1 00:00:00,000 --> 00:00:00,000 Hello guys. 2 00:00:00,000 --> 00:00:08,000 So in this video we are going to develop a multi-agent generative AI application using line graph uh, 3 00:00:08,000 --> 00:00:10,000 wherein we will try to create a chatbot. 4 00:00:10,000 --> 00:00:16,000 Along with the chatbot, we will also be integrating with some third party tool so that whenever we 5 00:00:16,000 --> 00:00:22,000 ask a query with the chatbot, if the query is being able to handle by the chatbot, uh, which is specifically 6 00:00:22,000 --> 00:00:26,000 being using some llms, then we will directly get the response. 7 00:00:26,000 --> 00:00:32,000 But if it does not has the answer, it will go ahead and ask this specific external tool. 8 00:00:32,000 --> 00:00:37,000 So, uh, the entire flow or the architecture looks something like this. 9 00:00:37,000 --> 00:00:41,000 Here we will go ahead and start this process here. 10 00:00:41,000 --> 00:00:46,000 Any user if if any user is specifically asking any query, first of all the query will go to the chat 11 00:00:46,000 --> 00:00:47,000 bot. 12 00:00:47,000 --> 00:00:52,000 Then the chat bot will decide whether it should specifically provide the answer, or should it. 13 00:00:52,000 --> 00:00:56,000 Probably go ahead and ask some third party tool. 14 00:00:56,000 --> 00:00:56,000 Right? 15 00:00:56,000 --> 00:00:59,000 The third party tool can be, uh, Wikipedia. 16 00:00:59,000 --> 00:01:01,000 It can be Google Search anything. 17 00:01:01,000 --> 00:01:01,000 It can be right. 18 00:01:01,000 --> 00:01:08,000 And this chatbot that you will be seeing, it will also be interacting with some open source LLM models. 19 00:01:08,000 --> 00:01:10,000 And this open source LLM model. 20 00:01:10,000 --> 00:01:13,000 We will be accessing it from the grok API. 21 00:01:14,000 --> 00:01:20,000 So now we are going to go ahead with respect to one additional step with respect to land graph like 22 00:01:20,000 --> 00:01:22,000 how we can integrate some tools. 23 00:01:22,000 --> 00:01:26,000 You know here I will be going showing an example with respect to one tool like this. 24 00:01:26,000 --> 00:01:28,000 You can add any number of tool. 25 00:01:28,000 --> 00:01:31,000 And when the interaction is done with the tool you get the response. 26 00:01:31,000 --> 00:01:33,000 And then you finally end this entire flow okay. 27 00:01:33,000 --> 00:01:37,000 So let's go ahead and implement this step by step. 28 00:01:37,000 --> 00:01:42,000 Now first of all, what we really need to do is that we need to go ahead and install all these libraries. 29 00:01:42,000 --> 00:01:47,000 That is uh, land graph lang smith lang chain lang chain underscore grok. 30 00:01:47,000 --> 00:01:52,000 You know why lang chain underscore grok is specifically used so that you can actually communicate with 31 00:01:52,000 --> 00:01:53,000 the grok API. 32 00:01:53,000 --> 00:01:55,000 Then you have Lang chain community. 33 00:01:55,000 --> 00:01:55,000 Right. 34 00:01:55,000 --> 00:01:58,000 So I will go ahead and execute this. 35 00:01:58,000 --> 00:02:02,000 So once you specifically execute this here line graph is going to get this. 36 00:02:02,000 --> 00:02:04,000 All these libraries will be installed. 37 00:02:04,000 --> 00:02:05,000 Okay. 38 00:02:05,000 --> 00:02:10,000 Now uh, after installing this we will go ahead and import two more libraries. 39 00:02:10,000 --> 00:02:12,000 One is annotated and type dict. 40 00:02:12,000 --> 00:02:14,000 The reason is very simple. 41 00:02:14,000 --> 00:02:20,000 Uh, if you have already seen my previous video, we are going to use this two libraries in our state 42 00:02:20,000 --> 00:02:23,000 class which will be managing the state of the line graph. 43 00:02:23,000 --> 00:02:24,000 Right. 44 00:02:24,000 --> 00:02:25,000 So let's go ahead and execute this. 45 00:02:25,000 --> 00:02:31,000 So once we have executed this I will go ahead and start writing my code right now. 46 00:02:31,000 --> 00:02:32,000 First thing first. 47 00:02:32,000 --> 00:02:38,000 What we really need to do is that I will also go ahead and install some tools, you know, and Lang 48 00:02:38,000 --> 00:02:44,000 already provides you some amazing tools and tools here that we are going to use is specifically like 49 00:02:44,000 --> 00:02:46,000 Wikipedia or RCF, you know. 50 00:02:46,000 --> 00:02:52,000 So if you have already explored Lang, so already there are a lot of third party tools specifically 51 00:02:52,000 --> 00:02:53,000 available, right? 52 00:02:53,000 --> 00:02:56,000 So for that, I will go ahead and write a comment here. 53 00:02:56,000 --> 00:03:00,000 We are going to work working with tools or external tools. 54 00:03:00,000 --> 00:03:12,000 So first of all, I will go ahead and write from long chain underscore community dot utilities dot utilities. 55 00:03:12,000 --> 00:03:16,000 I'm going to go ahead and import two tools I can actually use. 56 00:03:16,000 --> 00:03:19,000 But right now I will just show you an example with respect to one tool. 57 00:03:19,000 --> 00:03:21,000 One is rc API wrapper. 