GPT Researcher: Deploy POWERFUL Autonomous AI Agents

hey what is up guys welcome back to another YouTube video at the world of AI in today's video we're going to be focusing on a new amazing application which is called GPT researcher this has been developed by tablet and it's an advanced autonomous agent specifically designed to conduct comprehensive online research across a wide range of tasks now its primary function is to generate detailed factual and unbiased research reports now additionally to this gbt researcher offers customization options that allow you as a user to focus on relevant resources outlines lessons as well as enhancing the efficiency and effectiveness of the research process now drawing inspiration from Auto GPT as well as the plan and solve paper what GPT researcher has addressed is the two critical challenges in AI research which is speed and determinism Now by employing a parallelized agent work approach you're able to see that instead of focusing on a singer awareness approach we're able to see that gbt researcher ensures that there is more of a stable performance as well as an increased speed now this design Choice allows you as a user to obtain research results way more quickly and reliably which optimizes the productivity as well as saving a lot of valuable time now this is something that we're going to be checking out in today's video we're going to also take a look at some of the features as well as taking a look a little bit more about the architecture of this application we're going to also dive deep as to showing you how you can actually install it locally on your desktop and how you can actually access it on your Local Host so make sure you stay tuned and before we actually get into the gist of the video it would mean the whole world to me guys if you guys can go subscribe to World of AI if you guys haven't already I post a lot of content and a lot of value every single day so that you can benefit from it and that you can learn a lot from it so I highly recommend that you check it out if you guys haven't already and there's a lot of content that is has been posted already and it's a lot of value so you should definitely check it out if you guys haven't now if you guys haven't followed the Twitter page definitely do so as I'm going to be posting the latest AI content over here in terms of the AI news so with that thought let's get right into the video the mission of GPT researcher and its creators at halvey is to empower individuals and organizations by providing them accurate unbiased and factual information this Mission reflects a commitment to Leverage The Power of AI to combat misinformation as well as ensuring that users can access reliable and trustworthy knowledge now with GPT researcher it possesses a remarkable ability to devolve into the vast amount of online data enabling it to analyze diverse sources and extract valuable insights and knowledge now one of the most notable and aspects as well as like the most like intriguing aspects of GPT researcher is its customizable nature we're able to see in the demo at the start of the video that you're able to actually have the flexibility to tailor your own research requests according to your own specific need and preferences you can focus on relevant resources that aligns your own research topic and Ensure that you're able to gather the right information directly applicable and useful to what you give in terms of its prompt and this is something that we'll showcase later on in the video now I wanted to talk a little bit more about the architecture as we see over here let me zoom in a little bit so you can get a better idea now the architecture of GPC researcher involves the utilization of a planner and execution agent to facilitate the research process this is something we saw with auto gbt as well as with baby AGI now these agents work in a tandem to generate research questions to find relevant information and to ultimately produce a comprehensive research report now the architecture takes advantage of both gbt 3.5 turbo 16k parameter as well as the gpt4 models to accomplish the research task effectively now the planner agent plays a crucial role in the architecture as it's responsible for generating research questions based on the given desk as well as the user your input now these questions serve as a prompt for the sequential research process now the planner agent leverages the capabilities of the underlying language models to formulate high quality and informative questions that guide the research process we can see over here in this architecture you're given a task but the task is then given to the research question generator which focuses on different queries so this execution agent are then deployed to execute these research questions which are generated by the planner and this is where these agents employ a variety of techniques to locate the most relevant information based off of each question as well as each research question that is deployed from a task they explore online resources databases as well as other repositories to gather this information and the extract valuable insights to get you the best type of formulated answer now the execution agents are basically designed to efficiently navigate vast number ounce of data that is available and they retrieve the most like I would say the best detailed related answer to your research query so with the planner agent we're able to then see that it takes on the role as a filtering as well as like uh aggregated like way to filter out any bad information that is collected by the execution agent it then shifts towards retrieving the data and identifying the most relevant as well as the most valuable pieces of information so that it can create a cohesive research report and what's amazing is that it's able to collect so much data in like really quickly and you're able to get such a formulated detailed PDF that's up to like even 20 pages and this is something that is absolutely astonishing as to how fast you can get a research paper created within a couple of minutes uh now to accomplish these tests I want to talk a little bit more about the ending because I keep getting sidetracked talking a little bit more about the overall scheme of this project but to accomplish the whole execution what GPC research the researcher then leverages is the power of GPT 3.5 which we talked about as well as gpt4 models and what these models are they're basically Advanced language models that have been developed by openai and what they've done is that they use utilize these models to possess a deep understanding of natural language so that it can generate coherent as well as contextual relevant responses from searching online and this is by utilizing these models to enhance the quality and the accuracy of the research process so it starts off with the root task that is given from a user the user is then able to deploy This research question which is selected through the preferences that you can set with the application once you're able to generate the research question it is then set out to different types of queries and there's infinite queries that you can select towards and then what these queries do is that scraps through different types of content on the internet and then this is done by the execution agent and this is what it does is that it gathers all the information in different query Bubbles and from these bubbles it is then able to summarize it through visited sites as well as visited PDFs content files and it clogs it up into different Bubbles and now what the actual planner region does is that it then scraps through each and every different bubble that has gathered information and it then gets the best cohesive and relevant knowledge as well as the contextual relevant responses that are best guided for the actual research question and it formulates a response based off of whatever bubble was that was executed through the execution agent now you might be asking why might you use GPT researcher when you can just basically use any different type of application that can generate such large contextual reports well one of the main things is that most of these different LMS as well as other trained like application don't really focus on new data with gbt researcher it focuses