Build ANYTHING With AI Agents For FREE! (DeepSeek-R1 Beats ChatGPT)

Summary

In this video, Ishan Sharma demonstrates how to build AI agents without any programming knowledge. By using browser-use and various language models like DeepSeek-R1, users can automate tasks such as shopping online by following straightforward steps. Sharma walks viewers through the installation and setup of necessary tools, including Python and browser-use, to create agents that can perform a multitude of automated tasks, all for free.

Highlights

• Build AI agents in less than 30 minutes without coding! 🚀
• Automate purchasing on Amazon with AI - hands-free shopping! 🛍️
• Setup browser-use with Python easily for AI tasks! 🐍
• Explore versatile language models like DeepSeek and Gemini! 🧠
• Create and manage AI tasks through a simple web UI interface! 🌐
• Utilize local and API-based AI models in your automation! 🌍
• Experiment with AI tools like n8n for custom workflows! 🔧

Key Takeaways

• Learn to build AI agents without coding skills! 🤖
• Automate online shopping with AI agents! 🛒
• Use browser-use and language models effortlessly! 💻
• Create custom automation workflows with ease! 🌐
• Save time and effort with AI-driven tasks! ⏰

Overview

Ishan Sharma takes us on an exciting journey into the world of AI agents. He shows how, without a single line of code, anyone can set up an AI agent capable of automating tasks such as purchasing from Amazon. This tutorial video is packed with insights on leveraging browser-use and Python to ease into the world of AI automation.

The video delves into the technical aspects of setting up Python and installing the browser-use library to create AI agents. Users can follow along to learn how to use different language models, such as DeepSeek-R1, to perform tasks automatically. It’s all about making complex AI accessible to everyone!

For those interested in exploring further, Sharma introduces the n8n platform for more customizable AI automations. By the end of the video, viewers are left equipped with the knowledge to take control of their digital tasks, saving both time and resources using AI-driven solutions.

Chapters

• 00:00 - 01:00: Introduction to AI Agents In the chapter 'Introduction to AI Agents', the author describes their experience building an AI agent in under 30 minutes without any prior coding knowledge and at no cost. They demonstrate the capabilities of the AI agent by showcasing how it can automate the process of purchasing items on Amazon. Specifically, they explain how the agent logs into an Amazon account, searches for a product, adds it to the cart, and completes the purchase using an existing payment method and address. This example illustrates the potential and ease of use of AI agents in automating online transactions.
• 01:00 - 03:00: Installing Python and Browser Use The chapter focuses on the future of AI agents, demonstrating how users can create their own AI agents without any prior coding knowledge. It emphasizes the wide range of tasks these agents can perform, such as booking flights and recommending YouTube videos, automating tasks typically done manually.
• 03:00 - 05:00: Setting Up Web UI and Playwright Chapter Title: Setting Up Web UI and Playwright Summary: In this chapter, the focus is on creating an AI agent by first installing Python on your system. The technology used is 'browser use,' and the choice of language model (LLM) is up to the user, with options such as deep seek R1, Gemini, GPT 40, and ov1 models. The chapter includes a demonstration of setting up the AI agent for free using Gemini, utilizing a free open-source tool available on GitHub.
• 05:00 - 09:00: Creating AI Agent with Gemini LLM The chapter provides guidance on creating an AI agent using the Gemini LLM repository. It reassures readers that despite the complexity of code, the process is simple and straightforward. Readers are advised not to worry about comprehending all the code and are encouraged to use tools like ChatGPT to assist with installation queries, such as installing browser use on Mac OS, by following straightforward instructions given in the readme section.
• 09:00 - 16:00: Using DeepSeek Locally and through API The chapter titled 'Using DeepSeek Locally and through API' provides a detailed, step-by-step guide on installing and setting up Python on your system to use DeepSeek. It starts with instructions on downloading the latest version of Python from the official website, followed by guidance on installing it on your computer. Once installed, you are advised to move the installation file to the bin. The chapter concludes with a command, likely meant to be executed in the terminal, using pip3 to install necessary packages. This reflects a focus on enabling the user to effectively utilize DeepSeek either locally or via API.
• 16:00 - 21:00: Building AI Automations with n8n The chapter discusses the process of setting up and installing libraries and tools required for building AI automations using n8n. It begins with the installation of 'browser use' library, followed by 'playwright'. The focus is on following instructions step by step, ensuring each component is installed properly, highlighting the process of integrating web UI for interacting with browser-based automations.
• 21:00 - 22:00: Conclusion: Potential of AI Agents The chapter discusses the usability of AI agents, particularly in enhancing browser interactions. It emphasizes the ease of testing browser features with a user-friendly interface, making it accessible for users to experiment and develop with AI. The example provided involves creating a simple folder for an AI demo and navigating through terminal commands to manage files and directories. This highlights the potential of AI in simplifying technical processes, making them more approachable for users.

