WEBVTT 00:00.200 --> 00:07.920 Welcome in this episode, before we dive into building some amazing automations and agents with N810, 00:08.320 --> 00:10.840 let's start by clarifying a few basics. 00:11.440 --> 00:14.480 So what exactly do we mean by automations? 00:14.520 --> 00:16.800 A Automations and agents. 00:17.280 --> 00:19.760 There is plenty of confusion around these terms. 00:20.080 --> 00:25.720 So today, instead of just giving you the definitions, I will show you the differences using clear 00:26.040 --> 00:28.360 practical examples directly in N810. 00:28.920 --> 00:34.960 By the end of this episode, you will have a crystal clear understanding of each concept, and you will 00:34.960 --> 00:39.680 also get a sneak peek into the workflows we will be building together. 00:40.120 --> 00:42.080 So let's jump right in. 00:45.040 --> 00:46.600 Let's start with automation. 00:46.920 --> 00:50.120 So if we ask ChatGPT, what is automation? 00:50.120 --> 00:57.800 In simple terms, it says that automation means setting things up to happen automatically without needing 00:57.800 --> 00:59.560 a human to do them each time. 01:00.240 --> 01:03.450 So for example, when someone fills out a form online. 01:03.930 --> 01:10.170 The system automatically sends the, uh, the welcome email without you having to do it manually. 01:11.450 --> 01:16.650 Automation means setting up tasks to run automatically without human intervention. 01:17.290 --> 01:24.130 So in other words, you let technology handle repetitive tasks for you to help you understand how a 01:24.130 --> 01:25.690 standard automation works. 01:25.930 --> 01:28.650 I have prepared a simple automation in n810. 01:29.490 --> 01:36.650 So a simple example could be collecting contact information from visitors to your website and automatically 01:36.690 --> 01:38.650 storing their details in a spreadsheet. 01:39.210 --> 01:43.610 Basically, every automation always includes two basic parts. 01:44.170 --> 01:50.170 So the first one is trigger, which means something happens to start the automation. 01:50.650 --> 01:53.010 And in that case we have a form submission. 01:53.570 --> 01:56.570 So we can embed this form on our website. 01:58.210 --> 02:04.870 So when the visitor fills out the form and provides all the details, our automation will be triggered 02:05.510 --> 02:09.590 and all the data will be passed to a second node. 02:09.870 --> 02:14.430 So to the action node and get stored in Google Sheets. 02:15.110 --> 02:18.110 I have created a Google sheet to store our data. 02:19.110 --> 02:21.830 We have two columns a name and email. 02:23.470 --> 02:25.310 So now let's see it in action. 02:26.110 --> 02:28.190 I'm going to test our workflow. 02:28.190 --> 02:29.390 So our automation. 02:31.870 --> 02:33.910 And the form submission pops up. 02:34.750 --> 02:37.550 So let's enter my name. 02:40.310 --> 02:41.470 And my email address. 02:47.430 --> 02:48.230 Hit submit. 02:49.190 --> 02:51.270 And now our automation has been triggered. 02:51.950 --> 02:56.110 And these details have been passed to our action node. 02:56.110 --> 02:57.390 So to Google Sheets. 02:57.750 --> 02:59.710 Now let me check our Google Sheet. 03:00.550 --> 03:02.190 As you can see it has been updated. 03:05.830 --> 03:12.200 In a real world scenario, you would place a form on your website to capture your visitors details. 03:13.320 --> 03:21.920 And when this workflow is active, every time when a new visitor fills out the form, this workflow 03:22.560 --> 03:28.760 will be triggered and all the details will be passed to your Google Sheet automatically, so you can 03:28.760 --> 03:30.560 reach out to your prospects later. 03:32.480 --> 03:38.960 Now let's discuss AI automation and illustrate this with a practical example. 03:39.400 --> 03:43.280 By the way, we create this automation from scratch in a future lesson. 03:43.880 --> 03:49.840 But for now, imagine you are running a business and your inbox is constantly flooded with hundreds 03:49.840 --> 03:58.040 of customer emails every day, like questions about payments, requests for services, consultations, 03:58.040 --> 03:59.320 or general inquiries. 03:59.920 --> 04:04.800 And manually sorting through each message is tedious and time consuming. 04:05.760 --> 04:17.770 That's why we can Use large language models inside our automations, and I could handle this automatically. 04:19.650 --> 04:21.650 And here is exactly how it works. 04:22.370 --> 04:23.530 So you have trigger. 04:24.690 --> 04:32.330 You receive a new email captured by the Gmail trigger node and the large language model of your choice. 04:32.610 --> 04:40.010 In that case, we use GPT four mini reads and understands the email content. 04:40.930 --> 04:50.290 It then automatically categorizes the email into a few groups, such as service requests, consultation 04:50.290 --> 04:53.130 requests, payments, and others. 04:53.810 --> 04:59.970 And based on this classification, the automation then routes each email to the correct department or 05:00.010 --> 05:02.730 sends a suitable automatic replay. 05:04.650 --> 05:12.830 For example, a customer sends an email asking how can I pay my latest invoice, and the AI instantly 05:12.830 --> 05:15.470 recognises this as a payment question. 