WEBVTT 00:00.160 --> 00:01.200 Welcome back to the course. 00:01.240 --> 00:03.720 In this tutorial I'm going to show you something really cool. 00:05.000 --> 00:10.640 This new N810 MCP will create for you complex workflows in just a few minutes. 00:11.600 --> 00:16.440 You just describe what you want in plain English and it builds everything automatically. 00:17.280 --> 00:21.360 We are talking about workflows that would normally take hours to build. 00:21.520 --> 00:24.640 Now you can have them done in minutes using a simple prompt. 00:24.840 --> 00:29.840 N810 is the best platform for building agents and automations without coding. 00:30.120 --> 00:30.960 That's for sure. 00:31.200 --> 00:37.200 You can connect almost any service or app you can think of, and create sophisticated automations that 00:37.200 --> 00:39.720 would normally require a team of developers. 00:40.200 --> 00:42.360 But the problem is there is a learning curve. 00:42.400 --> 00:47.280 There are hundreds of different nodes and configurations to learn, and even though the drag and drop 00:47.280 --> 00:52.840 interface is nice, it can take hours to build complex workflows, so it's really time consuming. 00:53.400 --> 00:55.320 But this MCP changes everything. 00:55.440 --> 01:00.810 You can just describe what you want to an AI agent, and it will build the entire workflow for you. 01:01.090 --> 01:05.250 Not just generate some broken code, but create working workflows. 01:05.650 --> 01:09.210 You just set up your credentials and can start using them right away. 01:09.930 --> 01:15.730 Now in this tutorial, I'm going to show you exactly how you can set it up to build powerful workflows 01:15.730 --> 01:17.330 in minutes instead of hours. 01:17.690 --> 01:23.690 Now, you might be wondering, how is this different from just asking ChatGPT or cloth to build a workflow? 01:24.170 --> 01:29.210 Well, when you try that approach, you usually get something that looks decent but doesn't actually 01:29.210 --> 01:29.450 work. 01:29.450 --> 01:35.930 When you import it into an A-10, the nodes don't connect properly, the parameters are wrong, and 01:35.930 --> 01:40.490 you end up spending more time fixing it than if you build it yourself. 01:41.050 --> 01:43.970 That's because AI models are basically guessing. 01:44.210 --> 01:47.490 They don't really understand how anything works under the hood. 01:47.810 --> 01:51.250 But this nmcp is built completely differently. 01:51.610 --> 01:59.270 This thing has direct access to the actual documentation, not some outdated information or descriptions, 01:59.270 --> 02:06.350 but the real technical specs that the developers use it understand about 90% of the official documentation 02:06.550 --> 02:09.270 and knows exactly how each node should work. 02:09.710 --> 02:15.710 It has dedicated tools that pull up the exact documentation it needs before building anything, so it 02:15.710 --> 02:18.190 works with real knowledge instead of guessing. 02:18.630 --> 02:20.550 And here is what makes it so powerful. 02:20.750 --> 02:22.510 It follows a three step process. 02:22.950 --> 02:24.990 Let me break down how this actually works. 02:25.270 --> 02:28.190 So the MCP has three main sets of tools. 02:28.350 --> 02:29.750 First you have core tools. 02:30.270 --> 02:31.750 So documentation tools. 02:32.230 --> 02:37.750 These search through all the available nodes, pull up the right technical specs and gather everything 02:37.750 --> 02:39.430 needed for your specific workflow. 02:40.510 --> 02:46.110 So it looks up the right documentation, figures out which nodes you need, and understands how they 02:46.110 --> 02:46.950 should connect. 02:48.110 --> 02:50.230 Then you have the advanced tools. 02:50.630 --> 02:56.270 So the workflow creation tools, these take all that information and actually build the workflow. 02:56.270 --> 02:57.440 Well, so be it. 02:57.480 --> 03:01.280 The JSON structure with proper connections and configurations. 03:01.280 --> 03:07.240 So it builds the actual workflow and it knows the correct parameters, the right connections, everything. 03:08.480 --> 03:10.920 Finally, there are management tools. 03:10.920 --> 03:12.120 So deployment tools. 03:12.480 --> 03:18.720 These push the completed workflow directly into your account where you can start using it immediately. 03:18.920 --> 03:22.760 So it deploys the workflow directly to your workspace. 