AutoGPT World Record 🏆How I Setup Flowise on Replit with Pinecone Vector Database (NoCode Langchain)

are you ready to witness a world record in the making today we're not talking about just Ai and automation we're setting a new standard welcome to a groundbreaking tutorial where we'll break the world record for setting up an auto GPT instance from scratch in this video we're gonna dive into the world of Auto GPT a revolutionary AI agent that's transforming industries by autonomously achieving set goals we're gonna set up an instance of Auto GPT faster than anyone has ever done before right here on Redland a collaborative browser-based IDE that's revolutionizing how how we at langlabs develop apps but we're not stopping there we're going to be using flowwise an extraordinary open source project that allows you to build customized llm chat flows with ease this powerful tool will allow us to get our Auto GPT instance up in record time next we'll introduce you to Pine Cone the number one vector database that's making high performance llm applications a breeze to run even across billions of records and the best part you don't need to be a coding expert to do any of this with the power of all of these different no code tools you'll be able to create powerful software learn how to code or help others learn to code without even knowing how to write a single line of code so if you're ready to witness history in the making stay tuned and before we dive in if you're interested in Ai and automation development services be sure to check out our website langlabs.io we have a team of AI and automation experts that are ready to help you on your next project so without further ado let's begin all right guys I will be uh removing my splint for this one Daddy's better now I will be able to move freely come on you really think you'll be able to get through gara's defenses just by dropping a couple of pounds of weight so we're gonna start off on replit.com and that'll be the uh the starting point for this challenge I got the timer up I'm going to be creating a new replit and getting Auto GPT up and running and it'll be done when I send a message and there's no no errors and whenever I get the auto GPT response back so um for the challenge I already got my API Keys ready um that's not cheating that's just being prepared for a world record so without further ado oh let's break the world record start create Rebel node JavaScript we'll make it pro public Auto GPT world record let's go fading reply hitting the code the code that will be the first code will run is the installing flowwise script so we got the shell command open now we're installing fiddling my thumbs so yeah it's my thumb that's been fractured uh I wear this blunt for my thumb because it's been healing up kind of in the wrong direction and hopefully it's not too light for it to heal up properly but I'm going to make a sacrifice for this challenge so while the clock is ticking there's not much I can do just yet besides getting the start command ready so we'll be getting that ready to go so uh yeah how's the weather today got any cool uh weekend plans me personally plan on uh still have to hit my Murph it's a Memorial Day weekend gotta pay homage to a lieutenant Murphy and uh yeah so I guess I'll also use this opportunity to share that we are going to be launching our uh newsletter called The Lang letter this will provide Insight on artificial intelligence news tools with an emphasis on actionable and useful insight for businesses small businesses um like yourselves or aspiring entrepreneurs if you're not on that path already um still loading so uh Let's uh maybe another pitch let's talk about uh we provide at langlabs Ai and automation development services we will our our high our done for you offering is to completely take the development out of your hands and allow you to continue doing the work that you love to do and we're back in business I'm gonna start up the uh flow wise now it's spooling up something's going on let's see if it's loading up all right turning on night mode dark mode clicking Auto GPT looks good to me again some API Keys ready to go so boom API key for chat GPT cat GPT and then let's go uh yeah that's all we need serp API API key copied and pasted Pinecone memory index uh name I have one Auto GPT API key paste it in their environment key paste it in there I gotta get a path so this will be going on to replit so I'll just call it agent copy that copy that give them a name world record is uh let's get rid of the space just in case there's an error there and he's uh assistant he orgy turn the loop to three save the flow world record saving and then don't save the password let's see if we're up and running uh tell me about okay create a plan to learn LMX in send error Pinecone error so I have an error in the uh there we go paste it in the wrong key but we're back in business copy and paste my last prompt enter error saving troubleshooting clicking clearing chat history oh this thing got in the way clear it click thingy clear it clear it clear the history fine go loading making sure that this time we're good not sure should be good if it's good we'll see this uh console print out some internal talks hasn't populated yet I'm going to do a quick review maybe I missed something did something wrong everything looks like it should be all right check fee oh yes perfect you can see we got the standard um Auto GPT thought process reasoning process plan process criticism and speak so for my Loop it'll have three um it'll do a Max of three Loops before providing me a response so we'll stop the clock as soon as I get that response checking the tab this thing is getting in the way so I see [Music] thank you foreign [Music] so I'm not sure why the camera cut out and the audio stopped recording I got a really uh pretty terrible webcam so uh if uh there's some contention for this world record let me know and I will do my best to uh Rectify this in the future however I didn't cheat that first video uh the first six minute video was one take I let it run the entire time kind of couldn't see where the timer actually teleported to but I was still able to click it so six minutes had nothing from the start all I did was run two lines of code had my API Keys ready to go and we're in business so I know that was a little bit fast and the the actual Auto GPT output was kind of terrible probably because I left it with the GPT 3.5 but for the purposes of speed the agent completed its task it just kind of was talking about like creating the plan and it didn't really actually it talks about creating the plan for creating a plan to learn llm Ops which is one of the services that we provide at langlabs which includes fine-tuning and optimizing your production level LOM applications so for uh pretty much larger scale even Enterprise use we provide that as an offering at our agency so this would reduce the incidence and likelihood of incorrect responses by your llm application hallucination and increase accuracy and the desired output from from the prompts that you use in your application so I know uh I know personally I'm a little disappointed but did it in one shot I set the world record at six minutes if you can beat it let me know tag us and just consider it War no matter what time you set I'm coming for that record baby langlabs is number one the premier Ai and automation development agency no cap as they say so we had one of our community members uh he had asked us in our Lang Labs Discord group how to