58 00:03:21,000 --> 00:03:25,000 And the second tool that I'm actually going to use is something like Wikipedia API wrapper. 59 00:03:25,000 --> 00:03:31,000 Right now RCF is basically if you are looking for any research papers, you can basically go ahead and 60 00:03:31,000 --> 00:03:34,000 interact with this particular wrapper instead. 61 00:03:34,000 --> 00:03:38,000 If you just want to get the search from Wikipedia, you can go ahead and take it from this particular 62 00:03:38,000 --> 00:03:39,000 library. 63 00:03:39,000 --> 00:03:39,000 Okay. 64 00:03:39,000 --> 00:03:49,000 Then the next step will be that from Lang chain underscore community dot tools import rc query run. 65 00:03:49,000 --> 00:03:52,000 This will be responsible in running any queries that you specifically give. 66 00:03:52,000 --> 00:03:55,000 And then you have this Wikipedia query run okay. 67 00:03:56,000 --> 00:03:59,000 So this is for the RC tool. 68 00:03:59,000 --> 00:04:02,000 This is specifically for the Wikipedia tool. 69 00:04:02,000 --> 00:04:09,000 Okay, then, uh, after this, I'm just going to go ahead and call my RCF, uh, and Wikipedia tools. 70 00:04:09,000 --> 00:04:12,000 So let's say Wikipedia tools okay. 71 00:04:12,000 --> 00:04:23,000 Then here I'm going to write my RCF underscore wrapper is equal to RCF, rcf API wrapper here. 72 00:04:23,000 --> 00:04:29,000 Uh, let's say in this API wrapper how many results I want, I will go ahead and provide a parameter 73 00:04:29,000 --> 00:04:31,000 that is top-k results. 74 00:04:31,000 --> 00:04:32,000 Let's say top-k results. 75 00:04:32,000 --> 00:04:38,000 I just want one result and I will go ahead and write my document content character Max. 76 00:04:38,000 --> 00:04:40,000 Like how many characters I specifically want. 77 00:04:40,000 --> 00:04:42,000 Let's say I will go ahead and write 300. 78 00:04:42,000 --> 00:04:42,000 Okay. 79 00:04:42,000 --> 00:04:48,000 Then uh, with the help of this RCF, what I'm actually going to do, I'll go ahead and create this 80 00:04:48,000 --> 00:04:49,000 specific tool. 81 00:04:49,000 --> 00:04:54,000 I will use this R safe query run, and I will give the wrapper as a parameter over here. 82 00:04:55,000 --> 00:04:55,000 Right. 83 00:04:55,000 --> 00:04:59,000 One amazing thing about Google Colab is that it is now providing you code suggestion, which will be 84 00:04:59,000 --> 00:05:02,000 very helpful whenever you are writing any code. 85 00:05:02,000 --> 00:05:07,000 Similarly, if I go ahead and write my API underscore wrapper, I will go ahead and use my Wikipedia 86 00:05:07,000 --> 00:05:08,000 wrapper. 87 00:05:08,000 --> 00:05:13,000 Here also, I will go ahead and write my top k results and document content characters. 88 00:05:13,000 --> 00:05:14,000 Max characters. 89 00:05:14,000 --> 00:05:16,000 I really want 300, right? 90 00:05:16,000 --> 00:05:21,000 That basically means whenever we are interacting with Wikipedia, I'm going to get the response as one 91 00:05:21,000 --> 00:05:23,000 document at a time. 92 00:05:23,000 --> 00:05:26,000 And inside that document, the maximum number of characters will be 300. 93 00:05:26,000 --> 00:05:31,000 So finally, I will go ahead and also write about Wikipedia tool where I'm going to use this Wikipedia 94 00:05:31,000 --> 00:05:32,000 run query run. 95 00:05:32,000 --> 00:05:34,000 And we are going to run this particular API wrapper. 96 00:05:34,000 --> 00:05:36,000 So two things I have actually imported it. 97 00:05:36,000 --> 00:05:39,000 Now let's say let's go ahead and just uh okay. 98 00:05:39,000 --> 00:05:42,000 One uh error is basically there. 99 00:05:42,000 --> 00:05:45,000 Let's say okay, I have not installed. 100 00:05:46,000 --> 00:05:54,000 So we need to install RCF and Wikipedia okay. 101 00:05:54,000 --> 00:05:57,000 So without this this will not work okay. 102 00:05:57,000 --> 00:06:00,000 So let's go ahead and execute this quickly. 103 00:06:00,000 --> 00:06:02,000 So this is going to get installed. 104 00:06:02,000 --> 00:06:05,000 Here you can see that it is getting installed. 105 00:06:05,000 --> 00:06:09,000 Because without this particular packages we will not be able to execute this code. 106 00:06:09,000 --> 00:06:15,000 Right now it should basically run up here I'm getting an error saying that base tool underscore unit 107 00:06:15,000 --> 00:06:19,000 okay RC tools takes RC wrapper okay. 108 00:06:19,000 --> 00:06:21,000 This is my RC wrapper. 109 00:06:21,000 --> 00:06:26,000 I need to provide the parameter because over here I will go ahead and give my API underscore wrapper 110 00:06:26,000 --> 00:06:26,000 as my parameter. 111 00:06:26,000 --> 00:06:27,000 Okay. 112 00:06:27,000 --> 00:06:30,000 So I'm not going to edit any of this particular errors. 113 00:06:30,000 --> 00:06:35,000 So please make sure that and this is the key that is specifically used. 114 00:06:35,000 --> 00:06:36,000 That is the reason it was giving an error. 115 00:06:36,000 --> 00:06:38,000 Now it should run okay. 116 00:06:38,000 --> 00:06:39,000 Now this is run perfectly. 