03:23.680 --> 03:31.080 Since this is built as an MCP, it's compatible with both cloud and cursor, so you can choose the AI 03:31.120 --> 03:35.200 platform that works best for you or the one you are already using. 03:36.240 --> 03:38.360 Now here is what makes this really consistent. 03:38.560 --> 03:41.720 The MCP follows a proper sequence when it works. 03:41.960 --> 03:45.640 So it won't try to build the workflow before it has all the documentation. 03:46.080 --> 03:49.880 It won't deploy something before it validates everything first. 03:50.120 --> 03:54.240 And you can make this even better by setting up some basic configurations. 03:54.480 --> 03:55.420 So in cloud cloth. 03:55.620 --> 03:59.060 You can add this to your project settings in cursor. 03:59.100 --> 04:01.100 You put them in your configuration file. 04:01.620 --> 04:03.020 Either way works perfectly. 04:03.460 --> 04:09.900 This structured approach prevents the random results you often get from AI tools when they don't have 04:09.940 --> 04:10.940 clear direction. 04:11.580 --> 04:17.140 With this setup, you get complete workflows every time instead of broken or unfinished outputs. 04:17.660 --> 04:21.260 So this is why I think MPs are going to change everything. 04:21.580 --> 04:24.580 We are not just talking about AI that generates code. 04:24.780 --> 04:31.340 We are talking about AI that actually understands how platforms like Nathan works and builds workflows 04:31.340 --> 04:32.620 that function properly. 04:32.940 --> 04:37.740 So the AI tools that actually understand the platforms they are working with. 04:37.900 --> 04:43.140 When you use N810, you are working with this visual builder where you drag and drop different nodes 04:43.140 --> 04:44.620 to create your automations. 04:50.980 --> 04:52.980 Think of it like building with Lego blocks. 04:53.260 --> 04:58.030 Each piece has a specific purpose and you snap them together to create something functional. 04:58.630 --> 05:00.470 So these blocks are called nodes. 05:00.750 --> 05:07.950 Each node represents a specific task or function in your workflow, maybe sending an email, processing 05:07.950 --> 05:10.070 data, or updating your CRM. 05:10.790 --> 05:12.670 But here is what's happening under the hood. 05:12.870 --> 05:17.110 So every workflow you create is actually stored as a JSON file. 05:17.270 --> 05:23.270 So this file contains all the details which nodes you are using, how they are connected, what parameters 05:23.270 --> 05:24.710 are set, everything. 05:25.230 --> 05:32.110 And the cool thing is if you have the JSON file ready, you can just import it directly into an A-10. 05:34.230 --> 05:38.510 And your entire workflow appears instantly in the Visual builder. 05:39.390 --> 05:45.270 Now, ChatGPT or other Llms don't have the specific knowledge needed to build something that actually 05:45.270 --> 05:46.310 works in an A-10. 05:47.150 --> 05:53.520 But this MCP is different because it's trained on the documentation and has access to multiple tools 05:53.800 --> 05:55.840 and can work like an agent. 05:56.480 --> 06:00.640 If you are using Cloud desktop app, you can actually watch this happen. 06:01.000 --> 06:08.120 So the JSON gets built right there in your chart using the artifact feature, and you can see each part 06:08.160 --> 06:08.960 being added. 06:09.400 --> 06:12.840 So it's pretty cool to watch how your workflow comes together like that. 06:13.040 --> 06:18.120 Even if you are using something else, the MCP does all this work for you behind the scenes. 06:18.560 --> 06:24.680 And once it knows what it needs, it puts together your workflow, makes sure everything looks good, 06:25.120 --> 06:29.760 and sends it straight to an account where you can start using it. 06:30.680 --> 06:35.720 Now let me show you how to actually set this up, because it's way simpler than you might think. 06:36.680 --> 06:43.560 When you check that GitHub repo, you can see that we have a few different options. 06:44.160 --> 06:46.880 The main prerequisite is to install NodeJS. 06:47.720 --> 06:50.880 First you need to have Node.js installed on your computer. 06:51.220 --> 06:57.820 If you don't have it already, just go to Nodejs.org or just simply access it through this repo. 07:00.460 --> 07:02.340 And download the latest version. 