1 00:00:03,160 --> 00:00:12,970 In this basic application, this basic Elm application, we are going to ask using natural language, 2 00:00:13,240 --> 00:00:15,070 a database. 3 00:00:15,070 --> 00:00:18,280 We are going to to ask questions to a database. 4 00:00:18,280 --> 00:00:25,150 And this is really revolutionary because until now, as you know we are having this problem. 5 00:00:25,150 --> 00:00:31,330 So to get information from a database we had to use complex languages like SQL etc.. 6 00:00:31,330 --> 00:00:38,500 So this limited a lot the type of people that can ask questions to a database. 7 00:00:38,500 --> 00:00:44,320 If you are not familiar with the SQL language, you cannot do anything in this in this field, right? 8 00:00:44,320 --> 00:00:51,040 But now what we want is that anyone in our company, anyone in our organization, using natural language 9 00:00:51,040 --> 00:00:55,360 or common language, common English, can ask questions from a database. 10 00:00:55,360 --> 00:00:56,860 This is a revolutionary. 11 00:00:57,550 --> 00:01:02,440 So launching can do this in a very, very simple way. 12 00:01:02,440 --> 00:01:09,460 As you will see we are going to use a predefined chain and it will do the job for us. 13 00:01:09,460 --> 00:01:12,910 So the process is going to be very, very simple. 14 00:01:12,910 --> 00:01:15,010 We are going to load the database. 15 00:01:15,010 --> 00:01:21,790 We are going to create a predefined chain and going to we are going to start asking questions in natural 16 00:01:21,790 --> 00:01:22,480 language. 17 00:01:22,480 --> 00:01:29,950 So if you look at the code in the right side of the screen, the first thing we do is to load the OpenAI 18 00:01:29,950 --> 00:01:34,600 API AP api key as we do always. 19 00:01:34,600 --> 00:01:41,800 And once that is complete, we uh, we are going to load. 20 00:01:41,800 --> 00:01:45,400 Well, we are going to create an instance of the of the Elm. 21 00:01:45,520 --> 00:01:50,320 And uh, we are going to import a couple of components. 22 00:01:50,320 --> 00:01:55,480 One is SQL database and the other one is SQL database chain. 23 00:01:55,810 --> 00:02:01,930 This second one we are getting it from long chain experimental okay. 24 00:02:01,930 --> 00:02:07,450 So this is not the, uh, final approved version of long chain. 25 00:02:07,450 --> 00:02:17,230 This is just like a candidate, uh, for, uh, you know, uh, a final version of a long chain. 26 00:02:17,800 --> 00:02:22,930 This is we are using this way because it's interesting for us right now. 27 00:02:22,930 --> 00:02:27,280 You will see that we have different, many different options in next chapters. 28 00:02:27,280 --> 00:02:29,860 But right now this serves us well. 29 00:02:29,860 --> 00:02:33,880 So the first thing we do is to load the database. 30 00:02:34,420 --> 00:02:42,100 And once we have that, the only thing we have to do is to create a chain using this SQL database chain 31 00:02:42,640 --> 00:02:45,730 and chain does everything for us in this change. 32 00:02:45,730 --> 00:02:49,540 In this chain we enter the Elm and the database. 33 00:02:49,540 --> 00:02:53,080 This one is is is nothing but was true. 34 00:02:53,110 --> 00:02:56,350 We can just remove this thing here. 35 00:02:56,350 --> 00:03:00,880 And the only thing we have to do is to execute this chain. 36 00:03:01,180 --> 00:03:03,760 Once we execute this chain. 37 00:03:03,760 --> 00:03:05,290 And with a question. 38 00:03:06,100 --> 00:03:10,300 ChatGPT starts providing responses. 39 00:03:10,300 --> 00:03:13,870 So if we ask the database we have loaded. 40 00:03:13,870 --> 00:03:24,010 Here is a short version of a public database from the, uh, San Francisco city. 41 00:03:24,910 --> 00:03:31,060 So in this database we have information about the trees in the city, blah blah, blah, blah. 42 00:03:31,060 --> 00:03:31,690 Right. 43 00:03:31,690 --> 00:03:38,080 This is a huge database we have reduced with only 1000 rows. 44 00:03:38,080 --> 00:03:42,130 I think the original one has like 80,000 rows or something like that. 45 00:03:42,460 --> 00:03:45,790 So we are making a couple of questions. 46 00:03:46,090 --> 00:03:49,960 The first one is how many species of trees are in San Francisco. 47 00:03:49,960 --> 00:03:54,850 The second one, how many trees of the species Ficus nitida are in San Francisco. 48 00:03:55,030 --> 00:04:01,930 This is the answer that ChatGPT is going to provide, and in both cases the answers are correct. 49 00:04:02,440 --> 00:04:11,290 We can check that the answers are correct if we are use if we use SQL, for example, or if we use a 50 00:04:11,500 --> 00:04:19,450 Gpt4, we load the database as a file in ChatGPT four and we ask questions about that. 51 00:04:19,450 --> 00:04:24,370 The difference of doing this or using the long chain approach is that with the long chain approach, 52 00:04:24,370 --> 00:04:27,940 we don't need to load the whole database to long chain. 53 00:04:27,940 --> 00:04:34,150 And the process of searching information is going to be much more convenient in terms of time and cost, 54 00:04:34,150 --> 00:04:34,720 etc.. 55 00:04:34,720 --> 00:04:35,140 Right? 56 00:04:35,140 --> 00:04:44,290 So as you can see, a long chain has a another, uh, very easy and fast way to solve a very important 57 00:04:44,290 --> 00:04:45,070 problem for us. 58 00:04:45,070 --> 00:04:53,290 So imagine in every company and organization that has this huge, uh, databases that until now were 59 00:04:53,290 --> 00:04:57,490 reserved for the software guys, you know, the database guys. 60 00:04:57,490 --> 00:04:59,710 And now they are available for everybody. 61 00:04:59,710 --> 00:05:04,930 You know, the CEO, people in the finance department, in the marketing department, everybody in the 62 00:05:04,930 --> 00:05:08,020 company can start making questions to the database. 63 00:05:08,020 --> 00:05:10,330 It is really, really amazing.