we are going to use files dot Upload. Next, we are going to create a conditional statement to check if the file was uploaded, if not uploaded. Then we are going to display a warning message, raise system exit. No file uploaded. Rerun the cell and upload CSV file. So that's going to be the message. Next, we are going to get the name of the first uploaded CSV file. Let's create a variable. I'm going to name it CSV file. Next, iter and let's copy and paste upload it and put it inside the parenthesis. Then after that, we are going to request user to enter the analysis objective. Let's create a variable. I'm going to name it goal. To get user input, we are going to use input functions. Enter one line Objective for the AI agent. Next, we are going to add dot strip Then after that, we are going to execute the analysis using the uploaded dataset and user defined objective. Let's create two variables. I'm going to name it Pport and D F. And we are going to call these functions, analyze CSV. And remember, guys, this function has two parameters. The first one is CSV pad, that's the file pad and the second one is the goal. So let's copy and paste CSV file and put it here, and let's copy and paste go and put it here inside the parenthesis. Next, we are going to add section header for the final AI generated report. Print. So let's end AI Data Analysis Report. Let's end. Next, we are going to print the analysis. Print report. Let's copy and paste this variable and put it here inside the parenthesis. Next, we are going to add the file name that will be used to save the analysis, report to do this. Let's create a new variable, and going to name it out named. AI data report TxD. So the file format will be TXT. Next, we are going to open textFile for writing using UTF eight encoding with opened, let's copy and paste name and put it inside the parenthesis. That's a W, and we are going to set the encoding to UTF eight. As F next, we are going to write the user analysis goal to the file F dot right. Okay, now we are going to add the goal, plus let's copy and paste goal and put it here inside the parenthesis. N. Next, we are going to write the AI generated report to the file. F right report N. Make sure the is lowercase, plus let's copy and paste report, and let's put it here inside the parenthesis. Plus N, N. Next, we are going to write a readable preview of the datasets for reference. F dot right DidaRview. So let's end. And we are going to call the saf preview functions. And let's put DF inside the parenthesis. Remember, guys, DF is the data frame plus N. Next, we are going to notify the user that the report file is ready. Print end report save to let's copy and paste Odd Name and put it here. Download below if you want. And lastly, we are going to trigger the browser download of the generated report. Files dot download. And let's copy and paste Out name, and let's put it here inside the parenthesis. Okay, guys, we are done. Before compiling the code, there is one mistake that we need to fix. There was a typo here, guys, we need to remove the safe preview. Okay, next, we are going to test this EI agent using this sample data. So I have these small datasets with five columns, supplier name, product, on time delivery rate, defect rate, and lead time days. So we're going to use the AI data analysis agent to analyze this supplier's performance. Okay, let's compile the code. Okay, now the AI agent is running, we are going to upload the CSV file. Let's click Open. Next, we are going to define the goal for the AI agent. Find key insights. And let's hit Enter. Okay, now the AI agent is running. I will take few seconds, maybe five or 10 seconds. Well, it didn't take too long, guys. It only took 30 seconds, and here is the analysis report. To find key insights, I will follow a step by step approach. First, I need to decide which analysis to run. Considering the dataset contains information about suppliers, products, on time delivery percentages, defect rates, and lead times, I will run a correlation analysis and look for patterns. This will help to identify relationship between these variables and potential areas for improvement. My reasoning is that understanding the correlation between on time delivery defect rate and lead times can reveal suppliers or product performance trend. Okay, very cool. And these are the key insights and actionable recommendations. Prioritize suppliers with high on time delivery rates and low defect rates such as Alpha C with product A and C, which have high on time delivery rates. 96% and 95% and low defect rates, 2% and 3%. Optimized lead times, focus on reducing lead times as shorter lead times are associated with higher on time delivery rates and lower defect rates. Monitor and improve supplier product combinations, identify and address specific combinations with low on time delivery rates or high defect rate, such as Beta Mart with product C, 87% on time delivery and 6% defect rate. Okay, very cool, guys. We have successfully built AI Data Analysis agent using Python and Llama. I think that's it. That's all you need to know. I'll see you guys in the next video. Bye. 16. Building AI Multi Agent System for Project Management: Hello, guys. Welcome back to the course. In this video, we are going to build AI multi agent system for project management. We are going to have three agents, Taskop agent, workload agent, and project timeline maker. So these