117 00:06:39,000 --> 00:06:41,000 Now let's go ahead and run any of this. 118 00:06:41,000 --> 00:06:50,000 So I will go ahead and write Wikipedia dot tool and let me go ahead and ask who is who is. 119 00:06:50,000 --> 00:06:54,000 I'll ask any question who is a famous actor? 120 00:06:54,000 --> 00:06:56,000 Let's say Shahrukh Khan. 121 00:06:56,000 --> 00:06:56,000 Okay. 122 00:06:57,000 --> 00:07:02,000 So here I'm going to write Wiki wiki wiki underscore two. 123 00:07:02,000 --> 00:07:04,000 It should be dot invoke. 124 00:07:04,000 --> 00:07:11,000 Okay, so if I really want to invoke any of this invoke and let me go ahead and execute it now, it 125 00:07:11,000 --> 00:07:13,000 should probably go ahead and search from Wikipedia. 126 00:07:13,000 --> 00:07:15,000 And I'm getting the response over here. 127 00:07:15,000 --> 00:07:17,000 The maximum number of characters will be 300. 128 00:07:17,000 --> 00:07:21,000 Similarly if you can go ahead and use this RCF tool also. 129 00:07:21,000 --> 00:07:28,000 So if I go ahead and write RCF underscore tool dot invoke okay. 130 00:07:29,000 --> 00:07:32,000 Attention is all you need. 131 00:07:32,000 --> 00:07:36,000 So this is one of the research paper related to Transformers. 132 00:07:36,000 --> 00:07:38,000 So here you can see I'm getting all the information. 133 00:07:38,000 --> 00:07:40,000 You do not need all the information. 134 00:07:40,000 --> 00:07:44,000 So this is basically coming from the RCF itself right. 135 00:07:44,000 --> 00:07:48,000 And all the author research paper names everything is basically available over here. 136 00:07:48,000 --> 00:07:49,000 So perfect. 137 00:07:49,000 --> 00:07:53,000 That basically means my external tools are working absolutely fine. 138 00:07:53,000 --> 00:07:54,000 Right. 139 00:07:54,000 --> 00:07:57,000 But since I'm just going to use one tool, you can also use any number of tools. 140 00:07:57,000 --> 00:08:01,000 So I'll just go ahead and write this as wiki over here or wiki underscore tool. 141 00:08:01,000 --> 00:08:03,000 So this will basically be my tool. 142 00:08:03,000 --> 00:08:07,000 Uh, afterwards let's say if you want to go ahead and add any number of tool, just go ahead and write 143 00:08:07,000 --> 00:08:09,000 comma and keep on adding all the tool. 144 00:08:09,000 --> 00:08:10,000 Right. 145 00:08:10,000 --> 00:08:13,000 So now my third party tool is done. 146 00:08:13,000 --> 00:08:17,000 Now I'm going to create my line graph application okay. 147 00:08:17,000 --> 00:08:20,000 Line graph applications. 148 00:08:20,000 --> 00:08:25,000 Now in order to start line graph application already in our previous video we have seen I'm going to 149 00:08:25,000 --> 00:08:26,000 create class state. 150 00:08:26,000 --> 00:08:28,000 And here I'm going to use my type dict. 151 00:08:28,000 --> 00:08:29,000 Okay. 152 00:08:29,000 --> 00:08:35,000 The reason why I'm using this because I'm going to create this messages variable which will be a list 153 00:08:35,000 --> 00:08:36,000 of variables okay. 154 00:08:36,000 --> 00:08:38,000 And I'm going to use the annotated. 155 00:08:38,000 --> 00:08:46,000 And I have to make sure that whenever the execution of a specific node gets completed, I'm going to 156 00:08:46,000 --> 00:08:50,000 go ahead and just make sure to update this messages variable. 157 00:08:50,000 --> 00:08:52,000 So here I'm going to get my list. 158 00:08:52,000 --> 00:08:55,000 And here I'm going to get my add underscore messages okay. 159 00:08:55,000 --> 00:08:59,000 So once this is done I add underscore messages. 160 00:08:59,000 --> 00:09:03,000 Also I need to import and this add underscore messages. 161 00:09:03,000 --> 00:09:07,000 If you remember from my previous video already it is already available in the line graph. 162 00:09:07,000 --> 00:09:14,000 So let me just go ahead and import this okay line graph dot message import add message. 163 00:09:14,000 --> 00:09:17,000 And here I'm going to basically use this specific state. 164 00:09:17,000 --> 00:09:18,000 Right. 165 00:09:18,000 --> 00:09:20,000 So all these are completed. 166 00:09:20,000 --> 00:09:25,000 And we have this particular add messages and we are good to go okay great. 167 00:09:25,000 --> 00:09:34,000 Now the next thing uh along with this that we really need to import is three important things from line 168 00:09:34,000 --> 00:09:34,000 graph. 169 00:09:34,000 --> 00:09:42,000 That is your state graph start and end I need to import all these things so that I get all the information 170 00:09:42,000 --> 00:09:43,000 that I really want okay. 171 00:09:44,000 --> 00:09:45,000 Uh, with respect to this. 172 00:09:45,000 --> 00:09:46,000 So this is done. 173 00:09:46,000 --> 00:09:50,000 Now let me go ahead and first of all create my chatbot. 174 00:09:50,000 --> 00:09:51,000 Okay. 175 00:09:51,000 --> 00:09:55,000 Uh, because uh, already we have created this particular state class. 176 00:09:55,000 --> 00:09:58,000 Now we need to probably design our entire workflow Right. 