FREE WebUI Beats OpenAI Deep Research & Operator!🤖 (ANY LLM) Open Source Browser Use AI Agent

All right guys so in yesterday's video I mentioned that I actually just purchased chat GPT Pro and now I'm spending $200 a month for operator and deep research and in that video I actually covered an opsource deep researching tool that you can run in the command line 100% for free of course you use the API and I mentioned that it's a good alternative because not everyone can really afford nor do some people even see the value in paying $200 a month

For open AI tools and a few days before that we actually made a video on browser use which is an open source alternative to open ai's operator and if you don't know about operator and browser use these are two of the main tools that you actually get in the Pro $200 a month plan deep research does PHD level deep research on any single topic and then open ai's operator is essentially a tool that can control browsers and AI agent do specific

Tasks now both of these of course are close source and they also in some extent are limited for example with deep research at the moment you only get 100 deep research searches a month it's very labor intensive and then open ai's operator is limited in some regard to visiting certain sites or doing certain actions so with the open source browser use and the open- source deep research that I covered both of these tools were really great Alternatives but in both of

These videos with both these tools you had to use the command line and there was no beautiful web interface gooey well in today's video I'm going to show you an open-source solution that literally cures both of those problems for open ai's operator and deep research this is a web guey that is made by the same creators as a browser use and it even has deep research capabilities you're going to be able to self-host this solution locally in the next 5 minutes

And use it with any llm of your choice if you don't want to spend $200 a month and you want to open source solution hosted on your computer ASAP then stick around and let's dive right into it all right guys so for those of you that saw my browser use tutorial um like I mentioned this is made by the same creators of browser use right here so if we go up here you can see it's made by browser use and this

Is a really cool tool they also have a h a cloud version too um that's 30 bucks a month but of course we're going to be looking at the um self-hosted versions The open- Source versions but um yeah if you want to check it out you can check out their website here I would also recommend checking out their docs especially if you're going to be using uh the browser use right here where it's in Python code Beauty behind web UI is

It's actually a gooey so you don't have to worry about any code custom code anything like that very very simple to use and you can easily customize it with any llm so we're going to get started first thing you're going to want to do is go to this repo which will be linked down below of course I'm also going to link the docks right here for the quick start guide so what we're going to do is simply just run get clone

And then the repo URL right here and then we're going to change directory into web-ui and those commands of course are simply right here then what we're going to do is set up our python environment so we recommend using UV is what they say right here so just download UV it's good for managing python environments then just run UV vv-- python 3.11 to create the virtual environment and you can see that we did that right here next we're going to run

Depending on if you're on Windows or Mac or Linux for Windows you're going to run uh either this in command prompt or this in Powershell right here um and then on macro Linux You're simply just going to run this so run those commands to activate the virtual environment then we're going to install our dependencies so you're simply just going to run this command UV pip install dasr requirements.txt and then we're going to install playright so you're going to run playright install

Now once that's done we're going to want to configure our environment with our environment varibles so I just like opening it in cursor you can open it in vs CL or whatnot so I just run cursor dot to open it up in vs code which is right here and then what you're going to want to do is copy your env. example which you will see make a clone of that and rename it Tov now you have two options you can either

Set your API Keys up in here or you can set it up in the UI or you can do both or either so you have a lot of options here as you can see open a I key anthropic Google Azure deep seek mist ol llama so you can use any open source model then you have some different stuff down here as well I'm just going to keep this all on default okay and once you configure those environment variables we're actually going

To start things up now you will see that there is a Docker installation option too I'm not going to cover that it's literally right here if you do want to use Docker I just prefer not to use Docker personally now we're going to run Python webui dopy and then with these flags right here so we're simply just going to copy this and run this Command right here okay once you run this command give it a couple seconds and it's going to