05:17.710 --> 05:26.150 And it sends a helpful reply automatically, so it saves you a lot of time and ensures prompt responses. 05:27.430 --> 05:35.190 N810 is amazing because it lets you effortlessly integrate powerful large language models like GPT for 05:35.190 --> 05:43.470 all essentially ChatGPT, as well as other popular models like Deep Sea, grok cloud, etc.. 05:45.190 --> 05:51.950 And don't worry, we'll go through large language models and how to integrate them in the future lessons. 05:53.830 --> 06:01.790 In this example, images are categorized and routed into relevant inboxes, so I have created corresponding 06:01.790 --> 06:04.470 labels in my Gmail inbox. 06:05.510 --> 06:13.160 And now every time I get a new email, this AI automation will automatically sort it and move it into 06:13.160 --> 06:14.880 the inbox with the correct label. 06:17.000 --> 06:18.520 Now what is an agent? 06:19.200 --> 06:27.000 An AI agent is software that actively makes decisions, performs tasks independently, and interacts 06:27.000 --> 06:29.000 with users or other software. 06:29.520 --> 06:36.400 So you can think of it as your personal digital assistant, but capable of understanding your requests, 06:36.440 --> 06:40.800 performing actions, and even remembering previous conversations. 06:41.440 --> 06:43.320 So let's take a look in this example. 06:45.080 --> 06:50.560 So as a trigger, a user sends a chat message through open chat. 06:50.960 --> 06:55.040 So for example you ask what's the latest news on open AI. 06:56.160 --> 07:04.800 And the agent is always powered by an LLM like ChatGPT for all and instantly understands your query. 07:06.680 --> 07:12.360 Just like with AI automations in AI agents, we also use large language models. 07:12.880 --> 07:16.940 So think of these this LMS as the brain of our AI agent. 07:17.380 --> 07:23.540 They are responsible for understanding your requests, making smart decisions, and using the right 07:23.580 --> 07:27.940 tools to carry out tasks independently. 07:28.700 --> 07:37.700 The agent then uses built in tools, so in that case, we connected SAP API to search the internet for 07:37.900 --> 07:41.180 real time information and memory. 07:42.060 --> 07:46.500 So it's window buffer memory node to keep track of previous conversations. 07:51.420 --> 07:57.140 Alright guys, so let's test our agent to see how it works in action. 07:58.300 --> 08:07.060 And because it uses SAP API to search the internet for real time information, we can ask it, for example, 08:07.100 --> 08:12.860 to retrieve the most recent news about a agent. 08:13.660 --> 08:18.150 And we can test this agent by clicking on open chat button. 08:20.750 --> 08:23.630 And here you can interact with our agent. 08:25.390 --> 08:26.950 So let's ask it for. 08:31.150 --> 08:31.910 The newest. 08:35.710 --> 08:36.670 Articles. 08:39.710 --> 08:41.270 About my agent. 08:49.950 --> 08:56.830 And now as you can see, it's looking for the newest articles about my agent. 08:58.150 --> 09:00.870 And it should get a response in a sec. 09:06.390 --> 09:08.830 Successfully retrieved some information for us. 09:10.430 --> 09:13.350 So here are the newest articles about master agent. 09:13.870 --> 09:17.190 The Chinese agent draws comparison. 09:17.190 --> 09:17.280 Competition. 09:19.400 --> 09:25.000 It also gave us links so we can go directly to the article and read it. 09:29.560 --> 09:35.200 So yes, it provided seven seven articles in total. 09:39.280 --> 09:47.280 On the right hand side, you can see the entire process on how our agent performed this search, but 09:47.280 --> 09:51.200 we will go into detail how it works in the next lessons. 09:53.400 --> 09:57.160 I'm going to dive deep into agents and how they work. 09:57.360 --> 09:58.400 In the next lesson. 09:58.920 --> 10:05.520 This was just a short introduction to help you understand how agents differ from AI automation. 10:07.400 --> 10:10.400 Now let's quickly recap what we covered today. 10:11.000 --> 10:16.920 So automation is about setting up tasks to run automatically without human intervention. 10:17.520 --> 10:19.820 It follows a simple structure. 10:20.460 --> 10:22.940 So in every automation we have a trigger. 10:23.580 --> 10:26.260 An event happens at an action. 10:26.820 --> 10:28.860 So a task is executed. 10:30.500 --> 10:34.540 A automation builds on this by adding artificial intelligence. 10:35.260 --> 10:43.580 So a large language model acts as the brain and enables the automation to make smart decisions, analyze 10:43.580 --> 10:46.460 data and respond intelligently. 10:47.580 --> 10:50.500 And agents, take it a step further. 10:51.180 --> 10:59.780 They not only use an LLM as their brain, but also have tools they can call interact with other systems 10:59.780 --> 11:07.460 and perform complex tasks independently, autonomously, like responding to messages, searching the 11:07.460 --> 11:10.060 web, managing emails, etc.. 11:10.980 --> 11:17.100 And in the next lesson, I'm going to dive deep into agents and explain the difference between a agent's 11:17.420 --> 11:21.380 assistance, multi-agent systems and rack.