FREE Agent-Q Controls Your Browser To Do ANYTHING!🤖 Complete Open Source AI Automation Agent MultiOn

What's going on everybody Josh pokok here and in today's video we are going to be taking a look at agent Q Advanced reasoning and learning for autonomous AI agents and this AI agent can literally do anything it can control your computer it can do anything within your browser from searching things up getting different information we've covered different tools similar to this such as open interpreter before on this channel I'm going to show you how you can set this up easily without

Writing a line of code it's 100% open source let's dive right into it [Music] all right guys so here is the link to the GitHub repo for agent Q all the links that I cover will be in the description down below now the creators of agent Q is this company called multi-on and you'll see the blog post right here introducing agent Q research breakthrough for the next generation of AI agents with planning and self-healing capabilities so in recent years the capabilities

Of L L Ms have transformed natural language processing and understanding achieving remarkable Milestones despite these advancements llms have significant challenges in interactive environments especially in task requiring multi-step reasoning like web navigation current training methodologies relying on static language data sets fall short in equipping these models for dynamic real world interactions enter agent q a major milestone for agents combining search self-critique and reinforcement learning to state-of-the-art autonomous web agents that can plan and self-heal our breakthrough method addresses the limitations of previous

Llm techniques by introducing a novel framework for Learning and reasoning for autonomous web navigation all right so agent Q innovates basically on This research paper that I'll uh briefly touch on in just a second by combining guided Monti Carlo tree search MCTS and AI self-critique with iterative fine-tuning leveraging reinforcement learning for human feedback rlf methods like Direct preference optimization DPO algorithm this method enables llm agents to learn from both successful and unsuccessful trajectories enhancing the generalization capabilities and multi-step reasoning tasks

So key components of agent Q guided search with MCTS Ai self-critique and DPO direct preference optimization now real world validation so in real world experiments on Open Table multi-on agents drastically improved the zero shot performance of llama 3 model from 18.6% success rate to 81.7% a 340% jump after just one day of autonomous data collection and further to 95.4% with online search these results highlight our methods efficiency and ability for autonomous web agent Improvement and then it links to the full

Paper here all right so this is the actual research paper that you can go ahead and check out I'll leave it in the description down below agent Q Advanced reasoning and learning for autonomous AI agents it goes in depth um based on you know this their research and their new framework and I actually created a stepbystep guide for agent Q so you can get up and running I have all the different resources you could possibly want or agent Q the Chrome

Extension the blog post the repo the research paper the docks the API reference sdks and all that good stuff outlined in this dog I'm going to leave it in the description for you guys for free so you can check this out follow along with this video if you get stuck you can reference this Doc it's going to have all the different commands to run to get this set up and pretty much anything you could possibly want when it comes to agent

Q getting it started okay all right and I did Briefly summarize the uh agent research agent Q research paper right here so I'm not going to go over it but it's pretty much a lot of the stuff we covered in that blog post there is a little bit more um that they talk about of course in the research paper but uh there is a summary right here so you can check that out if you want to dive deeper into that now

To get started with agent Q there's really two main ways you can use this now the first way is through the command line and I'm going to show you how you can set that up in just a second the second way is through the Chrome extension so you can use it either way or both ways the Chrome extension is really cool too especially if you don't want to do any sort of commanding or whatever the case may be all right so

This is the link to the Chrome extension you can go here download the Chrome extension for free and then you're just going to have to create create a multi-on account for free you can come up here to the Chrome extension you'll see there's different options API enabled multi-on home screen show assistant there's settings and configure here now once you're at the actual multi on home screen here you can write a command so there's really two ways to do this you can

Go through Auto or step by step um and they give you some different options here like book a meeting order me a burger on door Dash whatever the case may be so if I just do search Josh pook on YouTube and then click do it you can see we get a little chat right here a little chat gooey pop up and it's already at YouTube it's just loading right now and boom we just searched up Josh pook on YouTube if you're