177 00:09:58,000 --> 00:10:00,000 So here I'm going to go ahead and execute this. 178 00:10:01,000 --> 00:10:09,000 Now let me just go ahead and right quickly I will go ahead and create my graph builder okay is equal 179 00:10:09,000 --> 00:10:12,000 to graph Builder I will create this specific variable. 180 00:10:12,000 --> 00:10:18,000 And here I'm going to go ahead and write state graph or state graph. 181 00:10:18,000 --> 00:10:21,000 And here we are specifically going to take this state. 182 00:10:21,000 --> 00:10:24,000 Now this state graph is nothing but it is imported from here. 183 00:10:24,000 --> 00:10:26,000 This state is nothing but my class. 184 00:10:26,000 --> 00:10:32,000 The state is going to maintain all the all the all those different different states in this particular 185 00:10:32,000 --> 00:10:33,000 messages variable. 186 00:10:33,000 --> 00:10:33,000 Okay. 187 00:10:33,000 --> 00:10:36,000 So this will basically be my graph builder. 188 00:10:36,000 --> 00:10:39,000 Now it's time to probably call my LM model. 189 00:10:39,000 --> 00:10:39,000 Right. 190 00:10:39,000 --> 00:10:45,000 Because at the end of the day we if we really want to create any chatbot, we really need to use a specific 191 00:10:45,000 --> 00:10:45,000 LM model. 192 00:10:45,000 --> 00:10:51,000 Now, I have already shown you in our previous video how we can actually go ahead and use LM models, 193 00:10:51,000 --> 00:10:51,000 right? 194 00:10:51,000 --> 00:10:55,000 Or how you can use or how you can call the LM model from Chad Brock. 195 00:10:55,000 --> 00:10:55,000 Right. 196 00:10:55,000 --> 00:10:59,000 So here I will go ahead and right from Lange Lang chain. 197 00:10:59,000 --> 00:11:00,000 Underscore. 198 00:11:00,000 --> 00:11:03,000 Grok import chart. 199 00:11:03,000 --> 00:11:04,000 Grok okay. 200 00:11:04,000 --> 00:11:05,000 Chart. 201 00:11:05,000 --> 00:11:05,000 Grok. 202 00:11:05,000 --> 00:11:09,000 And inside this chart grok I'm just going to go ahead and execute this. 203 00:11:09,000 --> 00:11:15,000 And then I'm going to create my LM variable which is nothing but which is my chart grok. 204 00:11:15,000 --> 00:11:21,000 And here you can actually go ahead and see my grok underscore API underscore key. 205 00:11:21,000 --> 00:11:27,000 This API key I should be getting from my uh, secret keys. 206 00:11:27,000 --> 00:11:27,000 Right? 207 00:11:27,000 --> 00:11:31,000 I have already added the secret keys, and I've already shown you how you can basically add it in the 208 00:11:31,000 --> 00:11:32,000 previous video. 209 00:11:32,000 --> 00:11:32,000 Right. 210 00:11:32,000 --> 00:11:36,000 So in order to add this, I will go ahead and write one more code. 211 00:11:36,000 --> 00:11:47,000 I will go ahead and write from Google Dot Colab, import user data, and then we will go ahead and write 212 00:11:47,000 --> 00:11:49,000 grok underscore API. 213 00:11:49,000 --> 00:11:55,000 Underscore key is equal to user data dot. 214 00:11:55,000 --> 00:12:03,000 Get And here I'm going to specifically use my grok underscore API underscore key, which is the key 215 00:12:03,000 --> 00:12:05,000 name that I've written it over here in my secret keys. 216 00:12:05,000 --> 00:12:06,000 Right. 217 00:12:06,000 --> 00:12:07,000 Grok underscore API underscore key. 218 00:12:07,000 --> 00:12:08,000 Perfect. 219 00:12:08,000 --> 00:12:09,000 So this is fine. 220 00:12:09,000 --> 00:12:13,000 Now once I get my grok underscore API underscore key I will just go and click on grant access. 221 00:12:13,000 --> 00:12:17,000 And finally I will keep the value over here. 222 00:12:17,000 --> 00:12:20,000 So this should be grok underscore API underscore key. 223 00:12:20,000 --> 00:12:23,000 And here finally this is my LLM model. 224 00:12:24,000 --> 00:12:28,000 And this LLM model by default is using no no models. 225 00:12:28,000 --> 00:12:30,000 So I have to provide my model name. 226 00:12:30,000 --> 00:12:32,000 So the model name will be nothing. 227 00:12:32,000 --> 00:12:36,000 But I will go ahead and write model name is equal to gamma. 228 00:12:36,000 --> 00:12:37,000 I can use gamma. 229 00:12:38,000 --> 00:12:40,000 So let's go ahead and use gamma. 230 00:12:41,000 --> 00:12:45,000 Gamma two nine be it. 231 00:12:45,000 --> 00:12:46,000 Okay. 232 00:12:47,000 --> 00:12:49,000 So I've specifically used this particular model. 233 00:12:49,000 --> 00:12:51,000 And let's go ahead and execute this. 234 00:12:51,000 --> 00:12:52,000 Perfect. 235 00:12:53,000 --> 00:12:54,000 This is done. 236 00:12:54,000 --> 00:12:56,000 My graph builder is also ready. 237 00:12:56,000 --> 00:13:00,000 Now it's time that we go ahead and create our chat bots, right? 238 00:13:00,000 --> 00:13:02,000 So we need to go ahead and define our chat bot. 239 00:13:03,000 --> 00:13:09,000 Now with the help of this LLM, we can basically call or we can call with our chat bot itself, whatever 240 00:13:09,000 --> 00:13:11,000 nodes we are specifically creating. 241 00:13:11,000 --> 00:13:11,000 Right. 242 00:13:11,000 --> 00:13:17,000 So let's go ahead and go to go with the next step wherein we will go ahead and create our chat bot, 243 00:13:17,000 --> 00:13:22,000 which will be doing a functionality of managing the state and also interacting with the LLM model. 