Build ANYTHING With AI Agents For FREE! (DeepSeek-R1 Beats ChatGPT) Transcription

• 00:00 - 00:30 I just built an AI agent in less than 30 minutes without knowing how to code for completely free hi I'm isan and this is a crazy AI agent I've just built which goes on amazon.in logs into my account I gave it the password searches for a detergent selects a detergent it adds it to the cart and it then purchases it on my behalf using my Amazon pay balance selects the address and and makes the purchase that
• 00:30 - 01:00 is insane this is the future of AI agents and by the end of this video you'll learn how to make an AI agent just like this for completely free without knowing a single thing about coding watch this till the end there's so much to learn and you can ask AI agents to do anything book flights for you you can ask AI agents to recommend you videos to watch on YouTube you can literally ask AI agent to do all the things that you do yourself done automatically for you so
• 01:00 - 01:30 that it can save you the time and the money to do it let me show you how it works so to create an AI agent like this all you need to first of all do is to install python on your system the technology we use today is called browser use and the llm we use is up to your choice you can use deep seek R1 you can also use Gemini you can also use GPT 40 and ov1 models depends on what you want to do I'll show you how to do it for completely free using Gemini as well so browser use is what we'll be using this is a free open Source GitHub
• 01:30 - 02:00 repository which will help you to create any AI agent that you want all right let's go down into the readme section of it and if you are getting annoyed by looking at all the code don't worry it's very simple and very straightforward you don't even have to worry about anything you can just select all the text on this website go to chat. openai and just ask this how do I install browser use on Mac OS for example and just give it all of this
• 02:00 - 02:30 data and it will tell you step by step how can you install browser use package right so this is a very simple way to do it first you need to install the latest version of python so you can go to python you can go to downloads and you can start downloading python on your own system and once you install python on your computer you can basically close this move it to bin and then basically just copy this command go to terminal and paste this I'll use pip 3 install
• 02:30 - 03:00 browser use and it is going to install the library of browser use on my system so it'll take a couple of seconds to install that and now it is installed the next step is for us to install playright so we'll just copy the same text paste it over here again you don't have to worry anything about the code just follow the instructions and just do the things step by step so we install playright and once playright is downloaded the next step is to install web UI so basically with browser use web UI is going to to help you interact with
• 03:00 - 03:30 browser use with a nice interface so that's all that it really is about so now if you can go down you will see test with UI you can test browser use with a UI repository so we'll just open this with a new link I'll make a very simple folder on my desktop I'll just say new folder and I'll just call it let's say AI agent demo and all I have to do now is to open up my terminal and I can use LS to look at all the items in this directory I can do CD desktop and I can
• 03:30 - 04:00 enter desktop and after that I just have to say CD AI agent and I'm inside the AI agent folder right now and now all I have to do is to say that I want to get clone and I'll put this link that I got from this particular repository so I can just go to code and I can just grab this link right here this link that you can see I'll just copy this I'll put
• 04:00 - 04:30 this link over here and I want to install this repository in this particular folder so I'll just click on enter and it is now installed so now if you would go here if you would open AI agent you would have a web UI folder inside so we will again click on LS and we can just say CD web UI and now we are inside the web UI folder okay so now we we are pretty
• 04:30 - 05:00 much done with setting this up if you go down and look at how to start using web UI it will say you have to first of all use UV what is UV I don't know we'll just click on it and see so UV is an extremely fast python package and project manager written in trust again you don't need to know what that is or how that works you basically just need to install it so I'll just grab this link or this command I'll put it in my terminal and I'll enter and now I have
• 05:00 - 05:30 UV installed in my system yep it's good to go so this will help us create an environment right so we will now copy this First Command UV VNV python 3.11 again you don't need to know what that means just press enter okay it says UV not recognized I have to go to my a new terminal and I can just say CD desktop CD AI agent I can do CD the web UI and I
• 05:30 - 06:00 can just now say UV and this Command right here and now what will happen is it will create a environment for you for me to run this project so now I have this environment activated it says activated with this particular command I'll just put it over here and now I am in this environment so as you can see it says web UI so we good to go with that we'll just go and install the requirements so I will just put this
• 06:00 - 06:30 over here there's a requirements.txt file in this folder which got downloaded when you get cloned it so it's just going to install all the dependencies for this project to work so we let that happen and as you can see I have not even looked inside the folder I not even seen the code yet I don't even need to do that that's the power of using browser use and building AI agents with it so we let that happen and once that is done we'll also just install Play I
• 06:30 - 07:00 think it's already installed yeah it's already installed here and now we just have to have this one command python web ui. py and we'll just copy this and paste it right here what this does is it initiates a local host Port through which we can open up our AI agent window so now you'll be able to see browser use web UI it'll take a couple of seconds but then we can basically go to https SL this particular address so I'll just
