How To Build WhatsApp AI Agent Using n8n | WhatsApp Automation With n8n | n8n Tutorial | Simplilearn

Imagine [music] this. Your customers are messaging you on WhatsApp all day asking for prices, support, order status, appointments, and hundreds of other things. And most of these messages are either delayed or never answered properly. Now, here's the crazy part. WhatsApp messages have highly insane engagement. On average around 90 to 98% of the WhatsApp business messages actually get opened which absolutely crushes traditional channels like email that sit around 20%. So if you can automate even a part of those conversations with AI you're basically unlocking a 24x7 high conversion communication channel for your business or your clients. That's where AI agents come in. Instead of a static FAQ bot that only understand a few keywords, an AI agent can actually understand what the user is saying in a natural language, keep track of the conversation, and respond it in a human-like way. It can qualify leads, answer support questions, book calls, and even trigger other automations in the background without you touching a single line of backend code. In this video, I'll show you how to build exactly that. A WhatsApp AI agent powered by N8. We'll connect your WhatsApp number to N8N. Plug in an AI model like chatd. Add memory so it remembers context and then write everything together into a clean visual workflow. By the end, you'll have your own AI assistant running inside WhatsApp, replying to real users in real time. In the next part, you'll walk through the exact tools and accounts you need so you can follow along step by step. Before we dive into the technical details, here's something genuinely exciting for anyone who is serious about building the next generation of AI products. The applied agentic AI system design impact program in collaboration with Microsoft and simply learn is a 10-week practitioner level experience designed for professionals who want to move beyond basic generative AI and start designing orchestring and leading agentic AI systems. You'll gain hands-on expertise in multi- aent systems, rack, mcp, planning framework, workflow automation, and agentic UX while working with 25 plus industry tools like Langchain, Autogen, QAI, N8N, Langraph, Azure, and more. Learning is deeply practical with 40 plus demos, 10 plus guided practices, seven realworld projects, and a capstone focused on product strategy delivered through live interactive sessions, mentoring, and even cohort-based collaboration. This program also offers career support via job assistant, Microsoft learn badges, and a joint completion certificate from Microsoft and Simply Learn. If you're aiming to build autonomous AI systems, lead AIdriven product strategy and stand out in the fast growing market. This program isn't just an upgrade, it's your launch pad. Enroll now and step into the future of Agentic AI with confidence. The link is given in the description box below and in the comment section. Before we begin, here's a small question for you to answer. What is the biggest advantage of using WhatsApp AI agent for your business? It reduces internal usage. It replies instantly to your customer 24x7. It replaces your website completely. It only works during business hours. Let me know your answers in the comment section below. Now, let's dive in. Today, we are building a 24x7 WhatsApp AI assistant powered by NATO. Let me break down what this means. Now you're creating a 24x7 assistant, not a dumb chatboard. A fully intelligent agent that understand context and has a natural conversation. Your customers get instant smart response anytime, anywhere. First, users simply message your WhatsApp business number. Everyone already has WhatsApp. They don't need to download any apps or learn new interfaces. They just message you like a texting a friend. That's the power. You meet customers where they already are. Second is that the AI understand the natural language. When someone asks, "Do you have this in stock or what's your return policy?" The AI comprehends the actual meaning, not just scanning for keywords. It graphs helpful natural responses. And here's the best part. Everything is wired together visually in NAT with zero backend coding. No Python, no JavaScript, no severe management. [snorts] Just drag and drop and connect. If you can understand a flow chat, you can build this. Whether you're a non-technical founder, a content creator, a freelancer, or even an agency, there is no barrier. You'll have a fully developed system running 24/7 on your WhatsApp business number handling real conversations with real customers. Let's build it. Here's exactly how this works. End to end. Understanding this flow is key to everything. Your customer opens WhatsApp and the message is your business number. That message travels through the WhatsApp business cloud API, the official secure pipeline WhatsApp provides for businesses. Then it reaches N8N. N8N is a orchestration. When the message arrives, a web hook, think of it like a doorbell notifies N8. It says new message incoming and N8 springs into action. N then sends that message to your AI model. Whether you choose chat, chippity, claude or even gemini, the AI reads the question, understands the context and generates a intelligent response. This is where the real AI happens. Finally, the response travels back through the WhatsApp API and lands your customer WhatsApp inbox. From their perspective, they just got an instant helpful reply. Here's what's beautiful. It's a simple, clean flow. messages, API, NA10, AI, API, and back to a response. Five steps, each with a clear purpose. And it's also modular. Want to use Claude instead of Gemini? Just swap the node. Want to add data sets, logging, add another node. The system scales. WhatsApp handles billions of messages daily. NA10 powers hundreds and thousands of workflows. Everything here is production ready. This architecture works at any scale. Five messages a day or even 5,000. That's the power of this design. Now let's talk about the key pieces of this. There are four essential pillars. Miss one and nothing works. Get all four right and you have got a powerful system. The