244 00:13:22,000 --> 00:13:30,000 So let me go ahead and create my definition chat bot because see uh, if, if I show you over here right. 245 00:13:30,000 --> 00:13:32,000 We need to create this first. 246 00:13:32,000 --> 00:13:32,000 Right. 247 00:13:32,000 --> 00:13:33,000 My LLM is ready. 248 00:13:33,000 --> 00:13:35,000 It is interacting with the grok API. 249 00:13:35,000 --> 00:13:37,000 Now I'll create my chat bot. 250 00:13:37,000 --> 00:13:38,000 I will be interacting with this. 251 00:13:38,000 --> 00:13:43,000 But I will also be adding one very important thing that is nothing but my tools, right? 252 00:13:43,000 --> 00:13:45,000 My LLM should also be integrated with the tools. 253 00:13:45,000 --> 00:13:48,000 So let's see how to probably do that. 254 00:13:48,000 --> 00:13:51,000 So first of all, uh I will go ahead and create a cell over here. 255 00:13:51,000 --> 00:13:54,000 I've created my graph builder. 256 00:13:54,000 --> 00:13:55,000 Everything is ready. 257 00:13:55,000 --> 00:14:02,000 I will take this LLM and I will say hey, go ahead and bind underscore tools, whatever tools I'm going 258 00:14:02,000 --> 00:14:03,000 to use. 259 00:14:03,000 --> 00:14:05,000 So this will be my tools is equal to tools. 260 00:14:05,000 --> 00:14:10,000 And if you remember this tools has specifically what does it has a binding as Wikipedia tool. 261 00:14:10,000 --> 00:14:11,000 Right. 262 00:14:11,000 --> 00:14:14,000 So I'm going to basically bind this particular tool itself. 263 00:14:14,000 --> 00:14:17,000 And after binding I will store this in a variable. 264 00:14:17,000 --> 00:14:25,000 This is just giving a additional functionality to the LM that it has this particular tool access also. 265 00:14:25,000 --> 00:14:25,000 Right. 266 00:14:25,000 --> 00:14:28,000 So this LM with underscore tools will be used. 267 00:14:28,000 --> 00:14:34,000 Now once I'm creating this particular functionality I will first of all derive my uh, I will inherit 268 00:14:34,000 --> 00:14:40,000 this particular state inside this chatbot so that it has all the information related to the message. 269 00:14:40,000 --> 00:14:46,000 And here I'm going to probably return my output with, with the help of this messages variable, which 270 00:14:46,000 --> 00:14:50,000 is the key that is present inside the state colon. 271 00:14:50,000 --> 00:14:55,000 And here I'm going to probably call or invoke with the help of my LM model. 272 00:14:55,000 --> 00:15:01,000 But this time my LM model will also have the functionalities from the tools, right? 273 00:15:01,000 --> 00:15:05,000 Because it also has the tools functionality which is nothing but the Wikipedia tool. 274 00:15:05,000 --> 00:15:07,000 And here I will go and write Dot invoke. 275 00:15:07,000 --> 00:15:12,000 And I'm going to probably take this particular state, this state which I'm inheriting, which is coming 276 00:15:12,000 --> 00:15:13,000 from this and inside this state. 277 00:15:13,000 --> 00:15:16,000 You know, what exactly is present messages variable is present. 278 00:15:16,000 --> 00:15:22,000 And if I take the state of messages, this is basically going to have the query of the user. 279 00:15:22,000 --> 00:15:22,000 Right. 280 00:15:22,000 --> 00:15:25,000 So once I execute this this is what it is basically going to return. 281 00:15:25,000 --> 00:15:26,000 Right. 282 00:15:26,000 --> 00:15:28,000 So my chatbot functionality is ready okay. 283 00:15:28,000 --> 00:15:30,000 This chatbot functionality is ready. 284 00:15:30,000 --> 00:15:33,000 This LM is being integrated with the LM tool. 285 00:15:33,000 --> 00:15:34,000 So it is basically going to interact. 286 00:15:34,000 --> 00:15:40,000 So whatever query comes, if this chatbot is able to probably provide the response from the LM, then 287 00:15:40,000 --> 00:15:41,000 it is going to get the output. 288 00:15:41,000 --> 00:15:44,000 Otherwise it is going to interact with this particular tool. 289 00:15:44,000 --> 00:15:44,000 Right. 290 00:15:44,000 --> 00:15:46,000 And finally we are just going to end this. 291 00:15:46,000 --> 00:15:50,000 Now we need to create this entire flow the flow of the execution. 292 00:15:50,000 --> 00:15:51,000 Right. 293 00:15:51,000 --> 00:15:55,000 So in order to create the flow of the execution I will again take my graph builder. 294 00:15:55,000 --> 00:16:03,000 So let me go ahead and write Graph builder dot add underscore node. 295 00:16:03,000 --> 00:16:04,000 Okay. 296 00:16:04,000 --> 00:16:07,000 And then we are basically going to write chatbot. 297 00:16:07,000 --> 00:16:10,000 So this will basically be my node name chatbot. 298 00:16:10,000 --> 00:16:14,000 And finally you will also be able to see this chatbot. 299 00:16:14,000 --> 00:16:14,000 Right. 300 00:16:14,000 --> 00:16:16,000 So this is my node. 301 00:16:16,000 --> 00:16:19,000 I will add the start with this chatbot node. 302 00:16:19,000 --> 00:16:20,000 Right. 