• 07:00 - 07:30 copy this and if I paste it over here there you go this is browser used web UI this will help us run any agent that we want to so I'll close the previous windows and this is all that you have to focus on let me just make it bigger oops Yeah so you'll see you have multiple things over here you have first account settings you then have llm configuration this is where you'll put who is the LM provider what is the name of the model and the temperature which is basically how creative you want the model to be
• 07:30 - 08:00 and then if you are using some API then you can enter this over here then you have the browser settings which you don't have to mess around with then you have the important place which is called run agent this is where you will describe what do you want this agent to really do and once you do that then the agent will do it on your behalf then you have configuration you don't have to worry about this and the results page is where you'll be able to see the recording of what this agent has just done and in the final result it will show the output and there are some errors if there are any to it will show
• 08:00 - 08:30 over here then you also have some recordings in this particular instance but what you have to first of all do is to create a llm provider for it so this is the browser use part of it the next part is picking a llm provider now you have so many of them right you can go to chat GPT you can use this API of chat GPT you can go to Gemini and you can use their API key or if you want to do something else then you can always go to chat. deep seek.com and use the API of
• 08:30 - 09:00 deep seek let's just search Google AI studio so we'll do that we get Google AI studio right here and let's create a API key right so we don't want to use Gemini we want to get the API of Gemini so let's click on get API key and I'll just say create API key and I'll just use Gemini API it'll take a couple of seconds and now I have an API key generated and I
• 09:00 - 09:30 can use this API for completely free you can go on to LM provider and browser use and click on Gemini and over here let's just use the 2.0 flash experimental this is a model that Gemini has launched uh this is a smaller faster model also experimental but we can basically just paste the API key right here and we are good to go that's all you need now you can go to the Run agent section and specify a task that you wanted to do very basic thing let's just say go to
• 09:30 - 10:00 google.com and search let's just say deep seek and click enter and give me the first five URLs that you get that's it that's all I wanted to do and I'll click on run agent and it will create a window automatically and it will do the things on my behalf look my hands are here and it is doing the thing automatically it's going to go on Google it will
• 10:00 - 10:30 analyze the CSS of Google it will find out where is the search bar it will then enter deep seek it will then hit enter and in some time it will fetch the top five links that it sees on Google automatically without me having to do anything this window will close and you can just go to the results tab you can go down and you will see the top five links from the search that it did automatically this this is incredible
• 10:30 - 11:00 now it might seem very lame but you can do anything that you can think of I said go to amazon.in login into my account using email as I entered my email and password as I gave it my password and find a let's say 2 kg detergent and add it to the
• 11:00 - 11:30 cart and order it for me using my Amazon pay balance to the address which has I can enter my address I just entered my flat number and I said in the saved addresses list this is it this is the only prompt I entered and with this what it did was
• 11:30 - 12:00 it was able to go let me just grab that video to show it to you it literally just went to Amazon it logged in entered the password searched two 2kg detergent it picked a detergent it then added it to the cart and it then proceeded to check out and I had some balance in my account left and it was able to order that using the address and just clicked on place
• 12:00 - 12:30 your order that's it my order was placed automatically without me having to do all the things myself this is mindblowing you can ask it to book flights for you you can ask it to book movie tickets for you you can ask it to do anything that you might want to do this agent was made using browser use and Gemini however you can also use chat gbt API but that is paid but you can use the Deep seek model locally that is stored on your system or you can also access is the Deep seek API I'll show
• 12:30 - 13:00 you how to do both so if you want to use a local deep seek model on your computer you can download ama ama is basically a way for you to run any large language model on your system itself and once you do that you can basically just go to terminal I'll close this you can go to terminal and you can just say o Lama and we can install any model that we want to so I'll just go on models I'll go to deep seek R1 and I'll ask it to install the 14 billion parameter model so with
• 13:00 - 13:30 this I'll be able to locally Run Deep seek in my system I showed you how to do this before in the last video but I'll go here I'll grab this I'll go here and I paste it and with this command I can download the 14 billion parameter model that I need it is recommended that you use a larger model if you want to do AI agentic work with deep seek stored locally so you should do this and once you do that I will show you the models that I currently have
• 13:30 - 14:00 so AMA list and I have the 8 billion parameter model so now that I have that I can basically just go onto browser use I'll have to launch this again sky is the limit to what you can do with AI agents as you can see you can book your own flights using the system that I just showed you you can even give it a simple prompt to read your CV and find machine learning jobs save them to a file and even start applying for those jobs on your behalf this is insane today it is possible for