Start on Local Host port 7788 and this is the browser use web UI interface and boy isn't it beautiful I mean it's I mean it's not too beautiful but at least it's you know a nice UI and once you actually see what it's going to be able to do it's very impressive so and as a side note as you can see there's the open AI endpoint right here so if you did want to change it to something like together AI or

Whatever the case may be say if you wanted to use a hosted version of deep seek instead of using the Deep seek API directly through China you wanted to use one that's in the US then you can definitely do that with an open AI like base URL right here and I'll probably cover that in different videos using this browser use web UI tools showing you different use cases cuz there's a lot of different configurations with different llms and different scenarios um

And so yeah I'm excited to use this more often okay so you can see different options here agent type so select the agent type you want to use or or custom we'll just keep it org and then Max Run steps so maximum number of steps the agent will take you can see Here Right Now the default is 100 Max actions per step all right so the default here is 10 and then enable visual processing capabilities we're going to turn that on

And then llm configuration so you can select your uh llm provider right here and then you can select the model name and so we could use uh 03 mini if you want which is really nice this is a very very powerful model and I really enjoy it so far and then you can change the temperature right here so control the randomness in model outputs oh you could also add the base URL right here which is really nice that they have this

Configuration in the UI and you could also add your API key here too so you can see here leave blank if you just want to use a EnV but if you want to override that you can do that here as well next is the browser settings so you can use existing browser instance keep browser open between tasks run browser without headless mode disable browser security features enable saving browser recording so this is really cool too browser width height recording video path

Trace path uh agent history save path so you can see these are all saving in the temp file are you tired of pouring thousands of dollars into appointment Setters 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 centers that work 24/7

Never get tired and book appointments while you sleep trained on thousands of successful conversations our AI agents outperformed 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 so you'll be able to access that

By just simply going to the temp file right here and then run agent will come back to this in just a second deep research will also come back to this in just a second the results are going to show right here latest recording all that stuff recordings are going to show right here and then configuration so you can load config file or you can save a current config file if you save a current config file it's going to save in this

Temp folder right here and it's a pkl file all right so here we can put our task description here so for example I'm saying go to youtube.com search for Josh pook find the names of One his last five most recent videos two how many subscribers he has and three create a Content plan for his channel to create videos make it in- depth and custom to him based on your research any additional information if you want so you can add hints for

The llm to complete the task we're just going to go ahead and click run agent now you can see we're getting those logs in the terminal and it actually just opened up the browser so you can see that it went to YouTube already we can see the memory the next goal the action right here where it navigated to so that completed step one now it's analyzing the page and it's going to use step two so you can see it's using the

That same kind of uh interface it uses with browser use but now we're using it in the web guey now it's typing in Josh pook my okay we pulled up my YouTube channel here it's analyzing the uh screen again okay and boom we got our report here so let's see so Josh has 16.8k subscribers um we can see the names of these recent YouTube videos right here so it's pulling them right here and then content plan for Josh's channel so deep

Dive reviews you can see the description here howto series industry Trends case studies Community engagement collaborations Insider insights and keep in mind guys this is a very basic prompt I could get it to maybe do more do some competitor research get some ideas of different other um topics and channels to cover but let's go ahead and actually check the results right here so if you look at the results we can see the final results so this is the output we can

See the model actions right here the model thoughts we go to recordings here and refresh we can actually see the recording so if I play this we can see it going to YouTube right here now it's searching my name going to the channel you can literally see the whole entire recording which which is really nice and really cool okay now let's try the Deep research this is something that everyone's been raving about and yes open ai's deep research is really cool

And Powerful but let's see what an open-source version can do and this is within a guey we covered one yesterday that was in the command line and it's really great as well but we're going to go ahead and put this in for our research task so compose a report on the use of reinforcement learning for training large language models encompassing its Origins current advancements and future projects substantiated with examples of relevant models and techniques the report should reflect original insights and