Not subscribed make sure to subscribe and it actually clicked on my YouTube channel already so you can see here in the chat we're literally talking to it and it said I'm navigating to YouTube to search for Josh pook I clicked on the channel link and now the search has been completed you can give it a rating you can actually have uh different prompts that suggest like subscribe to the channel you can even click voice right here so let me just move

Let me just move my head a bit but you can see here there's voice there's next step like there's so much you can do start over send play really really cool Chrome extension and you probably noticed I haven't even entered like an API key yet so it's not even using your um open AI API key or your claw um it's literally just using the free Chrome extension all right you could also click personalize here um turn on voice like this is

The future when it comes to browsing the web you're never going to be in the future in like a year or two years from now you're never going to be going manually clicking clicking on different things I mean you may do that a bit but most of the time you're going to be using your AI assistant your AI agent so I think this is a huge stepping stone in the right direction and like I mentioned we've covered other tools like open

Interpreter there's other ones too that I will continue covering on this channel soay stay tuned for that and this one's really cool though and I'm definitely going to dive into it a lot deeper because they even have an API reference right here where you can have autonomous API reference a browse retrieve create session step session close session screenshot list sessions and then their documentation and then they have sdks for node.js python so if you're a developer and you want to add

Something like this into your app you can definitely explore and build some cool things with this now let's go ahead and get it set up in our actual um command terminal so you can go ahead and there it does have the instructions right here in the GitHub but now there are a few different little things that is missing um and I'll just fill in the blanks for you I actually have it all outlined in this document right here so the prerequisites

Of course that you're going to need before we even get into the installation process is git Python and poetry all right so you can go to poetry here and you can literally just run pip X install poetry if you don't already have poetry installed and then you're going to need an open AI API key or I'm going to also show you how you can use you know together AI Gro grock um you could use olama you can really use any single

Model with this which is really really powerful so you can use closed Source models or open source models so first things first you're going to get clone the repo here so you're going to open up your command terminal I would recommend running it as an administrator and then you're going to get clone the repo into your desire location I'm not going to do that cuz I already did it all right once you do that you're going to CD into agent Q

So you're going to run this command to change directory into agent Q so once you CD into agent Q then you're going to run pip X install poetry so that's actually if you don't if you didn't already install poetry like I said so after you install poetry or if you already have poetry installed you're going to run poetry install so this going to install all the dependencies are you tired of pouring thousands of dollars into appointment set only to watch leads

Slip away imagine having a team of elite sales agents booking qualified appointments for you around the clock no more wasted time on training no more frustration with performance and no more draining your budget on inconsistent and expensive call centers introducing stride agents AI powered appointment CS at work 24/7 never get tired and book appointments while you sleep trained on thousands of successful conversations our AI agents outperform human teams at just one tenth of the cost join the ranks of businesses that

Doubled their appointments and booking rates in just a matter of weeks don't get left behind in the AI Revolution visit stride agents.com now and transform your entire sales process with Cutting Edge AI technology it's time to accelerate your stride with AI agents okay so depending on if you're on Mac or Windows um you're going to want to set your open aai environment variable so um if you don't have this set already all right if you've been doing different things with uh

Tools like then you probably already have your API key set but if you don't for mac and Linux you're going to run export openi API key and then equals your open AI API key and then for Windows it's set X open AI key space and then your open AI API key and this is going to be the same thing for if you're using Gro or together AI which I'll show you in just a second you're going to make sure you're wanting

To uh set that environment variable in your operating system okay once you have all that set up you're going to start Chrome in Dev mode okay so there's a couple ways to do this for Mac you're going to run this Command right here I'm not on a Mac so um yeah and then for w Linux you're going to run this command and then for Windows they say to run this command for me that one didn't work I actually had to add

An and in front of it okay so I ran this command well to be actually 100% accurate I ran this command either one should work this one just basically uh automatically sets a user profile for you and what this is basically doing is starting Chrome in developer mode on Port 9,222 now one thing I will say is if you have for me personally I had uh Adobe Premiere open up which actually uses that Port too um so you you if you