first one is WhatsApp business number. This is where customers messages you. It's not a regular phone number. It's a configured through WhatsApp business ecosystem to handle messages at scale 247. This is your front door. Without it, there is no entry for users to send messages. Second is the WhatsApp API and the web hook. When a message arrives at your business number, the API doesn't just store it somewhere. Instead, it sends a web hook notification to the N8 like ringing a doorbell. The moment a customer messages, N8N knows about it and start processing. No manual checks. Everything is instant and automatic. Third is N8N workflow. This is your visual canvas. You can drag and drop. nodes connect them together. One node receives the messages, another processes it, another sends it to the AI, another formats the response. You're not coding, you're stringing. Your workflow is a rule book and your AI agent. Fourth is Gemini model. This is the brain. Google's AI model actually understands language, read the context and also generates intelligent responses, not keyword matching, genuine artificial intelligence. You can customize it with your business knowledge, products, policies and tool. Here's how they connect customer messages with your business number. Messages goes through API via web hook to NAN. N8 routes it to Gemini. Gemini generates response and those responses flows back through N8 back through API to your customer message. Each piece is specialized. Each is essential. Each is battle tested. WhatsApp N8N and Gemini are a productionready platform trusted by millions. Let me show you what this actually does for your business. First is to answer common question instantly. Pricing, delivery times, business hours, refund policies. Customers ask these constantly. Your team answers manually over and over again. Same questions, same answers, wasted time. Your AI agent handles this instantly 24/7. immediate response and no human fatigue. Second is smart FAQ instead of rigid keywords. Regular chatboards are brittle. They only recognize exact phrases for your program. Your agent understand meaning. What's your price? How much does it cost? What's your rate? It recognizes these as the same question phrased differently and responds intelligently. Customers speak naturally. Your agent understands. Third is the short-term memory. The agent remembers conversation, contacts customer, ask about closing costs. Agent responds. Customer follows up with another question. The agent remembers it's still about closing cost and real estate. It answers with full context, not starting from scratch. Conversations feel natural and humanlike. And the fourth one is to extend with CRM, email, Sheets. Beyond conversation, Nen can do more. Automatically log every conversation to Google Sheets. You're building a database. Create CRM records. Never lose a lead. Small email alerts complex questions to the humans. Update spreadsheets. Your sales team knows what customer asked about. And here are some of the real world examples which includes service provider gets a message at 11 p.m. Instant AI response. Customer impressed. In e-commerce business, they track which product gets most questions, improves the production description. and in SAS which handles 80% of the onboarding questions automatically without support team less burdened out all this runs 24 by7 while you sleep while you're in meetings while you're on vacation your agent is working this is leverage before we start let me be clear about what you are getting first is a working AI agent replying to real messages not half finished code a fully deployed active system on your WhatsApp business number real customer messaging real conversation happening. You can screenshot them, show them to clients, investors, and even your team. Real tangible proof it works. Your customers get instant answers. Frequently asked questions are handled automatically. Your team focuses on high value work, not repetitive typing. You scale without hiring. You deploy automation that actually works. Second is you'll understand the full innate end flow, not just clicking buttons. You'll know what's happening at every step. how web hook works, how data flows and how to transform data, how to use conditional logic and how to handle errors. Why? Because in automation something always eventually breaks. When issues happen and they will, you won't be stuck. You'll reason through it. If message comes in here, gets processed here, sent to AI here, response comes back here. So the issue is this note. You can troubleshoot it yourself. When you want to add features, add logging, analytics, CRM integration, payment processing. You'll know how to do it. You're not limited to what we built. You can customize, extend, and scale it yourself. This understanding futureproofs your system. Technology changes, platforms evolve, but understanding the principles and you can adapt. Third is to ready to extend with logging, analytics, CRM integration. See those logs on your laptop. That's inate ecosystem. Hundreds of integrations once your core agent works. Endless possibilities. Add logging. Build complete customer integration history. Add analytics. See which question is asked most. Average response time. Resolution rates. Add CRM integration messages automatically logged to your CRM and payment processing customers complete transaction via WhatsApp. You can add an escalation logic which is complex question route to humans. Add a multi- language support which explain globally. All of these build on the foundation you create. You're not just learning how to build this one. You're learning enan. You're learning automation, learning workflows. These skills transfer to every automation project you do. By the end, you'll have a productionready WhatsApp AI agent, deep and understanding, a foundation you can extend infinitely. skills for any automation project. A system working for you 24 bar 77. This isn't complicated. We break it into a logical steps. We explain not just what to do but why you see progress at each step. This is a hands-on practical real learning. So let's get started with how to automate. So to start your integration with