three AI agents will interact with each other. Okay, let's get started. Firstly, we are going to import two Python libraries. We are going to import requests. So this is the AtTTP library that will be used to send requests to the Open Router API. Then we are also going to import JSON. Let's compile the code, and let's add another block of code. Next, we are going to add APIkey for authenticating requests to open Router API. Let's create a new variable. I'm going to name it APIKey Let's go to the previous file and we are going to copy and paste this API key. This is my API key guys, do not use my API key. Make sure to use your own API key. Next, we are going to create a list of team members with their skill sets and available hours used for task assignment logic. I'm going to name the list team. You can name it wherever you want, and it is totally up to you, and let's create a break it. Okay, let's add the first employee. I'm going to set the name to Alice. Next, we are going to define her skill sets. The first one is front end and the second one is react. Just a random skill sets, guys. Next, we are going to set her hours. 20 hours per week. Let's add another employee. I'm going to set the name to Bob. Next, we are going to define his skill sets. Big end and Pitnt And he will be working for 25 hours per week. And let's add one more employee. I'm going to set the name to Cara. Just a random name, guys. Next, we are going to define her skill sets. QA or quality assurance and testing. And she will be working for 15 hours per week. And let's add coma. Next, we are going to prompt the user to enter a short descriptions of the project. Let's create a new variable. I'm going to name it project. And to get user input, we're going to use input functions. Describe the project. Next, we are going to copy and paste these three EI agents, Tscope agent, workload agent, and project timeline maker. So let's copy and paste them and put them here. Okay, now I'm going to explain one by one. So the first EI agent is task coop agent. Let's read the prom. You are a Tascope agent. Break project into task and estimate hours per task. Return only just on so this AI agent will take the input from the user and it will break down into smaller pieces and estimate the durations to complete its task. And the second AI agent is Workload agent. Let's read the prom. Workload agent, assigned tasks to team members based on skills and available hours, written only just on format. This EI agent will assign WTS task is suitable for Alice. WTS test is suitable for Bob, and WTS test is suitable for Kara. And the third EI agent is Project timeline Maker. Let's read the prom. You are a project timeline maker. Create a realistic timeline for the project based on the task and assign work loads. So this EI agent will be able to create a comprehensive project timeline based on the complexity of the project and available resources. Okay, let's continue. Now we are going to create a main controller prom that coordinates agents, instruct reasonings, and requests final human readable summary. Let's create a new variable. I'm going to name it controller prom. Okay, let's write the prong. You are the main controller. And we are going to define all the agents that are available. The first one is task scope agent. So let's copy and paste this and put it here. The second one is work load agent. Let's copy and paste this and put it here. And the third one is project timeline Maker. Let's copy and paste this and put it here. Okay, I'm going to scroll down a little bit. Next, we are going to define the rules. The first one? Think internally and print your reasoning step by step. Rule number two, decide yourself when to call its agent. Rule number three. After reasoning, output final summary, not as Jason, but as bullet points. And the summary should include test estimates. Assignments. And timeline. Next, we are going to define the project. Let's scroll up a little bit, and let's copy and paste this variable, and let's put it here. Then after that, we are going to define the team. Let's scroll up a little bit and let's copy and paste this list, and let's put it here. Next, we are going to define the agents. The first one is Tescope agent. So let's copy and paste this. Let's copy and paste this again and put it here inside the brace. And we're going to copy and paste this part. So in this case, we embed the Tescope agent instructions block for the controller from the code. And the second one is workload agent. So let's copy and paste this. Well, it is going to be exactly the same format, so you can copy and paste this line, and we are going to replace the Tescope agent with workload agent. The third one is project timeline Maker. Let's copy and paste this and put it here, and the format is going to be exactly the same. So let's copy and paste this, and we are going to replace Workload agent with project timeline Maker. Next, we are going to close the prom. Then after that, we are going to send a post request to the Open Router chat completions endpoint with controller prom and reasoning enabled. Let's create a new variable. I'm going to name it response, and we are going to utilize request dot post Next, we are going to add the API endpoint. Let's go back to this file, AI Data Analysis agent, and we are going to copy and paste this URL at DTPSUen Router, AI, slash API, slash V one, slash chat, slash completions, and