303 00:16:20,000 --> 00:16:26,000 I will add that particular edge because we need this specific edge right from start to the chatbot. 304 00:16:26,000 --> 00:16:26,000 Perfect. 305 00:16:26,000 --> 00:16:28,000 This node is basically added. 306 00:16:28,000 --> 00:16:35,000 Now I have to also make sure that I need to add uh the tool node okay. 307 00:16:35,000 --> 00:16:41,000 So see whenever a tool is basically needed to add because here we will be adding this tool over here. 308 00:16:41,000 --> 00:16:42,000 Right. 309 00:16:42,000 --> 00:16:44,000 So this tool needs to be added to the chatbot. 310 00:16:44,000 --> 00:16:46,000 Then only the LM will be able to interact with the tool. 311 00:16:46,000 --> 00:16:52,000 So for that we basically use another library over here which is called as tool node. 312 00:16:52,000 --> 00:16:52,000 Okay. 313 00:16:52,000 --> 00:16:55,000 So let me just go ahead and import one more library. 314 00:16:55,000 --> 00:16:58,000 So I will go ahead and right from Landgraaf. 315 00:16:58,000 --> 00:17:04,000 In order to add any third party tools or any custom tools, you can also create this custom tools. 316 00:17:04,000 --> 00:17:11,000 I will go ahead and import one tool node and then along with this I will also give one tool condition. 317 00:17:11,000 --> 00:17:17,000 Like for this particular tool node there should be some kind of conditions we need to add right now. 318 00:17:17,000 --> 00:17:24,000 Now let's see what is the error tool node prebuilt from tool node okay. 319 00:17:24,000 --> 00:17:26,000 It should be tool conditions tools condition okay. 320 00:17:26,000 --> 00:17:29,000 No worries I will just go ahead and import this. 321 00:17:29,000 --> 00:17:29,000 Perfect. 322 00:17:29,000 --> 00:17:30,000 So this is done. 323 00:17:30,000 --> 00:17:31,000 Now what I have done. 324 00:17:31,000 --> 00:17:32,000 I have created a chat bot. 325 00:17:32,000 --> 00:17:36,000 I have connected from start node to the chat bot node. 326 00:17:36,000 --> 00:17:39,000 Then I will go ahead and call this tool node. 327 00:17:39,000 --> 00:17:41,000 So let me just go ahead and write tool node. 328 00:17:41,000 --> 00:17:45,000 And this tool node will basically have my tools. 329 00:17:45,000 --> 00:17:47,000 My tools is nothing. 330 00:17:47,000 --> 00:17:50,000 But whatever tools I am going to use that is nothing but Wikipedia. 331 00:17:50,000 --> 00:17:53,000 And I'll create a variable which is called as tool node. 332 00:17:53,000 --> 00:17:53,000 Okay. 333 00:17:53,000 --> 00:17:58,000 So this tool node is also created that basically means this tool node is also created. 334 00:17:58,000 --> 00:18:01,000 Now we need to connect from this chatbot. 335 00:18:01,000 --> 00:18:01,000 To the tool nodes. 336 00:18:01,000 --> 00:18:03,000 How do I connect it? 337 00:18:03,000 --> 00:18:04,000 It is very simple. 338 00:18:04,000 --> 00:18:05,000 I will go ahead and write. 339 00:18:05,000 --> 00:18:12,000 Graph graph builder dot add underscore node. 340 00:18:12,000 --> 00:18:14,000 And here I'm going to basically go. 341 00:18:14,000 --> 00:18:17,000 Ahead and create a node which is called as tools. 342 00:18:17,000 --> 00:18:19,000 And this tools is going to get connected. 343 00:18:19,000 --> 00:18:22,000 To tool underscore node right. 344 00:18:23,000 --> 00:18:26,000 So this way what we have done this entire node is getting. 345 00:18:26,000 --> 00:18:27,000 Getting created. 346 00:18:27,000 --> 00:18:28,000 Right. 347 00:18:28,000 --> 00:18:28,000 Right. 348 00:18:28,000 --> 00:18:31,000 So here this entire tool node is basically getting connected. 349 00:18:31,000 --> 00:18:37,000 Now we need to probably connect this tool to our chatbot now in order to connect it. 350 00:18:37,000 --> 00:18:40,000 Obviously we need to write some conditions. 351 00:18:40,000 --> 00:18:43,000 That basically means what does this basically mean. 352 00:18:43,000 --> 00:18:47,000 Like if I'm getting a query, this chatbot may either interact with the tools or it may. 353 00:18:48,000 --> 00:18:49,000 Interact with the LM and give you the output. 354 00:18:49,000 --> 00:18:50,000 Right. 355 00:18:50,000 --> 00:18:53,000 So this kind of edges if there is a. 356 00:18:53,000 --> 00:18:58,000 Bidirectional edge like this, we really need to use something called as add conditional edges. 357 00:18:58,000 --> 00:18:58,000 Right. 358 00:18:58,000 --> 00:19:09,000 So here what we are going to use is that I will write graph builder dot add add conditional edges. 359 00:19:09,000 --> 00:19:11,000 See add conditional edges. 360 00:19:11,000 --> 00:19:14,000 That is the reason we had imported somewhere. 361 00:19:14,000 --> 00:19:15,000 Let's see where we have imported it. 362 00:19:15,000 --> 00:19:17,000 Have we not imported this? 363 00:19:17,000 --> 00:19:18,000 Let's see. 364 00:19:19,000 --> 00:19:21,000 Add conditional edges. 365 00:19:22,000 --> 00:19:23,000 Okay, we have not imported it. 366 00:19:23,000 --> 00:19:26,000 So first of all we really need to import it. 367 00:19:26,000 --> 00:19:30,000 But no no need to import it because this is an inbuilt function inside Graph Builder. 