WhatsApp, your business page which is a Facebook page which acts as a meta platform should be connected. So make sure you have to have a business page in Facebook. So here I can show you a fake skincare brand that I have built a Facebook page on. Make sure your online time that is after your creation of the Facebook u account or anything you are at least 1 hour after the creation point else it won't take. So next just type in developers do Facebook and enter into the portal using your Facebook account. So this is the startup here. You can see my apps. Just click on my apps and then here you can see I have already simply care which is added. So how do you add a new app? So just create and enter your details. Suppose the name of your app whatever it is and then give you a Gmail account and click on next. And here it shows you use cases. So either is it you're creating an app for create and manage ads with marketing APIs or is it for create and manage apps add we have access threads to API we can launch a game on Facebook authenticate request data from Facebook login users and also connect with customers through WhatsApp. Now our major goal is to connect with customer through WhatsApp profile. So just click on that click on next. So next here is businesses. So if you don't have simply care it will ask you to create a business portfolio. So as you can see want to connect a new business portfolio just click on that and give your name ID etc. So as you can see portfolio name it's asking. So let's say testing and enter your first name and your last name and create a portfolio. Now again the name you choose for this business manager is not valid. Please choose another. So it's asking us to change the business name. So let's say testing pad with a T capital. So if I create a portfolio, it'll be created because the first letter should be capital. So it'll take some time to create and it'll ask you to verify your portfolio. So then you can verify via your phone, via your email, etc. So after authenticating these two like in the meta page that I have simply care, you guys will have your own products listed over there. So I'm going back because it's needed two factor authentication which I already have simply care as the brand right here to show you people. So after authenticating in the user page itself you'll see simply care with an ID mode in deployment and business simply. Now here for N8N another important platform where we going to automate the systems keep this like this and go back to N8N. So click on N8N.cloud.io etc. And here get started. So here N8 platform provides 14 days free trial and then you have to pay for the version. You can login via your email and then again two factor authentication happens there also. Again after 14 days you might have to pay for this model. So after logging in you'll get a portal like this. I have few works going on. So here we have a overview and my productions etc failure rate my personal where workflows are there admin panel templates insight help settings create workflow etc everything are provided now let's go back you can also chat also so any has inbuilt AI chat system also available so if there is any error or if you want immediate help then and there you can ask Gemini AI which is also integrated in N8 so let's start with this so in my personal account. I'll be creating another folder create workflow and let me name it WhatsApp in 8 in. So the basically we are integrating WhatsApp business with our AI integration to answer the question and later we can train the model as per our business needs. So firstly what are the equipments? Instead of using a manual trigger I'm going to use WhatsApp business cloud first as a trigger and the first one would be on message. So here as you can see we have WhatsApp O account which I'm already added but how you going to add it is click on add credentials and you'll have a client ID and client secret. So go back to your Facebook and you'll see your apps over here. Right? So just click on it, open it. So here you can see uh customize the content with the customers through WhatsApp. So just open that. So if you come down here, you can see API. So start using the API. You just need to click on it and you'll have a API setup configuration. So this is for the API access which we'll be using in the WhatsApp 2. But for the quick setup, go to app setting and basics. Here you'll find the client ID as well as the passcode required. So you'll see an app ID here and app secret here. So just copy the app ID to the client ID and the app secret as well. Just copy and paste it. And when I save it, it should be green. That is credential successfully created. So after this when you go back it's showing couldn't connect because I've already created the same one using o one and here trigger is a message passing through WhatsApp and then you just stop it. Now how will you test if this WhatsApp trigger is working? So if you go back to the Facebook page and go back to where I saw the API configuration setup, you'll see a number test number that is here it is different that is 1 + 1LE 5 1 9 0 1 914. So if you send a message to this when we start the workflow in WhatsApp trigger I should be able to see a green signal in the so just copy this number to your WhatsApp and text it. So let me just show you. So here I have a number. So I'll just execute workflow and I'll be testing texting high. So as you can see this trigger got activated and it is turning into green. So this should be the case with you guys also to know if the trigger is working or not. The next part is we have to attach an AI agent. So just type in AI agent and you'll see an AI agent. Just click on it. And here you can see uh few things which we will say. So for this what is the major concern is how do you connect the trigger? We'll define it below. This trigger will be started only if the message comes. So what is the message that we sent? You can see it in the input box which is hi. So just drag and drop the body part here and it will immediately pick up the results that is the text that I've already shared in WhatsApp here and then this is good to go and here you have to add a API key whether it's chat GPT