implement the tree of thought prompt chaining technique on flowwise so this is one of the the many services that we offer our community for free free of charge if you're engaging with us if you're sharing our content and you like what we're doing um we're we're here to support you too so this none of this can be done without your support he asked and uh one last thing is like if you have any input or questions you'd like addressed make sure you you connect with us message us on our socials and join our Discord group and we'll be happy to help that's what we're here to do we love AI we know how important important it is to learn and actually Implement these tools in your regular uh your existing systems and processes so we're here to help by helping you we're only getting better at ourselves so uh back back to it um from our Discord he kind of asked about the tree of thought prompting technique which is where the first prompt is like you have a problem describe your problem can you brainstorm three distinct Solutions please consider a variety of factors step two is prompt uh for each of these three proposed Solutions evaluate their potential consider their pros and cons initial effort um you can you get the gist of it if you join our Discord group you can see the The Prompt and what have you so um I've built this on replit on flow wise rather so this is it right here I kind of just tested it to make sure it works uh had one kind of Hiccup on the first go-around but I got it working now so while I let it do its thing The Prompt that I gave it is my problem is growing a following on Twitter our perfect factors include growing an audience that is looking for Automation and AI development services so starting from that first prompt chain um we have our context is objective and I know from last time I was going a little bit fast from the last video so I'll be saving this prompt uh this chat flow this flowwise chat flow and I'll be hosting it and uploading it to our Lang Labs public Discord group so be sure to join on in if you want to be able to download this for your own usage so it goes our context is objective I have a problem related to describe your problem area could you brainstorm three distinct Solutions um I don't want to click it because I know that the chat bar will might disappear so I'll see if I can just scroll down a bit could you brainstorm three distinct Solutions please consider a variety of factors such as your perfect factors so we're doing a little thing where um in in virtues prompt he wanted it to be in like the square brackets uh just to kind of fit his intent into what flow wise is capable of doing you uh you can only pass like kind of one variable with flowwise if you want to do something a little bit more complicated it'll require kind of like a more custom solution but we were able to get it working in flow wise so you just feed it those square brackets as is and you'll import your prompt as uh has seen above so moving on to the next prompt chain so prompt chain nothing nothing crazy here you have LOM chain is like the basically the end the first node in this uh prompt chain you have the language model which is jet GPT and the prompt template above so I'm getting an error chain llm machine expects multiple inputs cannot use run it's saying that but I know that this is the console print here it's working so I'll have to brainstorm a little bit more and in like the final in the uh the final go around I'll share like the correct version but I'll just walk through the my initial thought process for getting it up and running um so this is this is just like a draft and doing a quick thing to show that we support our community so the next prompt is our context is objective our evaluations are evals from each of the three proposed Solutions developing their potential and then I have it set for objective and eval okay I think that's what it is um I saw that I had the prompt or the curly bracket as eval but it should be or evals and it's supposed to be evalued so that probably fixes that at least a little bit um and then in the third chain it goes objective is objective once our our Solutions we have evaluated our solutions for each solution deepen the thought process generate potential scenarios strategies for implementation you can give a read later on and then I have objective and solutions and everything does look good on that end going to the third and final prompt we have okay another so this one I probably figured it out what was wrong I just named the variables incorrectly so try this again so maybe it'll give us an answer by the time we're done our context is objective objective variable our scenarios we've evaluated our results oh now I see that I will have another hiccup so I might as well just cancel that at the very end our results our results cancel we're going to call this result if you don't have the the variables set up properly it'll just give you errors so I'll have to rerun The Prompt again to get a good answer so same thing my problem is growing if only on Twitter our perfect factors include growing an audience that is looking for automation AI development services and you kind of saw on the fly like a lot of this stuff is trial and error no one's perfect like I whip this thing up in like a minute just because uh that's how quick these things take it doesn't have to be perfect on the first go around what matters is that through Rapid iteration you'll get your the desired outcome faster than just kind of Meandering around with the plan go ahead and execute and that's kind of what we believe at langlabs is just push push push we're going to fail here and there along the way but that doesn't matter as long as we're learning we don't make the same mistake again and we continue to violently execute we will have success so that's what we believe in if that message resonates with you check out the rest of our our content and check out our team so goes to the another llm chain feeding to the chain tool because I wanted this to be a conversational agent with chat uh history so you'll you can only really do that if you have uh the conversational agent online on full Lane chain and flowwise the chain that the L1 chain is connected the chain tool the chain tool says tree of thought use this when the user provides input on their problem area and their perfect factors so that's kind of the go around uh our our way to get around not being able to send two variables to The Prompt that virtus has requested so again here I'll keep troubleshooting some more but that was the gist of it the the conversational agent has that chain as one of the allowed tools so it's the chain tool we're using chat GPT GPT 3.5 and we have chat history so tune in again next time folks we'll go into a little bit more depth make sure you check the Discord group we're going to host the 100 accurate chat flow for the tree of thought that vertuss has recommended or requested and again this is something that interests you we offer a full Suite of services we'll develop for you we'll optimize for you we're a community of AI Lang chain LOL application experts so make sure you connect with us and I look forward to seeing you next time

Channel: LangLabs 🔥

Views: 2,015

Keywords: ai finance, auto gpt, auto gpt install, autogpt, chatgppt, flowise, langchain, langflow, pinecone, pinecone vector database, replit, vector database, vector database openai, vector database tutorial

Id: CprzRA-emL4

Length: 21min 46sec (1306 seconds)

Published: Mon May 29 2023