368 00:19:30,000 --> 00:19:31,000 Okay. 369 00:19:31,000 --> 00:19:36,000 So here I'm going to basically go ahead and write add graph add conditional edges. 370 00:19:36,000 --> 00:19:47,000 And inside this particular edges here what edges will specifically add from chat bot to to tools underscore 371 00:19:47,000 --> 00:19:48,000 condition. 372 00:19:48,000 --> 00:19:49,000 Right. 373 00:19:49,000 --> 00:19:51,000 And this tools underscore condition is nothing. 374 00:19:51,000 --> 00:19:53,000 But it is basically coming from here. 375 00:19:53,000 --> 00:19:59,000 Now understand over here why we are adding this conditional edges because just from the diagram you 376 00:19:59,000 --> 00:20:00,000 can see that from the chatbot. 377 00:20:00,000 --> 00:20:02,000 I need to interact with the tool and again get it back. 378 00:20:02,000 --> 00:20:08,000 So whenever we have this kind of bidirectional we really need to add this particular edge which is called 379 00:20:08,000 --> 00:20:09,000 as conditional edges. 380 00:20:09,000 --> 00:20:15,000 I need to give my node name and the tool conditions are the default conditions that is available over 381 00:20:15,000 --> 00:20:15,000 here. 382 00:20:15,000 --> 00:20:20,000 This tools condition is going to make that bidirectional directions itself right from the chatbot. 383 00:20:20,000 --> 00:20:25,000 Now after that, uh, you know that my tools also needs to get connected to the chatbot. 384 00:20:26,000 --> 00:20:31,000 See this tools when I'm using this, add conditional edges, the chatbot, tilde tools, the communication 385 00:20:31,000 --> 00:20:34,000 will happen, but I need to reconnect it back. 386 00:20:34,000 --> 00:20:40,000 So for in order to reconnect it back I will just go ahead and write graph underscore builder dot. 387 00:20:40,000 --> 00:20:45,000 Add edge from tools to chatbot see from tools to chatbot. 388 00:20:45,000 --> 00:20:49,000 And finally we are going to add an edge which is from chatbot till the end. 389 00:20:49,000 --> 00:20:51,000 That is what we really want this end. 390 00:20:51,000 --> 00:20:53,000 So it is very simple. 391 00:20:53,000 --> 00:20:57,000 First of all we went this is the entire flow. 392 00:20:57,000 --> 00:20:58,000 And this is really important. 393 00:20:58,000 --> 00:21:01,000 Inside this flow we use three important things. 394 00:21:01,000 --> 00:21:09,000 One is add edges, add edges, add edges, add node. 395 00:21:10,000 --> 00:21:11,000 Right. 396 00:21:11,000 --> 00:21:14,000 And then you also had conditional edges. 397 00:21:16,000 --> 00:21:16,000 Right. 398 00:21:16,000 --> 00:21:18,000 All these three nodes we had. 399 00:21:18,000 --> 00:21:18,000 Right. 400 00:21:18,000 --> 00:21:21,000 So first of all we connected start to chatbot. 401 00:21:22,000 --> 00:21:25,000 So we created uh we created a node for chatbot. 402 00:21:25,000 --> 00:21:27,000 And we add this edge from start to chatbot. 403 00:21:27,000 --> 00:21:31,000 Then we created this LM tool integration. 404 00:21:31,000 --> 00:21:34,000 Then from chatbot to tools we specifically went right. 405 00:21:34,000 --> 00:21:40,000 So if we go ahead and just revise this step by step, see first of all we added a node chatbot which 406 00:21:40,000 --> 00:21:42,000 is having this particular functionality. 407 00:21:42,000 --> 00:21:46,000 And this functionality is nothing, but it is using this LM underscore with tools dot invoke. 408 00:21:46,000 --> 00:21:50,000 It is just trying to get the output from this particular message right. 409 00:21:50,000 --> 00:21:54,000 Whatever message is there, LM will either give the response or the tool will give the response. 410 00:21:54,000 --> 00:21:55,000 Perfect. 411 00:21:55,000 --> 00:21:59,000 Then we added the edges with start to chatbot at this specific node. 412 00:21:59,000 --> 00:22:01,000 Then we created our tool node. 413 00:22:01,000 --> 00:22:04,000 Then this tool node got added as a node in the graph builder. 414 00:22:04,000 --> 00:22:10,000 Then we added conditional edges between chatbot and we applied a tools condition. 415 00:22:10,000 --> 00:22:13,000 Then we added the edges back from tools to chatbot. 416 00:22:13,000 --> 00:22:19,000 And then finally we have this graph builder dot add edge where we connected the chatbot with the end 417 00:22:19,000 --> 00:22:19,000 right. 418 00:22:19,000 --> 00:22:21,000 So all these things were executed perfectly. 419 00:22:22,000 --> 00:22:24,000 Now let's go ahead and compile this. 420 00:22:24,000 --> 00:22:30,000 I'm going to basically use graph underscore builder dot compile. 421 00:22:30,000 --> 00:22:31,000 Okay. 422 00:22:31,000 --> 00:22:32,000 Compile. 423 00:22:32,000 --> 00:22:36,000 And then this will basically be my graph graph. 424 00:22:36,000 --> 00:22:38,000 And we will just go ahead and execute this. 425 00:22:38,000 --> 00:22:39,000 Okay. 426 00:22:39,000 --> 00:22:45,000 Now finally in order to display this graph I can go ahead and use the same thing which we have used 427 00:22:45,000 --> 00:22:48,000 earlier in our videos with respect to IPython dot display. 