How to Connect DeepSeek to Make.com + 3 Real-World Automations

First, I will show you how you can connect DeepSeek to Make.com and then we’ll walk to three real world examples that you can start using AI in your scenarios right away. So the first thing we need to do is we need to create an account at DeepSeek. So go to DeepSeek.com and then here at the top right, we want to go to their API platform. API stands for Application Programming Interface. That’s like a very techy name. It’s basically just a way for Make and DeepSeek to talk to each other. So we go to the API platform and then over here we go to sign up. Now then enter your email address, enter a password, and then request a code to be sent. So then here it says verification code was sent. Go over to your inbox, get the verification code. So in this case, that is this over here. Then I go here and then I click sign up. And then the first thing you will see is your dashboard with your usage and your balance, which of course says zero right now. So the first thing we need to do is we need to add a few bucks to our account so that we can actually use it. My suggestion would be to add five bucks. That’s already a lot already. That’s like thousands upon thousands of completions. It’s very affordable. So select five bucks, choose your payment method, go to the next steps, and then afterwards you’ll see that you have five bucks in your account. Now then the first thing that we need to do is we need to create an IPI key for Make.com. So it’s basically like a little password that we can use to connect to our account. So we click that over here and then I like to name this Make.com and then create API key. Now then we need to copy this over here. And then what’s very important is that you do not lose it because you’ll never see this again. So if you want to reuse it maybe for something, then copy that. But what we basically do is we copy it, we click here done. Then we go over to make and then what you want to look for is the deep seek module. Now then look here for deep seek AI verified. So, this is like the official DeepSeek module. And then over here, we are going to use create a chat completion. Now, then it will ask us to create a connection. So, I just call this one DeepSeek. And then we paste that API key from earlier in here. So, I click save. Then it is going to connect. It will take a second. And then you will get all the options for the module over here. Now, then just a very quick explanation. So we have user and system. So system is basically to tell it, okay, what is your purpose? And user is like adding more details to whatever you’re trying to do. So here, system. For example, we say you are a helpful assistant. And then name you can leave empty. And then we add another item. And then we say here, user. And you just say, hi there, for example, right? Now then down here, we select the model. so we have here deep seek chat and we have deep seek reasoner and chat is very handy for like quick tasks like you want to like summarize something you want to generate some text you want to create content something like that and reasoner is more handy if you want to analyze something I’ll also show you an example of that in a bit so right now I’m just going to set here this to chat and then we have here the maximum amount of tokens it’ll use I’ll just leave this the default but if you want you can reduce this and then over here we have the temperature this is quite important it’s a basically tell it how creative it’s going to be so the lower the number the less creative it is and the more to stick to the rules that you give it so usually I like to set this to 0.2 so this will make it a tiny little bit creative but it’s not going like a rogue and like it’s going to like ignore our rules. Now then we have a response format you can choose between text and a JSON object. JSON is basically a text-based data structure, I’ll show you another example of that in a bit but now I’m gonna leave this to text. I click here save, click save again and then now I’m just gonna click here win once just to test see if it’s working. And then over here we have the result hey hello how can I help you today? So it’s like a very simple result right we did not tell it anything complex to do to create or to write so it gives us a very simple response but this is how you connect deep seek to Make.com now then now I want to show you three really cool real-world examples now the first one is to create a Facebook post whenever a new video is published on my youtube channel you can also use it for something else like an article on the website or even another website but this is how I use it So over here I have the youtube.com feed for my YouTube channel. So then whenever a new video gets posted this one starts once a day at 9 o’clock and it’s going to check is there something new and then it’s going into the deep seek module and then over here you see again that we have a system role and we have a user role. So the system over here says you are a professional copywriter for social media and you write one short Facebook post based on a new video that I will provide the details of. The purpose of the post is to share and announce the videos and get people to watch it and then I give it a bunch of important rules. Now then below here under user I have new video details, title and then the title from the RSS module and also the description from the RSS module. Then I used the DeepSeek chat model, I did not limit the tokens and I set the temperature again to 0.2 so that it gets a little bit creative but I do not want to get it too creative because otherwise it’s just going to ignore these rules over here and then what I did is I get a result from this also demonstrated in a bit and then we get a Facebook post and then I’ll just share the result as a post and then also a link to that new video so let me demonstrate click here once so then you see over here we get here one of my tutorials on YouTube how to extract an email address from a text in Make.com. Then over here, you see that it generated that Facebook post. So here, need to pull an email address from a messy block of text. I just made a quick video showing you how to usually do that with