07:03.020 --> 07:03.860 It's really easy. 07:03.860 --> 07:07.900 Just download it, install it and you are ready for the next step. 07:08.380 --> 07:11.780 Next you will need Cloud Desktop app installed on your computer. 07:12.100 --> 07:16.940 You can download it from Cloud's website and obviously you will need an account to log in. 07:17.660 --> 07:23.980 This is important because the browser version of cloud doesn't support Mctp servers, so you need the 07:23.980 --> 07:24.900 desktop version. 07:25.980 --> 07:29.460 Now open Cloud Desktop app and go to settings. 07:31.820 --> 07:32.860 Click on developer. 07:34.860 --> 07:38.340 Then at the bottom left corner you will see Edit Config, so click on it. 07:41.300 --> 07:44.020 This will open up a configuration file on your computer. 07:45.700 --> 07:46.660 So let's open it. 07:50.510 --> 07:56.590 Now, if you are configuring MCP servers using Cloud Desktop app for you, it will be empty. 07:58.590 --> 08:01.470 We need to paste a few things into this configuration file. 08:03.070 --> 08:05.510 So let's copy it from GitHub step by step. 08:05.990 --> 08:13.430 By the way, thank you to the author of this repo to Gronkowski for creating this for us. 08:14.550 --> 08:18.510 When you scroll down you will see the instruction how to set it up. 08:20.070 --> 08:26.710 The easiest method is using Npx and we've got basic configuration and full configuration. 08:26.830 --> 08:28.230 I recommend using this one. 08:29.070 --> 08:33.710 So Cloud Desktop app will be able to import the workflow into your instance. 08:34.990 --> 08:39.350 So let's copy it and let's go back to our configuration file. 08:47.310 --> 08:48.510 And now let's paste it here. 08:50.000 --> 08:54.200 Now we need to replace N810 api url and N810 API key. 08:55.080 --> 09:02.120 Now, in order to find your N810 API URL, you need to login into your account. 09:02.840 --> 09:06.200 You will see your URL right there in the address bar. 09:06.640 --> 09:10.200 Just copy everything up to the cloud part. 09:10.240 --> 09:11.920 That's your API URL. 09:12.760 --> 09:17.920 Make sure to include https and paste it into your configuration file. 09:18.080 --> 09:19.560 So just replace this URL. 09:24.600 --> 09:25.040 Great. 09:25.840 --> 09:28.480 Now we need to grab your N810 API credentials. 09:29.000 --> 09:30.680 This is super straightforward. 09:31.240 --> 09:33.600 So when you are login into your account. 09:34.160 --> 09:38.440 Now to get your API key click on your profile icon in the bottom left. 09:39.680 --> 09:40.680 Go to settings. 09:44.040 --> 09:46.200 And you should see N810 API section. 09:47.220 --> 09:48.060 Click on that. 09:49.380 --> 09:51.060 And here you can create your API key. 09:53.700 --> 09:54.460 Give it a name. 09:59.740 --> 10:00.380 It's safe. 10:02.660 --> 10:03.500 Copy it. 10:04.540 --> 10:08.620 Go back to configuration file and replace this API key. 10:10.380 --> 10:12.300 So paste your API key right here. 10:14.460 --> 10:15.140 Like that. 10:15.740 --> 10:17.780 Now we need to save the configuration file. 10:21.700 --> 10:23.780 And completely restart cloud desktop app. 10:24.180 --> 10:25.340 Close it completely. 10:28.820 --> 10:31.580 This is important because just refreshing won't work. 10:31.620 --> 10:37.900 You need to fully restart the application if you prefer using cursor instead of cloud desktop app. 10:38.100 --> 10:41.900 The setup is almost identical, so open cursor. 10:42.380 --> 10:43.700 Go to cursor settings. 10:46.590 --> 10:47.230 Settings. 10:49.750 --> 10:54.590 Click on Tools and Integrations and create new web server. 10:57.910 --> 11:01.310 And paste the exact same configuration we just used for cloth. 11:01.710 --> 11:02.230 And that's it. 11:02.270 --> 11:06.470 Cursor will now have the same n810 MCP capabilities. 11:07.470 --> 11:10.550 Now here is a pro tip that will make everything work much better. 11:10.670 --> 11:18.150 So instead of just using cloth desktop app normally creating a project, you can name it whatever you 11:18.150 --> 11:18.550 want. 11:18.950 --> 11:21.030 The important part is adding project knowledge. 11:21.270 --> 11:27.870 So there is specific instructions you can add that help cloth understand how to use the MCP tools properly. 11:28.030 --> 11:31.430 So you can find these instructions in the GitHub documentation. 