428 00:22:48,000 --> 00:22:50,000 This is just from the documentation. 429 00:22:50,000 --> 00:22:51,000 Now we can see the entire flow. 430 00:22:52,000 --> 00:22:56,000 See start chatbot tools end right. 431 00:22:56,000 --> 00:22:58,000 Everything is basically here. 432 00:22:58,000 --> 00:23:02,000 So finally you can see the flow over here after executing it. 433 00:23:02,000 --> 00:23:03,000 Let's go ahead and test it. 434 00:23:03,000 --> 00:23:05,000 So see from start to chatbot. 435 00:23:05,000 --> 00:23:08,000 If the tool is required then again it is connected back. 436 00:23:08,000 --> 00:23:10,000 And finally we go to the end. 437 00:23:10,000 --> 00:23:10,000 Right. 438 00:23:10,000 --> 00:23:16,000 So all these things are specifically connected and we are good to go with respect to this right now. 439 00:23:16,000 --> 00:23:23,000 Let's go ahead and try out some of the flow like with respect to any kind of messages okay. 440 00:23:23,000 --> 00:23:26,000 And let's see how we are going to get the output right. 441 00:23:26,000 --> 00:23:35,000 So first of all I will just go ahead and give my user underscore input is equal to I will write uh hi 442 00:23:35,000 --> 00:23:36,000 there. 443 00:23:37,000 --> 00:23:39,000 Let me just go ahead and write hi there. 444 00:23:39,000 --> 00:23:42,000 And here I'm just going to write. 445 00:23:42,000 --> 00:23:46,000 My name is Krish okay. 446 00:23:46,000 --> 00:23:51,000 I'm just going to probably provide some basic input from the user. 447 00:23:51,000 --> 00:23:51,000 Right. 448 00:23:52,000 --> 00:23:56,000 Then I'm going to go ahead and use an events variable. 449 00:23:56,000 --> 00:24:00,000 And let me go ahead and write graph dot stream. 450 00:24:00,000 --> 00:24:06,000 And here inside the stream I'm going to give my messages in this format. 451 00:24:06,000 --> 00:24:11,000 This is really important because this messages will follow in the C any user message. 452 00:24:11,000 --> 00:24:13,000 We need to probably set it up in this way. 453 00:24:13,000 --> 00:24:18,000 Write messages is equal to user user input stream mode values okay, stream mode values. 454 00:24:18,000 --> 00:24:21,000 And then we are going to probably get our response. 455 00:24:21,000 --> 00:24:25,000 So I will go ahead and write for event in events. 456 00:24:26,000 --> 00:24:41,000 Let's display the event of messages minus one dot pretty print. 457 00:24:41,000 --> 00:24:43,000 We are just trying to print it in a better way. 458 00:24:43,000 --> 00:24:49,000 So whenever I give this particular message you can see uh, by default it is probably going ahead with 459 00:24:49,000 --> 00:24:50,000 Wikipedia call. 460 00:24:50,000 --> 00:24:56,000 Okay, this should not happen because it is just connecting my name with some famous personality name. 461 00:24:56,000 --> 00:24:58,000 So it is probably giving you some information, right? 462 00:24:58,000 --> 00:24:59,000 Sangeetha. 463 00:24:59,000 --> 00:24:59,000 Krish. 464 00:24:59,000 --> 00:25:00,000 It has selected. 465 00:25:00,000 --> 00:25:01,000 No worries. 466 00:25:01,000 --> 00:25:02,000 I will just go ahead and write. 467 00:25:02,000 --> 00:25:02,000 Hey. 468 00:25:03,000 --> 00:25:03,000 Hi there. 469 00:25:03,000 --> 00:25:05,000 My name is John. 470 00:25:05,000 --> 00:25:07,000 Let's execute this now. 471 00:25:07,000 --> 00:25:08,000 Okay. 472 00:25:08,000 --> 00:25:10,000 So query John Wikipedia. 473 00:25:10,000 --> 00:25:12,000 See it is going to probably Wikipedia. 474 00:25:12,000 --> 00:25:15,000 Uh, so no good Wikipedia. 475 00:25:15,000 --> 00:25:17,000 Uh search result was found. 476 00:25:17,000 --> 00:25:18,000 So it is saying hey. 477 00:25:18,000 --> 00:25:18,000 Hello, John. 478 00:25:18,000 --> 00:25:19,000 It's nice to meet you. 479 00:25:19,000 --> 00:25:23,000 Now see it is probably going and checking the tool message. 480 00:25:23,000 --> 00:25:28,000 Also whether I'm going to find this specific message or this particular name over there. 481 00:25:28,000 --> 00:25:31,000 So it did not get a good result, then it is probably interacting with us. 482 00:25:31,000 --> 00:25:31,000 Hey. 483 00:25:31,000 --> 00:25:33,000 Saying that hello John, it's nice to meet you. 484 00:25:33,000 --> 00:25:36,000 So guys, I hope you like this particular video. 485 00:25:36,000 --> 00:25:40,000 Uh, this was about building chatbot applications with third party tools. 486 00:25:40,000 --> 00:25:42,000 Uh, with the help of line graph. 487 00:25:42,000 --> 00:25:44,000 Uh, just go ahead and try it out. 488 00:25:44,000 --> 00:25:50,000 Try it out with multiple tools and see, uh, whether it goes to RC or not, just keep on adding and 489 00:25:50,000 --> 00:25:51,000 keep on asking queries. 490 00:25:51,000 --> 00:25:54,000 You should be able to get all the solutions over there, right? 491 00:25:54,000 --> 00:25:55,000 So yes, this was it from my side. 492 00:25:55,000 --> 00:25:56,000 I will see you all in the next video. 493 00:25:56,000 --> 00:25:57,000 Thank you. 494 00:25:57,000 --> 00:25:57,000 Take care.