• 14:00 - 14:30 you to do all of these tasks automatically this is the power with AI agents made with browser use and Gemini again with deep seek you can go onto this deep seek model install it and once you have it here let me launch browser use once again so this is working so I can go on to browser use and right over here I can go onto the llm configuration I have my amaama working so I have access to the Deep
• 14:30 - 15:00 seek model locally so I'll just go on to AMA and I'll just write down deep seek R1 deep seek Das R1 hyphen 8B that's my model um and after that I can basically just ask this agent to go onto google.com and type open Ai and click search and give me the first URL I just click on run this agent it opens up a browser it goes to google.com and with
• 15:00 - 15:30 just that you'll be able to build an AI agent using deep seek and browser use for completely free stored locally on your computer so that's how you do it with deep seek but let's say you want to use a deep seek API in that case you can open up open router this is basically going to help us connect with the Deep seek R1 API I will quickly sign in and once we are here we can just go on to deep seek R1 this is the free version but I can also go to the Deep seek
• 15:30 - 16:00 R1 I can go to the API I can create an API key I'll just say let's just say isan tries deep seek agent I'll click on Create and now I should be able to create this key and paste it right in the LM configuration so in this case I'll just go to deep seek I'll enter my API and I'll also enter my
• 16:00 - 16:30 base URL now this base URL should basically be this but API slv1 and with just that I should be able to use the Deep seek R1 API through open router on browser use to create a AI agent that's how you create AI agents using browser use this is incredible the amount of stuff you can do with it without any limits is insane openi is charging you $200 to access the operator which they have created which
• 16:30 - 17:00 is similar to this but that has so many limitations YouTube has banned it but you don't have to worry about it if you're using this particular setup right here again I will be deleting all of these API keys so don't try to use these make your own API key especially for open router you actually have to pay for it so for example I have some $4 in this account so that I can use it and this is how you are able to create any AI agent that you want with browser use and any of the llm providers deep seek Gemini CH
• 17:00 - 17:30 GPT another thing you can do is to go and set up your own AI automations with n8n and I'll show you an example of what I mean by that so n8n is a AI automation platform with which you can without knowing how to code create your own AI agents so I'll just log in and once that is done you can create your own workflow so I'll just click on create
• 17:30 - 18:00 workflow and in this this is a environment where I can create a workflow that I want I'll show you a very simple example of what I mean by this so I'll go to create your first step and I'll just say chat so this will have a chat trigger basically whenever I will chat with it it will be activated and once I send a text message I want it to reply through a AI agent so I'll just click here add a node I'll click on AI agent this is the AI agent that we have
• 18:00 - 18:30 now you can see three things right it has chat model it has memory it has tool so let's first add a chat model to it so I'll just click on plus we'll go to the open AI chat model and now I have one selected over here which is my own credential but let's do it from scratch so I will go to open router I'll go to this free R1 API I'll go on to create API key on open router let's just say trial create I'll copy this API key and
• 18:30 - 19:00 I'll paste it right here as for the base URL I will basically just grab this and paste the open router link over here I'll say API slv1 I'll click on Save and I will go back now I have to change the model ID so if you go back to this model it's called Deep seek deep seek R1 free so we'll just grab this and
• 19:00 - 19:30 we'll just paste this over here I'll go back I'll click on Save and now once we are done setting up our chat let's just ask it something I'll say hi who are you let's see what it says it'll take a couple of seconds and in some time we got the reply which says hi am deep seek light preview and AI assistant created exclusively by chinese company deep seek so this is now working for us and now we can add any automation
• 19:30 - 20:00 that you want in it for example you wanted to read your emails you wanted to reply to your emails as well so I'll just add one node over here I'll call it Gmail and I can select any of these things so I'll say get many messages and now this is one node that I have with me right now so I'll limit it to five I only want five emails to see at one time and that is all that I need to do and now have a Gmail get message
• 20:00 - 20:30 node right here so if you go over here you have to set this up you have to actually login with Google for it to look at all of your Gmail account but I have already done that so that I won't have to do again so I'll just say show me today's unread emails and I'll press on enter and you will see a problem so it says no prompt specifi I'll go onto this I'll look at what Json
• 20:30 - 21:00 we got I'll copy this Json and I'll go to Define below I'll enter this text right here and once I do that I will just go back and I'll say save and now if I ask this again it's going to go onto my Gmail and give me the top emails that I have for today which are unread and the best part is using these tools n81 or or browser use you can set up your own AI agents
• 21:00 - 21:30 and use any model that is performing the best it could be O3 mini which just launched by open AI it would also be deep seek it could also be Gemini the 2. 2.0 version which we just tried out the possibilities are endless it just depends upon what do you want to build and what thoughts you have about how can you build stuff from scratch that is all from me today that's how you create AI agents without knowing how to quote for completely free for less than 30 minutes I hope you enjoyed this video if you're still watching right in the comment
• 21:30 - 22:00 section I watch till the very end I'll see you in the next one subscribe share this with your friends let me know what other videos do you want me to make also let me know if you have any questions and I'll see you in the next video bye-bye