Are running to any issues it could be some external application that you have opened and you may just have to close that and rerun it okay once you have Chrome open up it's going to look something like this just like normal Chrome and then you're going to want to run pyth on-- mpace agent Q okay so just literally copy and paste this and once you copy and paste it it is going to boot up agent Q in the command terminal you'll

See current state agent current thought plan completed task and then you can enter in your task and we'll do that right now all right so my command is search up Josh pook on YouTube and click on the second most recent video okay so it's actually doing this in the new tab that we opened up so it's searching YouTube right here okay so we can see completed task YouTube okay so we can literally see typing in Josh pook into the search and

We can see that completed task right here next is identifying the list of video results for Josh pook uh on the YouTube search results focusing on the video timestamps so I probably would have just clicked on my channel when two videos and then looked at the second most recent one but I mean we'll see how this works identify the second second most recent video on Josh pook for the search results based on the provided video timestamps now keep in mind that

This is this could technically be right I should have been more specific in the prompt and this is something that I I just want to reiterate and um hammer down is your prompts are key be specific in your prompts okay so we got this actually yes this is the second most recent video so it did get it correct it did analyze the timestamps and that is pretty damn cool and this is using the open AI model now I'm going to show

You as you can see the final result has been done and you can literally get creative with this um and you know go wild now I'm going to show you how you can do this with any single model all right so first things first you're going to want to open up the repo into something like uh cursor or vs code which I have right here then you're going to want to go to the folder these are the this is the the

What you're going to want to go to base. piy and this is in the directory of agent q and then core and then agent so if you see here I'm in agent q and then in core and then in agent and then you'll see base. Pi now you'll see here I have this in the guide step by step if client is open AI it will use open Ai and if it's together or whatever you want to use then it will be

Use the base URL and API key that you set uh in the code and in the OS environment so you'll see here in line 26 in the code it says CL client and then open AI so if you change this to together for example then it will use the together base URL here as well as the together API key right here so you're going to need to set this in your environment variables now you can do this for Gro together ol

CLA you get the point you're just going to need to change that base URL to whatever you're using so if I go down to line I believe 50 52 and 53 you'll see base URL so if I wanted to use Gro I would just change this to the word Gro instead of together and I would go up here and change this from open AI to Gro and then it's going to use that second one right so it has this if client

Equals open AI it's going to use the regular open Ai and then if it equals whatever you write here whether that's olama together whatever Gro then it's going to use whatever base URL you set here and then the environment variable token uh or the environment variable name for the I key that you set here okay now I also left some additional videos to open interpreter I did two videos on open interpreter and I'm going to do more videos on tools like

This um if you want me to do more videos on agent Q or any specific agent that can control your computer do certain things let me know in the comments down below or if you want me to do a video on any other AI coding tool or whatever the case may be let me know in the comments Down Below guys hope you got some value from this video I just want to cover this AI agent tool like I said I think

This is the future I think this is huge and I think this is very very good like when you use it it's pretty damn accurate there's going to be bugs just like in any single tool but like I said this is the future that this is a trajectory we're going with AI searching AI browsing let me know what your thoughts are in the comments down below about this tool and where we're going in all this type of innovation that we're seeing

So rapid in AI all right other than that guys um hope you got some value from this video if you did make sure to smash that like button and subscribe to stay up to date with the daily uploads we upload videos every single day on AI automation business growth bunch of different stuff here and if you're new to the channel and you haven't already joined our free Facebook group and free Discord Channel links to that will be in the description down

Below stri community.com also too guys thank you so much for the recent 7,000 subscribers road to 10K on the way appreciate your guys' support I'm extremely grateful and if you guys do run a business and you need appointments booked for your business on autopilot from AI agents 247 that never sleep and are trained on thousands of successful conversational Pathways then visit stride agent.com link to that will be in the description down below and if it does seem like something that could

Help your business just book a call with myself and my team and we'll see if it's a good fit other than that guys I will see you in tomorrow's video keep hustling keep grinding and of course guys accelerate your stride take care