Make.com. Now, and then I posted this to Facebook. And then the second example is this scenario over here that helps me to organize my inbox in Gmail. So over here, this module is going to watch for new emails. and then with deep seek I’m going to analyze that email. So over here you see I have a system-based role that says here you are an assistant that organizes my emails analyze the provided email and return only the name of the most relevant label from this list and then I give it a list of labels that I have in Gmail you can also see those over here and then down here user I set here I give some details basically of the email. So I give here who the email was sent to, the subject line, and then also the text content of that email. And then here the model. So instead of the deep seek chat, I use deep seek reasoner because I’m analyzing something. So it’s good at doing that. So that’s why I use that one over there. And then again, I left the temperature to 0.2. Now then I say here, okay, return in text. and then we go into a router and then you see over here that I compare the results if it’s equal to the name of that label and then if that is so then I add that label with this module over here so you see here then it adds purchases so let’s test it over here click Save run once so then it’s gonna look for new emails and then you see that we have a new email then the email will be sent to deep seek for analyzing and then after that is done it will tell which label to apply to that. So then you see over here that we got that result subscriptions. So whatever email we got it said it is a subscription so it went over here the result is equal to subscriptions so we added the subscriptions label. Then the third example is processing a context form submission. That’s always cost quite a lot of time and I was thinking hey how can we make this like a little bit faster and now I use DeepSeek to summarize basically the email and then use this as a subject line so that I immediately know hey this is what that email is about and I also use it to personalize one little sentence that I send back to someone, not super necessary but I thought it’s fun. So over here we get to the contact form data, I will demonstrate that in a bit and then we go to the deep seek module and over here we say hey process this form submission and return only invalid decent object with the following structure. So we want to get a subject and we want to get a summary so here I said okay I want to have a short title maximum 40 characters summarizing message and then a summary and basically what I wanted is I wanted to have something that fits naturally into the sentence I just wanted to let you know that I got your email about dot dot dot and then I’ll use that to personalize the email now then down here under user I said okay here this is the content of the form submission then the model we use deep seek chat tokens I just left it to that and then the temperature is that again to 0.2 so that it sticks to our rules over here and then under response format instead of text and now set json object and this is quite handy to notice that’s why I wanted to include this example is that you do not always want to have just one result with the text sometimes you maybe want to get a bunch of stuff back that you would then want to use later on in your scenario so a json object so json is basically just like a text-based data structure it’s just like a way for programs to understand a format of data and make really works well with this I also have a guide about json I’ll put it in the description if we want to check that one out but so over here we have a decent object and then to make sure that make understands the json object we parse that json so we put our result in here and then we have something that we can then use inside our emails as a subject line over here and also here in my sentence I just wanted to let you know that I got your email about and then over here we have the summary. So let me demonstrate this. So click here okay then I will submit the form on my other screen and then over here you see okay we got the form submission so here first name Emma, email address Emmah Doe, question I’m working on a project to teach people how to bake bread could you help me to automate the student onboarding process. Just an example of something they could get. And then over here, you see now that we’ve got this as a result back. Subject, bread baking project help. Summary, automating the student onboarding process for your bread baking project. So then we have this JSON structure over here. And then now we set here parse to JSON. So now you see here that we have these variables that we can then use later on in our scenario again, like we did over here. So here we just clip here and then you can just like map here these variables, subject and summary and then now you see that it sent the email first to me about the submission that’s like hey we have a new submission and then also to that person to let them know like hey I got your email and that is this email over here so you see here bread baking project help so that is the subject line it generated hey max and then we have the personalized sentence I just wanted to let you know that I got your email about automating the student onboarding process for your bread baking project and it will get back to you within one or two days. I thought this is super fun to include. This is one of the things you can now do with automation, but if you want to make your life even a little bit easier. So now I showed you how to include the DeepSeek module into your Make.com scenario, then I also want to invite you to the world of OpenRouter. It’s a super cool tool to which you can basically connect to hundreds of large language models like DeepSeek. So this module over here, and what is really cool is you can select any kind of module from any kind of company without having to create an account at all of these companies over here. So if you want to learn more about how to use that, you can, for example, even use DeepSeek and then also a free model of them. So it is already very affordable, but then it’s even cheaper. So if you want to learn more about that, then watch this tutorial over here.