let's put it here. Next, we are going to add the headers. So let's go back to this file again, and we are going to copy and paste the headers. And let's put it here. Just a quick reminder. So this is the bearer token header for the API authentications. And we also declare the JSON payload content type here. Next, we are going to add JSON payload that will include the model selections and message history. Let's create a new variable. I'm going to name it data, and we are going to use JSON dot dooms. And let's create a parentheses Okay, now we are going to specify the AI model that we want to use. In this case, we are going to use Gemma. So let's go to open Router, and let's type in Gemma. Tree end. Let's click this, and let's click Quick Start, and let's scroll down a little bit. And we are going to use this model, guys. So let's copy and paste this, and let's put it here. The name of the model is Google Gemma three and e4b IT three. And let's add comma. Next, we are going to add single user message that will contain the full controller instructions. Messages. Let's create a break it. Okay, now we are going to set the role to user. Then after that, we are going to set the content to controller prom. Let's scroll up a little bit, and we are going to copy and paste this controller prom. And let's put it here. Let's add comma after the break it. Next, we are going to request that the model prints or expose its intermediate reasoning steps. We are going to set enabled to true. And we are going to parse the TTP response body directly as JSON into the response variable. So let's add dot Jason. Next, we are going to print the header end project management output. End. Then after that, we are going to create a conditional statement if choices in response So in this case, we are going to check for the usual open router or chat completion structure containing choices. Print, response. And let's put choices inside the break it. Let's put zero inside the second break it. Let's put messages inside the third break it, and let's put content inside the fourth break it. Else. Then we are going to print the fallback message that indicates the expected structure was not present. Full response. So let's end. Next, we are going to do one more thing. We are going to print the row response Jason for debugging and inspections. Print Jason dot Dums let's put response inside the parenthesis, and we are going to set the indent to two. Okay, guys, we are done. Okay, before compiling the code, we need to modify the controller prom a little bit. So we need to have at least one line space after team, and we need to have one line space after agents. Okay, that's it. And let's compile the code. Describe the project. Property management system. So that's the project that I want to build. Let's hit Enter and it will take a few seconds for the AI agent to run. So just wait, guys. Well, it didn't take too long, guys. It only took 40 seconds. Let's scroll down, and here is the result. Okay, I'm the main controller for the property management system project. Here is my plan. Task breakdown and estimations. First, I need to breakdown the project into smaller manageable tasks and estimate the time required for each Okay. Here is the task estimate. Design database schema. 8 hours, develop big end APIs. User authentications, property management and payment 60 hours, develop front end user interface 80 hours. Implement unit test 30 hours, perform system testing and QD 40 hours and deployment and infrastructure setup 12 hours. Okay, let's scroll down. Now that I have the test estimates, I will call the workload agent to assign them to the team members. Okay, so here is the first assignment, design database schema. The test is given to Alice. Develop BC and APIs. The test is given to Bob. Okay, let's scroll down. Finally, I will use the assignment to create a timeline. Okay, so here is the project timeline, and here is the project summary. Very cool. Ooky, guys, we have successfully built AI multi agent system for project Mateen. I think that's it. That's all you need to know. I'll see you guys in the next video. Bye. 17. Building AI Food Inventory Vector Database with Pinecone: Hello, guys. Welcome back to the course. In this video, we are going to build AI food recipe generator, and additionally, we are going to connect the AI model with Pinecone vector database. So the AI model is going to generate food recipe based on available food ingredients. Okay, let's get started. Firstly, we are going to install two Python packages. The first one is Pinecone. Let's copy this, Pip install Pinecone, and let's compile it. Next, we are going to install Mistral AI. Let's copy this. Deep install Mistral AI and let's compile it. It will take few seconds, maybe five or 10 seconds. So just wait, guys. Hopefully it doesn't take too long. Okay, we have successfully installed Pinecone and Mistral AI. Let's add another block of code. Okay, now we are going to import several pod libraries. Firstly, we're going to import Pinecone and several spec from Pinecone. Import Pinecone. Make sure it is uppercase, comma. Serverless spec. We are going to use these libraries for creating and managing vector indexes. Next, we are going to import Mistral AI from Mistral AI, import Mistral, Make sure it is uppercase. This is the AI model that we are going to use, and we are also going to import NumPi. Import NumPi as NP. We are