on relevant like contextual reports as well as relevant content that is best suited for your research we can see current LMS are trained on past and outdated information whereas GPT researcher focuses on relevant information now another thing is that to form objective conclusions from manual research tests it can take time and we can see that when you're able to generate from you can see from this demo too you're able to generate reports in like a couple seconds be able to get like 20 Page reports in a couple seconds and I have not seen any other application that is actually able to do this another thing is that solutions that enable web search such as chat GPC plus web plugins only considered limited limited resources that in some cases result in superficial conclusions or biased answers another thing is that users can select the users using only a selection array resources can create biases in determining the right conclusions for research questions or tasks so these are some of the reasons why you would actually want to use GBC researcher and one of the main things I personally would want to use this is because of the speed as well as how it's actually able to focus on relevant information you're able to see in terms of its feature it generates its researches research papers outlines resources and lesson reports you're able to get 20 web sources per research to form objective and factual conclusions oops my bad you're able to get even like pages that are up to even 20 Pages you're able to include an easy use web interface it scraps web sources as well as JavaScript supports it's able to keep track and context of visited and used web sources so you're able to actually get a citation of where the information is actually able to get like get come from you're also able to export the research reports to PDFs as well as other different types of files now with that thought that was actually getting into the next step where we actually start installing this application onto our desktop so first things first you're gonna have to have python 3.11 or later installed so if you guys haven't had python python installed definitely do so you also need to get Visa application that you will need to help clone the repository onto your desktop now the application is Visual Studio code we might use this to edit the code or to actually deploy the actual application so if you guys haven't had these three applications that if they do so and with that thought let's get right into installing it so first things first you want to go on to the repository I'm going to leave all the links in the description below so do not worry click on this green button copy this link open up command prompt once you have that open type in git clone paste the link and press enter oh shoot I pressed the wrong thing press enter and it will start cloning the repository onto your desktop now what you want to do is get into the actual folder by typing cdgpt researcher once you're in the folder you need to start installing the files and that is by copying this link right here which is PIP install Dash R requirements so what you want to do is copy and paste in command prompt click enter and it will then start installing the actual files that are needed for this application so once this is done I'll be right back you will also need an open API key so make sure you have this installed or sorry not installed but make sure you formulate your key on this link I'll leave it in the description below if you guys haven't already another thing that you want to know is that you need to make sure that you have a billing set up so that it's able to utilize open apis 2.5 model or gpt4 model now once you have that figured out you need to then go back onto the repo and you want to copy this command over here this is for Mac but in my case I have a windows so I'm going to copy this link right here but first i'm going to paste my key and once I have that copied out paste it right here I kind of messed up because I need to be in the right folder but once I'm in the right folder I'll copy and paste it again and I'll be right back so guys for some reason I do not know how to or what the command is for Windows to actually analyze uh my open API key with this application so what I just did is I just uh what do you call it I just opened up the file our I launched the agent by copying pasting this prompt right here so it will now start installing it and it'll start loading this application once it's ready so once it's done you can then now access it on your local host after pasting it so what I'm going to do for the open API key is that I believe there's going to be a settings tab on the application in which you can paste your open API key and that way you'll be able to input your key so that you can utilize and access this application so once the Local Host is ready I'll be right back unfortunately guys I cannot load this project because of this stupid W easy print it's not able to import some of the external libraries I tried playing around with it tried tweaking it tried installing it too I even uninstalled it and still kind of figured it out but that's okay guys uh I'm just gonna show you a live demo on Twitter this is just an example I just wanted to Showcase you how you can actually generate research papers uh at least we got a chance to look at some of the features as well as architecture but once you have it loaded up on their Local Host it will then look like this you are then put into this file looking thing over here in which you can ask what would you like me to research so in this case you can ask it should I invest in a semiconductor industry in 2023 obviously for this question you would want to select an agent that is related to finance you wouldn't want to select an academic research engine it in this case you would want to select the finance agent so then you can ask you what type of report would you like me to generate you can re uh there's different types of reports that you can generate in this case you would want to have a research report and then what it does is that there's an agent output it starts looking towards different types of tasks we can see that it's adding different sources from different websites to formulate your research report and at the end it starts summarizing the sources and as an output you can get this amazing research paper at the end that has a formulated uh response as to what you should do investment opportunities as well as a decision on what you should do and we can see it also puts it into a PDF you can download it as a PDF as well as copying it onto your clipboard and at the end you have a lot of different sources and references in which you can check back on and this is amazing as you're able to get such an amazing formulated answer because of this amazing application so in conclusion this architecture of GPT researcher employs yourself with the planner and execution agent to generate research questions as well as locating relevant information and producing comprehensive research reports it's a great way for you to actually formulate a lot of different research papers or planners that require a lot of time so I highly recommend that you check this out guys because it's very easy to access and it's very easy to install except for me I can't seem to install Easy print but I hope you enjoyed this video guys thank thank you so much for watching means the world to me make sure you subscribe turn on notification Bell if you guys haven't seen in my previous videos I highly recommend that you do so follow me on Twitter and with that thought guys have an amazing day spread positivity and I'll see you guys very shortly peace out fellas

Channel: WorldofAI

Views: 5,056

Keywords: ai agents, autonomous ai, gpt 4, GPT Researcher, Tavily, AI research, online research, autonomous agent, research reports, speed, determinism, auto gpt tutorial, auto gpt install, auto gpt, auto gpt ai agents, auto ai agent, superagi, babyagi, super agi, chatgpt, openai, pinecone, google search, ai, agi, autogpt, autogpt install, super agi install, superagi install, superagi github, autonomous agents, ai execution, intelligent agents, autonomous decision-making, auto ai gpt4

Id: ytE64RV9CLo

Length: 17min 0sec (1020 seconds)

Published: Thu Jul 13 2023