1 00:00:03,590 --> 00:00:15,200 So one of the things that, uh, we software developers, uh, do a lot is to interact with external 2 00:00:15,200 --> 00:00:16,160 APIs. 3 00:00:16,160 --> 00:00:19,790 So we request data from external APIs. 4 00:00:19,790 --> 00:00:27,800 And in order to do that, until now, we have to do it by writing code and understanding the API endpoints. 5 00:00:28,280 --> 00:00:36,200 So this is a process that is really it takes time effort you know, trial and error etc.. 6 00:00:36,200 --> 00:00:36,560 No. 7 00:00:36,560 --> 00:00:43,880 And now what we want is to request data from an external API using natural language. 8 00:00:44,420 --> 00:00:50,630 And as you will see, we can do this with an LM application very easily. 9 00:00:50,930 --> 00:00:54,680 And it's going to be a very simple process. 10 00:00:55,190 --> 00:01:00,620 Uh, just with these, uh, three steps here, you are going to define the API. 11 00:01:00,800 --> 00:01:07,580 You are going to create a chain, and you are going to start asking questions in natural language to 12 00:01:07,580 --> 00:01:09,770 request data to this API. 13 00:01:09,770 --> 00:01:17,900 So if you look at the right side of the screen, you see that as always we ask for the OpenAI API key 14 00:01:18,230 --> 00:01:23,000 and we start creating an instance of the LM model. 15 00:01:23,240 --> 00:01:32,870 Once we have that, and this is a crucial step, we are going to define what we can call our API documentation. 16 00:01:33,410 --> 00:01:40,760 This is very important because what we enter here is really what matters to this application. 17 00:01:40,760 --> 00:01:44,690 So in this case for example we are going to use an API. 18 00:01:45,360 --> 00:01:53,340 That is associated with a huge database with data about, you know, all the countries in the world, 19 00:01:53,340 --> 00:01:53,610 right? 20 00:01:53,610 --> 00:01:59,250 So they have information about, you know, the number of people living there, the language they speak, 21 00:01:59,250 --> 00:02:01,740 and thousands of different indicators. 22 00:02:01,740 --> 00:02:02,250 Right. 23 00:02:02,640 --> 00:02:14,040 But the only information we are going to access is the information we define here in our API documentation, 24 00:02:14,040 --> 00:02:19,770 because this is the information that is going to be passed to the GPT. 25 00:02:19,770 --> 00:02:22,710 So this is like the prompt right. 26 00:02:23,310 --> 00:02:31,290 So in this prompt what we are saying is this is the URL where the API is the base URL. 27 00:02:31,290 --> 00:02:40,950 And in this API we are going to show interest in three endpoints just in three endpoints instead of 28 00:02:40,950 --> 00:02:43,620 I don't know 100 endpoints that they provide. 29 00:02:45,540 --> 00:02:51,900 So in this case, the only thing we are telling ChatGPT that we are interested in is the name of the 30 00:02:51,900 --> 00:02:56,400 country, the currency they use and the language they speak there. 31 00:02:57,780 --> 00:02:58,860 So. 32 00:03:01,100 --> 00:03:10,850 Once we have that, the only thing we need to to do is to import a predefined chain from line chain 33 00:03:10,850 --> 00:03:11,780 called API. 34 00:03:11,810 --> 00:03:17,060 Chain this chain we are going to configure our our instance. 35 00:03:17,150 --> 00:03:23,630 Instance of this chain using the LM we created and this API documentation. 36 00:03:23,630 --> 00:03:26,090 We have configured this verbose through. 37 00:03:26,120 --> 00:03:26,990 Don't worry about that. 38 00:03:26,990 --> 00:03:30,110 You can just remove it and it will work the same. 39 00:03:30,710 --> 00:03:37,880 So once we have that we just make questions to the API. 40 00:03:38,600 --> 00:03:42,800 So we can ask this question. 41 00:03:42,800 --> 00:03:46,670 Give me information about France in less than 100 words. 42 00:03:46,820 --> 00:03:56,000 So ChatGPT is going to this external API is going to look for information about France using the endpoint. 43 00:03:56,000 --> 00:03:58,730 And it's going to provide us this summary. 44 00:03:58,730 --> 00:04:06,620 We can say we can ask questions more you know complex like list the top three biggest countries where 45 00:04:06,620 --> 00:04:08,330 the official language is French. 46 00:04:08,330 --> 00:04:12,680 So in this case they are looking for the official language. 47 00:04:12,680 --> 00:04:18,980 And also three countries, uh, from biggest to smaller. 48 00:04:19,970 --> 00:04:23,870 And probably they are using the population size here, I don't know. 49 00:04:24,080 --> 00:04:30,680 So, uh, the answer here is that the top three biggest countries with official where the official language 50 00:04:30,680 --> 00:04:34,880 is French are France, Canada and Belgium. 51 00:04:34,880 --> 00:04:41,240 And I understand that here they are using the population size, because my understanding is that Canada 52 00:04:41,240 --> 00:04:48,320 is a larger country than France in, in, in, in size. 53 00:04:48,320 --> 00:04:50,990 But, uh, well, I don't know. 54 00:04:50,990 --> 00:04:56,540 So this may be a question to, to to study further. 55 00:04:56,540 --> 00:04:56,990 Right. 56 00:04:56,990 --> 00:05:07,400 But right now what we have is a very, very simple application in order to request data from an external 57 00:05:07,400 --> 00:05:09,950 API using a natural language. 58 00:05:09,950 --> 00:05:17,840 This is really powerful and is going to change the way we as a software developers or software engineers 59 00:05:17,840 --> 00:05:19,250 work on a daily basis. 60 00:05:19,250 --> 00:05:21,050 So this is really revolutionary. 61 00:05:21,050 --> 00:05:29,720 And we are, just as I say, in the elemental, simple basic stage of LM applications.