going to use this library for generating random factors that will be used as embeddings. Alknex we are going to get the Pinecone APi key. Let's create a new variable. I'm going to name it Pinecone API. Then you will need to go to Pinecone website and just log in. Let's click this you can log in using your Gmail account. You can log in using your Github account. You can also log in using your Microsoft account. It is totally up to you, but in this case, I'm going to log in using my Gmail account. As you guys can see, I have successfully logged in, and here is my dashboard. To create a new API key, just click this API keys, and you will be redirected to this spade and just click APIKey and you will need to name the API key. For example, Vector Database, and just click Create Key. Okay, let's go back to Google Colab. So this is my Pinecone API key. I'm going to copy and paste this and put it here. And don't use my API key, guys. I will deactivate this API key once I'm done with this project. So be sure to use your own API key. If you use my API key, it will not work. Next, we are going to get Mistral AI API key. Let's create a new variable. I'm going to name it Mistral key. Okay, now let's go to Mistral website and let's click Try AI studio. You will need to log in. You can log in using your Gmail account, and let's click API keys. And let's click Create New Key. You will need to name your API key and just click Create New Key. Okay, now, let's go back to Google Colab. So here is my Mistral AI API key. So I'm going to copy and paste this and put it here. Do not use my API key, guys. Use your own API key, alright? Okay, let's continue. Now we are going to initialize Pinecone client using APIKey. Let's create a new variable. I'm going to name it PC. So PC stands for Pinecone client. And we are going to utilize this library, Pinecone. And we are going to set the API key to Pinecone API. Let's copy and paste this and put it inside the parenthesis. Next, we are going to add the name of the vector index where ingredient factors will be stored. Let's create a new variable, I'm going to name it index name. And the name will be inventory. Next, we are going to create a conditional statement to check if the index does not already exist on Pinecone. If index name not in PCLs indexes, dot names, then we are going to create a new index with specified settings if it doesn't exist, pc dot Create Index. Next, we are going to assign the name to the index. We are going to set the name to index name. Let's copy and paste this and put it here. Remember, guys, the name is inventory. Next, we are going to set the vector dimensions. We are going to set the dimensions to 384. And remember, it must match vector size that we will insert. In addition, we are also going to use cosine similarity search closes gradient vectors. So we're going to set metric to cosine. Next, we are going to configure Serverless deployment on AWS or Amazon web surface and US is Regent. Spec. We are going to use this library, Serverless Spec let's copy and paste that. And we are going to set Cloud to AWS we are going to set the regent to US is one. Okay, let's continue. Next, we are going to connect to the existing or newly created Pinecone index. Let's create new variable. I'm going to name it index pc dot index. Let's copy and paste index name, and let's put it inside the parenthesis. Then after that, we are going to initialize Mistral client for chat completions. Let's create new variable. I'm going to name it client, and we are going to use Mistral library. So let's copy and paste this and we are going to set the API key to Mistral key. Let's copy and paste this and put it inside the parenthesis. Next, we are going to define a simple dictionary representing ingredients and their quantities. I'm going to name the dictionary inventory. Okay, I'm going to add random food to the vector database. So we have 12 eggs, two chickens, one spinach, five garlic, three rice, and four carrots. Okay, now we are going to create a for loop. We're going to look through each ingredient and its quantity. Four ingredient quantity and inventory. So let's copy and paste this dot items. Then we are going to generate a random 384 dimensions factor to simulate an embedding. Let's create a new variable. I'm going to name it factor, and we are going to use Numpai library np random dot RND. And let's put 384 inside the parenthesis dot TLS. Next, we are going to upload the factor, and we are also going to upload the metadata into Pinecone using ingredient name as ID. Index dot ASED let's create or break it inside the parenthesis, and let's copy and paste the ingredient and put it inside the parenthesis. Comma factor. Remember, guys, ingredient and quantity is the index of the four loop. And we're going to set the ingredient to ING. That's the first index, and we are going to set the quantity to QTY. That's the second index of the four loop. Okay, next, we are going to confirm a successful data upload to Pinecone. Print inventory uploaded to Pinecone. But let's scroll down a little bit. Next, we are going to create a functions that will generate a recipe using user requests and available ingredients. I'm going to name the functions, generate recipe. You can name it wherever you want, and it's totally up to you, and this function has one parameter, and that's requests. Okay, now we are going to create a random query vector to ft