Analysis moving beyond mere summarization of existing literature and we're using a Max search iterations of four and a Max query per iteration of twoo so this is really nice that you can customize this because this is something that you cannot do in open AI deep research and we're going to go ahead and simply just click rund deep research and boom now we can see it actually opened up two brows ERS right here and we can see that it actually did two

Queries so history of reinforcement learning and large language models and current reinforcement learning techniques large language models so this one's looking up the history and this one's looking up the current techniques so now we're going to assembly. a right here as well as medium right here it's getting this information from the blog and now it's actually taking in these learnings from this blog as you can see the blog text right here and now it has a new memory and now a

Task progression so it's taking in that new memory progressing in the task having future plans thoughts and then summarizing and then moving on to the next action okay so it extracted multiple different pages that continued on doing this and then as you can see at the end task completed create a gif at agent his.gf now it's starting the second search so you can see here it opened up two more browsers this is the origins of reinforcement learning and large language models

So it can open up a bunch of different browsers at once as as you can see I set it to two but it's doing all these different queries future prospects of reinforcement learning and large language models and it's basically using the whole browser use features that we were just showcasing to do deep research as actual a real human would you can also scale this a lot more easily because you can put it to open up three browsers at once do different

Queries at once and really get in-depth knowledge around that specific topic as you can see here it's going to different blogs pulling up different archive files so as is processing if we go to our temp file right here we can see agent history right here we go to deep research we can actually see markdown files of the research that it is actually extracting from the web so right now we have four different markdown files and you can see all the other

Stuff here like recorded videos traces web UI settings really cool that they have this all right we can see that did two more queries here so future developments of challenges in resource learning for large language models examples of model using reinforcement learning in llms all right so we can see report saved right here under deep research and if we take a look in our guy we can actually see our deep research report so I'm not going to go ahead and read

Through it all of course but we can see reinforcement learning in training of llms Origins advancements and future projects the origins of reinforcement learning in llms current advancements in reinforcement learning for llms we can see it's referencing different citation notes right here which are at the very bottom where we could go actually look at these methodological insights and applications future prospects and challenges and then conclusion so we can see the references here if we click on these references it's actually going

To take us of course to the ones that we just saw you can see assembly AI right here medium right here and you can download your report in a markdown file like so or like I showed you before access it in the backend side in your IDE so here we can see under deep research query results final report and then boom there you have it all in all guys this is a very powerful open source alternative to open ai's operator as

Well as deep research I really like what the team at browser use is doing and what they're putting together here I like that they threw it all in to an actual web UI to make it easy for anyone to use this tool and really save you potentially $200 a month me personally I'm going to just be using both because you know I love open source I love open source tools I also don't mind at the moment just paying for a little

Bit more convenience let me know what your thoughts are in the comments Down Below guys are you guys currently using open ai's Pro Plan let me know if you actually bought that or you're just like no it's not actually worth that $200 a month and let me know if you guys have tried browser use have you tried the gooey version and have you tried the Deep research version and if you're into deep research definitely check out the video I made yesterday

On another deep research tool and you can compare them and see which one you actually prefer better for open source versions other than that guys if you're new to the channel we upload videos on the time ai ai agents AI coding NAD Marketing sales business grow if you like that type of content you got some value here I'd really appreciate it if you like the video comment down below and subscribe to stay up to date with the uploads also too guys

If you haven't already joined our free Facebook group and Discord Channel I'll leave a link down below Shri community.com and then also too guys you're definitely going to want to join our free stride AI Academy for behind the scenes resources tools training Etc and you can network with myself as well as other like-minded individuals in the AI space then also too guys if you run a business and you need help implementing AI agents into your business like AI appointment centers Ai

Call center AI automations in N or in Python whatever the case may be book a call down below at executives.com apply or if you want actual help selling these AI automations AI agents to other businesses for 2 to 10K plus and book a call down below to speak to my team and we can see if it's a fit or not other than that guys I will see you in the next video keep hustling keep grinding and of course guys accelerate your

Stride take care