AutoGPT & Langchain in Node.js

--- type: transcript date: 2023-04-28 youtube_id: lwx6ShL-V38 --- # Transcript: AutoGPT & Langchain in Node.js in this video I'm going to show you how to set up Auto GPT with langchain in node.js I'm going to cover everything that you need such as getting API Keys making sure the version of node that you have is up to date demonstrating how to install and configure your project will create an auto GPT instance using open Ai and hopefully by the end of it you have a solid understanding on how you could leverage Auto GPT in Lang chain using node.js so the first thing that I'm going to have you do is obtain a couple API keys so for the openai API key one thing to note with it you will have some credits that you can leverage without tying in a credit card so if you just head over to this link here you'll be able to sign up for an account and get an API key if you just go to the right hand corner where your initial is once you've signed up you'll be able to click that view API keys and generate a key similarly go to serp API this will be the service that we use to leverage Google in the application in future videos I'll cover other ways on how you can navigate the web and make queries similar to serp but for this one we'll be using serp as it also gives you some free credits that you can use off the bat in your application to get going so the first thing that we're going to do is we're going to npm init Dash y now once we have that I've already run this you'll have the package.json set up and then once we're in the package Json just make sure you make a type module within the package since we're going to be using Imports in this tutorial then once we have that set up go ahead and touch an index.js and a DOT EnV similarly I've already run this so I'm not going to run through this right now but as soon as you have the index.js set up just open them up and in the dot EnV you'll put in your API Keys as you see here so once you have those set up we're just going to ensure that our node version is up to date so I encourage you just to make sure that you have a recent version to make sure that all the packages are up to date and will work I'm using node version 18. so as long as you don't have something like node version 12 or 14 even 16 I'd encourage you anything over 18 and above you should be safe so once we have that we're going to npm install a couple things we're going to npm install Lang chain off the bat and then I'm going to revisit this hnsw lib later in the project so while we have that uh running through I'm going to go ahead and just run through some of the Imports that we have here so the majority of what we have that we're leveraging here is from open AI so we're going to leverage the experimental Auto GPT feature so it is experimental right now we're going to import a few tools for our file operations we're going to leverage the node file store for storing files within node.js we're going to use this library to store our Vector stores so if you're not familiar with Vector stores it's how you can create the relatedness between two words so it's what's being used under the hood to create an efficient way of leveraging something like Auto GPT then once we have that we're going to be using open AI as our actual interface llm that we're going to be interacting with and then finally from there we're going to require an import rather dot EnV so actually just make sure to install dot EnV as well so once you have those set up we're just going to head down here and configure our DOT EnV so what this allows us to do is reach into our DOT EnV files once they're all saved so another thing to note just make sure that once you've pasted in your API keys in the dot EMV just make sure that the file is saved so once we have that we're going to instantiate a node file store from there we're going to go ahead and create a couple tools for file operations and set up our serp API so in this one I'm going to specify that we're going to be using Google there's also some other things that you can leverage and pass in here such as passing in your location if say you wanted to use something like getting weather data you could do that based on your location there then next we're going to set up the hnsw lib with our openai embeddings so we're going to store our vectors locally so what this Library allows us to do is create a local version of our Vector store rather than having to use a database for it and then the open AI embeddings API is how we create the relatedness within the vector so a vectors essentially words represented in numbers and the closer that those numbers are is the closer in the quote relatedness of those numbers so once we have that set up we're going to go ahead and create an auto GPT instance we're going to specify the model in this one we're going to be using open AI like I mentioned we're going to use our Vector store for memory and then I'm just going to specify a name and a role for the auto GPT instance that we're going to set up so I kept it pretty broad I just called it developers digest assistant and the AI role is simply assistant but you can be more specific if you want in in in that uh in those fields there so next I just came up with an example I said create an index.html file with a working stock chart of Apple stock price using charge.js with Dynamic stock data oh I see with data with Data stock data from Google Finance so I might just pair this back a little bit I'll just say instead of dynamic let's just see if it will get stock data and then from there if we go ahead and type in node index.js and run this we should get this command where it specifies us to install the hnsw lib Dash node so I wanted to show this because often when you do install link chain I have noticed that a missing piece is actually remembering to require this Library so it's not necessarily going to be something that you use if say you're using an external Vector database but just so you're not freaked out when you see this command that's just something that you can take from the error and simply paste it in if you actually copy it you can just paste it into your terminal here and I'll just delete this for now since we had a bit of a false start in our example and then I'm going to go ahead and run this again so right off the bat you see this K4 is greater than the number of elements in the index setting K to zero so once you see this message you'll know that the auto GPT instance is running as it should and you'll start to see the output like you might have seen some examples on how Auto GPT runs so you'll start to have these outputs like thoughts okay I need to create an index HTML with a working stock chart of Apple stock price using charge JS etc etc it will start to give you the reasoning it's plan it's criticism so it's pretty interesting just to put in an input and watch this I'd encourage you to watch this closely at first at least actually even just in general in the early phases of this integration just to make sure that your API key doesn't run off the rails here and you end up in some sort of recursive Loop or just running for a very long time or much longer than expect it so yeah just keep an eye on it but it is pretty interesting if you haven't used it before just to see how it works so you see okay I need to find a tutorial or documentation on chart JS and how to get stock data so okay it has the thought I need to evaluate the source of the tutorial I found and ensure it's reliable okay I need to create an index.html file and add the necessary code so you see it it works pretty quickly and now the interesting part is I see that I have a write file command so it's actually written a file within our directory here we have an index HTML and we'll see I need to test the chart and ensure it's working properly so I'll just let it go through and see if it finishes here in a reasonable amount of time and then we'll open it up and see what it looks like but this is just to give you an idea so it's less on okay is this going to work and give me a perfect working stock chart that's interactive and dynamic and whatnot but more just to give you an idea on how you could use this so with a more specific prompt you could likely have a very impressive result but the thing to note with this is you can use this to write to files like code and it doesn't necessarily need to be a txt file of say a summary or a brief like you might have seen in some examples and you can actually make files such as the code like I have in this example example rather so um actually instead of awaiting here I'm just going to go ahead and open it to the side here or rather open it with let's just open it with live server and see what it's given us okay so we see here that we have a stock chart now whether this is accurate or not I'll sort of leave it up to you to double check so today's date is April 28 2023 and you can let me know if this data is accurate and if it's not you can go through the the tasks and see okay why wasn't it accurate what sort of prompt might you be able to input in the future to make it more accurate and yeah so hopefully you found this useful if you did please like comment subscribe I'll have plenty more videos going more into depth into the Lang chain into Auto GPT and the whole AI ecosystem and how it relates to software engineers and web developers and without further Ado until the next one