nearby vectors. Let's create a new variable. I'm going to name it query effect. Np dot random dot Rent, and let's put 384 inside the parenthesis dot tols. Next, we are going to retrieve top ten closest ingredient vectors with metadata. Let's create a new variable. I'm going to name it RS. And in this case, we are going to use index dot query. And we are going to set the vector to query ft. So let's copy and paste this and put it here. Next, we are going to set top k to ten. Because we want to retrieve the top ten clauses gradient vectors. And lastly, we are going to set include metadata to True. Okay, let's continue. Now we are going to format ingredient names and quantities into readable lists. Let's create a new variable, I'm going to name it available. And let's create a break it. Inside the break it. Let's create a brace. And let's put metadata inside the first break it. Let's put ingredient inside the second break it. And let's create a parenthesis. And let's put metadata again inside this break it. And let's put quantity inside the second break it. Next, we are going to create the four loop. For and res and let's put matches inside this break it. Okay, now we are going to write the prompt. Let's create a new variable. I'm going to name it prompt. So we are going to build a prompt instructing Mistral AI to generate a recipe using only the retrieve ingredients. Okay, let's write the prompt. Create a recipe based only on these available ingredients. Let's copy and paste this variable available, and let's put it here. So we store all available food ingredients in this variable. Next, we are going to define what the user wants. Let's copy and paste requests. And let's put it here. Do not include ingredients that are not listed. And we are going to close the prompt. Okay, next, we are going to call Mistral hat Model to generate recipe text. Let's create a new variable. I'm going to name it response. We are going to use client dot chat dot complete next, we are going to define the AI model that we want to use. We are going to set the model to Mistral small latest. Next, we are going to send Pm as user message. Messages, let's create a break it, and we are going to set the role to user. Then after that, we are going to set the content to prompt. So let's copy and paste prompt and put it here. Next, we are going to extract the generated recipe from the response. Returned response choices. Let's put zero inside the break it dot message dot content. Okay, guys, we are done with these functions. Okay, next, we are going to test this AI food recipe generator to make sure it is working. Let's create a new variable. I'm going to name it recipe, and we are going to call these functions Generate recipe. And let's stipend the user request. Something healthy for dinner. Then after that, we are going to print the header. Send AI recipe. Send. And lastly, we are going to display the generated recipe to the user. Print. Let's copy and paste recipe and put it inside the parentheses. Before compiling the code, there are a few mistakes that we need to fix. The first one is a typo, guys. So if you scroll up, this is supposed to be ingredient with E, not with I. So Make sure that you fix that. Also fix this one, and let's scroll down. Also fix this one. So Make sure that you fix this typo. Next, we are going to fix this one. Let's close the brace and let's add the string here. And let's close the brace here. Additionally, we are going to comment these two lines because those two lines were only for simulation, so we don't need them. And we are also going to replace this line with the new one. So we are creating a for loop ingredient and quantity. Remember, guys, these are the index of the for loop and inventory dot items. Okay, we are done, and let's compile the code. Let's click this Okay, now it is running. It will take few seconds, maybe ten or 20 seconds. So just wait, guys. Hopefully, it doesn't take too long. Okay, here is the food recipe. It didn't take too long, guys. It only took 10 seconds. Here is a healthy and balanced dinner recipe using only the available ingredients, curling chicken with spinach and rice. The preparation time is 10 minutes, and the cooking time is 25 minutes. Okay, here is the ingredients, and here is the instructions, and here is the reason why it is healthy. And if you opened your Pinecone vector database and just click database and just click Index. As you guys can see, we have successfully created this food inventory. Let's click that, and let's scroll down a little bit. You will be able to see all the food ingredients like carrot, rice, egg, garlic, chicken, and spinats. Okay, very cool. I think that's it. That's all you need to know. I'll see you guys in the next video. 