Imagine an AI that doesn't just browse the web, it controls it. An AI that can scrape data, file JRO tickets, fill out forms, trigger workflows, and even automate complex 30inut research projects, all within your own browser or straight from the cloud. Well, allow me to reintroduce Retriever, the next evolution of the AI web agent, [music] which has been now rebuilt from the ground up with MCP, cloudscaling, and multi- aent intelligence. Previously, Retriever AI was only known for being a powerful local first browser extension, one that ran completely inside your own Chrome browser. It let you automate websites applied to jobs or collect data securely [music] and privately all without relying on external servers. But with this new version 2 update, Retriever takes a massive leap forward. Now your browser itself becomes a remote MCP server, meaning other apps and AI agents can directly control your browser

remotely. For example, if you were to tell Retriever to send a connection request to Google CTO Jeff Dean, the agent instantly understands and it's able to open up LinkedIn in the background, find a profile, and send the request all within the browser using credentials securely. This is because Retriever runs locally and it can handle websites that block cloud bots like LinkedIn while still executing multiple tasks in parallel across different tabs. And the best part is is that you can save that task as a shortcut and rerun it instantly or scale the exact same automation to the cloud with a single click. Essentially, with Retriever version two, your browser becomes a remote MCP server and it's going to let other apps and AI agents control it a lot better. You can instantly scale any automation to thousands of cloud browsers and run multi- aent workflows for 30 plus minutes using Google Sheets as memory or you can even trigger automations on the go through WhatsApp. This is a new version that is going to

support a lot of new features with slash commands, record and repeat task, and a lot more which we're going to be uncovering throughout today's video. And a huge props to them for sponsoring today's video. This is definitely a great tool and I highly recommend that you go ahead and upload them on product hunt as they have just released this new version today. But what sets Retriever apart from other AI web agents is its overall performance on various sorts of benchmarks. On the web bench, it scored an 81.39 percentage for its overall success rate. It also is super fast with a blazing fast 0.9 minute task completion and an ultra low 12 cents per task cost. It is something that also excels at both reading and writing tasks. It outperforms competitors like Open AI as well as Enthropic. And it's a local first architecture which avoids bot detection. It reuses authenticated sessions. It can even handle captas and support multi-tab workflows. While

DOMbased intelligence is going to ensure accurate interactions with complex web pages, it's fast, it's reliable, and it's costefficient, which is why I really wanted to showcase it on the channel. Before we get started, I just want to mention that you should definitely go ahead and subscribe to the World of AI newsletter. I'm constantly posting different newsletters on a weekly basis. So, this is where you can easily get up-to-date knowledge about what is happening in the AI space. So, definitely go ahead and subscribe as this is completely for free. But let's get started. You have three ways to access it. First is through the WhatsApp bot that lets you trigger automations and fetch data directly from your phone. And it's simply as just texting the bot. You can also access it through the Chrome extension, which is what we showcased before as a remote MCP server, which is going to let your AI agent control your browser remotely. You also have the cloud and API platform, which we'll showcase. But to install it, you can easily click on this button, which will take you to the Chrome store, and

you can add this to Chrome as a single click, and it will add the extension. Once installed, you can then click on the retriever extension. And this is where you can now go ahead and create an account or sign up using Google. It is completely free to access and you can even access a free API via Gemini. But here is your AI web agent. You can do almost anything from automating tasks on the web, extracting data where you can even automate data entry. You can crawl different sorts of pages, create docs, call APIs and MCPS as well as work on different sheets. Now, to get started, you can simply go ahead and automate anything based off the natural language prompt that you paste in. And what you can also do is you can add context based off the page that you want to work on. So if you want it so that it has context for my world of AI YouTube channel page as well as my newsletter, you can add that as context so that it can simultaneously perform web-based

task across these two different pages asynchronously. On the bottom, you can click on teach a trick and this is essentially where you can add in a new recording to teach the AI web agent to perform a task. What you can also do is if you click on the MCP URL, this is where you can copy a URL from retriever so that you can connect your browser as a tool to other AI agents or apps so that it can control and let them trigger real actions on your browser remotely. Also, keep in mind you can use this completely for free. This is the beauty of Retriever because if you wanted to, you can have it so that you can add in your own API from Gemini and Gemini offers a free API tier that you can use on a daily basis. There's also different tools that you can have your web agent use from different MCP servers to user tools like Slack messages. You can also use system tools that are already implemented within Chrome. You can add in different sorts of files and you can

even attach Google Sheets. But now let's test this out. I can go ahead and I can request it to scrape the contents of my YouTube channel page as well as my newsletter. And I can scrape the data as markdown. We can send in this prompt. And you can see that it is going to start off by analyzing the request. This is where you can add in further guidance or corrections if needed. But it's going to now work on planning to execute this task. And you can see that the agentic tabs right now are this uh YouTube channel link as well as the newsletter link. And quite rapidly it was able to do a great job in extracting all the contents. You can see all of the YouTube links that it was able to extract from my YouTube channel and it was able to also extract the link the total views as well as the upload date. And at that bottom it also is showcasing the summary it got from the newsletter. Talks about all the new posts I have on newsletter and it has extracted all the contents for it. And you can see that this was

something that it was able to do quite rapidly. You can share this task. You can copy the messages and you can even create a shortcut for this task to be rerun in the future. And like how you can access the extension for free, you can also access the cloud platform. This is essentially where you can automate and this is essentially where imagine you are able to take any automation you built locally and instantly scale it to thousands of browsers. This is where you can simply extract website datas at scale. You can process spreadsheets at scale and automate any workflow at scale. It is something that you can simply describe in natural language and you can execute it rapidly with the flashlight, the flash model or even a pro model. You can set a default proxy. You can also add in tools like you would with extension. This is where I can execute a task of extracting contents off of angel.co co finding different sorts of AI startups founded after 2023

having it precisely execute this task that is very detailed scrolled until all the results are loaded and then extract each company name description location and a couple of other fields and I can simply go ahead and request this task to be completed but the beauty of this is that you can scale it to a thousands of browsers in the cloud with a single prompt that's exactly what the retriever cloud does it can turn your local workflow into a production ready automation that can run in parallel across massive cloud infrastructure and it's going to give you supercharged scale speed and reliability and you can see it was quite rapidly able to extract all the contents and now we can open it up in a Google sheet. This is where we have the parent website you have the source website location tagline industries and all the different fields that we had requested fully scraped. This is the beauty of retriever. In the same manner, you can do a lot more at scale with the cloud version, but you

have a good hybrid approach of using retriever with obviously with the extension as well as the cloud if needed. But this is the new update of retriever that makes AI web automation a lot easier. I'll leave all these links in the description below so that you can get started right away. But with that thought, guys, thank you guys so much for watching. I hope you enjoyed today's video and got some sort of value out of it. This is, in my opinion, one of the best AI web agents available, and I highly recommend that you take a look at it with the links in the description below. But with that thought, make sure you go ahead and join the newsletter, join our private Discord, follow me on Twitter, and lastly, make sure you guys subscribe, turn on notification bell, like this video, and please take a look at our previous videos because there's a lot of content [music] that you will truly benefit from. But with that thought, guys, have an amazing day, spread positivity, and I'll see you guys really shortly. He suffers.