11:31.950 --> 11:37.310 Basically, they tell cloth to follow a specific workflow when building N810 automations. 11:38.390 --> 11:40.630 Just click on Set Project instructions. 11:41.750 --> 11:43.470 Go to the GitHub repo. 11:45.010 --> 11:45.730 Scroll down. 11:49.010 --> 11:50.890 And you will see cloud project setup. 11:52.890 --> 11:54.130 Just copy the prompt. 11:57.210 --> 11:58.730 And paste it as a knowledge base. 12:01.250 --> 12:03.130 To setup the project instructions. 12:03.770 --> 12:04.930 Save instructions. 12:07.330 --> 12:07.890 Perfect. 12:09.090 --> 12:14.690 So now once your project is set up, you should be able to see the MCP tools available. 12:15.010 --> 12:17.450 So in the project look for the tools section. 12:20.090 --> 12:24.890 And you should see an MCP with multiple tools available. 12:25.410 --> 12:28.010 There are 39 tools at the moment. 12:32.810 --> 12:35.130 If you see that everything is working correctly. 12:35.490 --> 12:39.210 If for some reason you don't see the tools, you might need to troubleshoot. 12:39.210 --> 12:46.340 So make sure NodeJS is properly installed and double check your configuration file or restart the cloud 12:46.380 --> 12:47.100 desktop app. 12:47.460 --> 12:51.060 A few things to keep in mind when you're using this setup. 12:51.300 --> 12:57.900 First, make sure you are always working within your project when you want to build workflows, so the 12:57.900 --> 13:01.740 project knowledge helps cloud understand the proper workflow. 13:02.460 --> 13:07.660 Second, I recommend using Cloud Opus for the best results. 13:07.980 --> 13:12.060 Is the best model and handles complex workflow building much better. 13:13.100 --> 13:20.420 And finally, the MCP has about 39 different tools and it will often use several of them for each request. 13:20.420 --> 13:24.620 So cloud will ask you to authorize these tools for the first time. 13:25.140 --> 13:31.100 I recommend clicking allow always so it will be able to create workflows for you much faster. 13:31.500 --> 13:36.460 Once everything is set up, you can start building workflows just by describing what you want. 13:36.740 --> 13:42.350 The MCP will automatically search for the right nodes, configure them properly, Validate everything 13:42.590 --> 13:45.430 and deploy the finished workflow to your account. 13:46.590 --> 13:53.310 Now, if you are using an internal cloud or self-hosted, then on this setup works perfectly fine. 13:53.590 --> 13:59.270 But if you are self-hosting and you turn on your own machine, there are a few things to keep in mind. 13:59.990 --> 14:04.590 If you want to run locally on your machine, you have a couple of options. 14:04.590 --> 14:07.270 You don't actually need Docker to run anything locally. 14:07.710 --> 14:10.510 You can simply use the npx command to start it. 14:11.230 --> 14:16.390 Just run Npx N810 in your terminal and it will start an internet localhost. 14:16.790 --> 14:23.790 However, Docker is generally the most stable option for local hosting since it provides better isolation 14:23.790 --> 14:24.990 and consistency. 14:25.590 --> 14:29.510 But for getting started quickly, the Npx method works perfectly fine. 14:29.870 --> 14:36.190 Now, the localhost might not work as a URL with cloud, desktop app, or cursor. 14:36.710 --> 14:42.570 These are applications that sometimes can access localhost the same way you can. 14:43.130 --> 14:44.290 The solution is simple. 14:44.290 --> 14:53.090 Instead of using localhost as a URL in the configuration file, use your local network IP address so 14:53.090 --> 15:00.050 you can find this by running command this command on Mac or this one on windows. 15:00.730 --> 15:07.810 Then use that IP in the configuration file like this example instead of localhost. 15:08.210 --> 15:13.890 The API key setup is the same regardless of whether you are using cloud or self-hosted N810. 15:14.730 --> 15:19.050 Just go to settings, look for the API section and create a new key. 15:19.810 --> 15:23.090 Then copy that key and paste it into your configuration file. 15:24.050 --> 15:24.810 And that's it! 15:24.810 --> 15:32.210 This is how you can use this amazing n810 MCP in apps like cloud, Desktop app or cursor to build powerful 15:32.250 --> 15:35.850 n810 workflows that actually work and save a lot of time. 15:36.450 --> 15:37.770 I hope this was helpful. 15:37.770 --> 15:39.730 Thanks for watching and see you next time.