key which is the Google AI studio or even open router is fine now how do you get the key first of all let me add the Gemini AI Gemini API keys are free you can use it whereas for open router or chat GBT or open AI platforms API key you have to pay. So for open AI you just need to add open router chat model here and again create an account give the API key and save. So since here I'm not using um open AI router or um open router chat I'm going to use Gemini and connect it. So again I already have an account but I'll show you how to create an account here. It's similar for open router as well. So create new credentials if you don't find any account and the host will be already there. API key. So for API key you should visit Google AI studio which just type in Google AI studio and the first link and just sign up for it and you'll get a interface like this. Now you can see an option of get API key. Open that click on it and either you can create a new API key. Give it a name and create key. Your key will be created. Here I've already created so I'll just copy the API key. Come back to N8 and paste it and just save it. So here you can see credential successfully created inside your personal space and it's a green one. So that is done. Now you can just select whichever account is comfortable for you and select the model which you want to work with and you're good to go. Now close this. So after AI agent having in connection with the chatboard the next one is we have to reply back with a message right. So for that we're going to use WhatsApp business cloud but here we are going to use send message as an option because we want to reply it back. So here it is asking another account. So here I'll show you how to create a new credential. So there are two kinds of WhatsApp credentials required. First one is to get the uh ID, client ID and the passcode. The second type here is access tokens. Now how do you get the access tokens? So go back to the meta app where you have uh API set up as an option in the same page and here you have a test number and a recipient numbers. So just add your recipient numbers and generate the access token. So it'll again ask your Facebook details to login, authenticate, process, etc. So I'll just say reconnect and you'll have a access token. So copy this and paste it in the NA10. Now what about business account ID? Again the business account ID will be here. WhatsApp business account ID. So copy this. Go back to the NA10 business addit and save it. So this will be your credential. As you can see connection tested successfully. That means your connection is tested successfully. Now how do you add credential? How do we know what message to reply what? That's the thing here. So I'll go with the normal WhatsApp account to which I already have. This is the message. Obviously we need a reply for the message itself and send operation is selected. So senders recipient numbers type in the recipient numbers including the country code and message will be text type and text body. So let's just add the recipient's phone number which and then execute this step. It's a text message and we'll figure out the body part and then let's execute the workflow. It's waiting for a response. So just go and type in hi. So let's just integrate AI agent as well. So here as you can see we'll just execute the previous node. And here you can see body is there. So just drag and drop it here. So your reply will be high. So if that is fixed. Yeah, I think it is fixed and close. So again we have green here, green here and the problem is here. So let's integrate this. Open the options here. As you can see uh we have an output. So what should be the text body trigger? Drag here and drop the output. So this will be our reply and you can cross it. We have test number, recipient numbers. Drag and drop the reply which you want to give back. This is the output from the AI agent. Close it. Now again if I ask any question in WhatsApp tester number send a message and test it out. It should give me the answer. So let me just show you. So previously high send when triggering the output. So let's let me click on execute workflow [snorts] and I will also type in so execute workflow and I'll be typing in hey AI agent is responding and there you go I got a reply saying hey there how can you help you today? That's our first reply. Now again I have to have other questions. So let's see ask let's try and ask the EI to tell us a joke. So let's say can you tell me a joke? So it is again triggered AI agent reads and sends us the response. So it has given us a joke and how you can integrate this part into your business is training the AI model. Now let's train the AI model. For that whatever Google Gemini chat model you're using just give all the information of your business data etc. and it'll be absolutely answering all the business related questions using natural language process. All right, so now we have gone from zero to a fully functional WhatsApp AI boat. Here are your final tips before you even ship this. Number one is to keep your WhatsApp token secure. Store it in an environment variable. Never hardcode it. If it ever leaks, regenerate it immediately from meta. Tip number two is to monitor your workflow executions. Nitn logs everything. Watch for errors in your early runs and iterate on a prompt until Gemini nails your use cases. And tip number three is to test with real message first. Don't go live to production users until you have sent at least 20 text messages and verified every response. Tip number four is to scale thoughtfully. Gemini's free tire has rate limits. If you hit them, upgrade the plan and adjust your workflow to batch messages if needed. Now, what are the following step? Take this workflow, customize the prompts for your business, integrate it with your CRM or Google Sheets, and start automating customer conversations. The possibilities are endless. If you build something cool, let us know in the comment section below. Thank you for watching. And if these are the kind of videos you would like to watch, make sure to like and share this video and subscribe to Simply Learn for more such content. Now go automate. See you in the next one.