18. Connecting Zapier MCP Server with Google Sheets: Hi, guys. Welcome back to the course. In this video, we are going to build Model Context Protocol server using Zapier. Additionally, we are going to connect MCP with Mistral AI. Let's get started. Let's click this Start Building. Make sure you've already logged into Zapier. You can log in using your Gmail account. Okay, now, let's click this new MCP server. Okay, now we will need to select the MCP client. Let's click this. There are many options like Cloud, Microsoft copilot, Julius AI, Mistral AI, Jet GPT, Manus, and still many more. I'm going to select Mistral AI, but you can select any AI assistant that you want. It is totally up to you. Next, you will need to name the MCP server, and let's click Create MCP server. Okay, we have successfully created the MCP server. Next, we are going to Add Tools. Let's click this Add tool. There are many tools available like Gmail, Google Calendar, Google Sheets, Google Docs, Google Form Notion, Google Drive, and still many more. I'm going to select Google Sheets. Let's click this and let's type and create. And I'm going to select this one. Create spreadsheet and let's click Connect. Actually, we need to add two more tools. So let's click Out Tool, and let's click Google Sheets, and let's type in Update. And let's check these two boxes. And let's click Add two tools. In total, we have three tools. Tree spreadsheet, Update spreadsheet through, and update spreadsheet rooms. Next, I'm going to click Connect Okay, now we are going to connect Zapier MCP with Mistral AI. Let's click this copy URL, and let's click this link and you will be redirected to the page. Make sure you've already logged into your Mistral AI. You can log in using your Gemma account, and let's click at Connector. And let's type in Zapier. And let's click at copy and paste the URL here, and for the authentication methods like out 2.1, and just click Connect. Well, in my case, Zapier has been ddt, so let's click this and let's click Connect. I'm going to click Low, so I'm going to give Zapier access to Mistral AI. As you guys can see, now, Zapier has been connected, and it is also important to make sure that you have already connected your Zapier with Google *****. Okay, now we are going to try this MCP server and make sure it is working. Let's click NewhaT and now we are going to use Dzepier MCP. Let's click this and we are going to type in the prompt. Create a sample dataset for Logistics, five columns and 15 rows and store the data in my Google ***** And let's hit Enter. And let's click Continue. Now we are giving Zapier access to Google *****. It didn't take too long, guys. It only took 10 seconds or even less than 10 seconds. So let's open the Google Sheets As you guys can see, Mistral AI has successfully created a new Google Sheet file. The name of the file is logistic data, and we have five columns, shipment ID, origin, destinations, weight and delivery status. Very cool, right? So you will be able to use Mistral AI to modify these Google Sheets. Okay, I think that's it. That's all you need to know. I'll see you guys in the next video. Bye. 19. Conclusion & Summary: Holly guys, welcome back to the course. In this video, I'm going to summarize all things that we've learned in this course, and I'm also going to share a few tips and tricks on how to improve the quality of your AI automations and AI agents. Okay, let's get started with the first topic, error handling. You need to design your workflow to be able to handle API failures, timeouts and unexpected responses. You need to build fallback mechanisms. So when your AI surface is down or return poor results, the system automatically retries, switches to a backup model, locks the issue, and notifies you without interrupting the entire automations. If you remember, in one of our projects, when we built AI Data Analysis agent, we created a conditional statement right here. So it will check if the data has not been uploaded, then it will raise an error message. No file uploaded, re run the cell and upload the CSV file. So this is one of the examples of error handling. Okay, now let's go back to the min map, and let's talk about the second topic. API cost optimizations. It is very important to monitor API usage because AI automation can become unexpectedly expensive if talking counts request frequency or model size are not controlled. Optimizing costs through tching requests, truncating context, sketching results, and choosing the right model insures scalability without wasting budget. So there are some AI models that are more expensive. There are some AI models are keeper for example, Cloud and Open AI are more expensive compared to Gemini. So make sure that you are fully aware of that. And if you are using open Rotor API, you can go to Activity TAR, and you will be able to track or monitor your API usage. You can see which AI model that is being used. You can also see how many API requests that you have made. Okay, now, let's go back to the mind map, and let's talk about the third topic, context and prom engineering. It is very crucial to have strong contexts and prompt engineering, so the AI model receive only the information that matters and produce consistently relevant responses. Structuring prom with clear roles, objectives, constraints and examples dramatically improves agent accuracy, reduce hallucinations and makes workflow more dependable. If you remember, in this project, when we build AI Data Analysis agent, before we are asking AI to do anything, we define the role. Look at this. You are an autonomous data analysis EI agent and data analyst. This is the first thing that we need to do before asking AI to perform task. And it is exactly the same thing like what we did in this project, AI project manager agent. Before asking AI to do anything, we define the role. Look at this. You are Tescope agent. You are workload agent. You are project timeline maker. You are the main controller. Okay, guys, this is the end of the course. Thank you so much for completing the course, and I wish you all the best. Bye.