Beyond APIs: How AI Web Agents Are Automating the "Long Tail" of Knowledge Work

Hi, we're Arjun and Bhavani, and we're here to present to you Rtrvr.ai. Rtrvr.ai is a universal AI web agent that can do tasks autonomously on the web, extract structured data, and call APIs, all for you with just natural language and with you as you browse, like as a side panel with you as you browse.

So we believe this will dive deep into use cases, as well as dive deep into how these kind of category of new universal AI web agents will transform people's workflows and be as transformative to the browser as its creation itself when Netscape came out.

So your browser is a bottleneck for most of your workflowsright now. Even for people with full-time jobs, they're spending hours of their day manually copying and pasting from one website to another, a website to, like, Google Sheets or to their CRM, or they even offshore the scraping to third parties that are expensive and also unreliable.

Yeah, and/or they set up RPA bots that just break whenever a website changes. And you also have silos where some data is only available on the website and some only on the APIs, leading to a hassle for users to combine this data across both, which leads to untapped potentials of what you can do and what you can leverage this data for.

So Rtrvr changes all this by being a Chrome extension that leverages being an AI web agent. So you can open the side panel and do tasks, give tasks to do autonomously across pages, as well as extract structured data to Sheets.

So let's dive into a use case. So say you're on LinkedIn and you have our Chrome extension open. You can give a prompt just, like, for the latent, like, find and follow the

latent space podcast page, and, you know, if following, ignore.

So just with this natural language prompt and clicking a button, our AI web agent will go to the task of interacting with the page to do this task. So it can fill out the prompt, it can fill out the search field for latent space podcast, it can interact with all elements on the page.

So it already noticed we're already following the page. So yeah, and yeah, it even more we can do even more complex use cases than doing tasks autonomously. We can also have this built-in feature to extract data to Sheets.

So say you put this prompt, like, for every article on the page, extract this data, and click this export button. We would write all of this to Google Sheets for you. And using our AI browser agent technology, you can also say, like, do a task on this page and then extract the data.

And, like, you can have all of these combinations of tasks and extractions for you. So it was able to extract all this data to Google Sheets, and we can say that our approach to solving this problem is very cost-effective and it probably costs less than a penny to do this page extraction.

And even more than doing actions on one page, we can do actions across tabs, across, like, Sheets. Like, you can give a Google Sheet column of URLs and we can open those to interact with and extract from. So we can lead to more complex use cases.

So say you're on, like, this archive search and you want, for the first five PDFs, extract these fields. We can break these down as subtasks to do and open them as new tabs and process these tabs simultaneously, independent of each other, and write this data to Sheets for you or interact with those pages.

Maybe they're LinkedIn job applications and you want to apply to them. So we were able to extract all of this data to Google Sheets, and

you can do more complex use cases of you can select tabs to extract from. So say you wanted to compare these Amazon product pages, and you could just leave the prompt empty itself, and our AI web agent will figure out what fields you probably want to extract from and from all these pages and extract those pages to Google Sheets for you, extract those fields to Google Sheets.

And so you can maybe effectively compare all three of these products without having to open them simultaneously. And yeah, so we can even extract URLs on the page. So maybe you wanted to compare these products and you want to see the images themselves, so we can extract the source image URLs and you can compare them.

Even more than that, we can also do actions on the tabs themselves before extraction. So maybe you want, instead of top reviews, the most recent, so select by most recent reviews

in the dropdown and extract the details of the

most recent review. So these tabs are selected, and then I just do extract, and our AI web agent will do actions on all of these tabs simultaneously in the background and extract this most recent review that we wanted.

So

they're all, like, on top reviewsright now, and in the next action, it should go to most recent review. So yeah, it clicked most recent here, and yeah, so it clicked most recent across all of them, and it will give us the most recent review for us.

And even more than doing actions just as basic extractions, we can also do

yeah, so it was able to extract these most recent reviews. And Bhavani will dive into more advanced use cases of doing research and asking across these tabs you have.

Yeah, if these use cases sound exciting to you, let's unlock more potential of Rtrvr in your day-to-day, like, ad job tasks,right? Say, for example, you have, like, a bunch of design docs open and you want to send a quick summary to your, like, colleague at work.

All you have to do is select your design docs and then tell Rtrvr, okay, like, extract key points and summary,right? And when you ask that, it goes through the docs that you, like, selected and extracts the data that you, like, asked for.

So all unlocked productivity in, like, a one-click of a button for you. Not just only Google Docs, you have, like, PDFs, you have Google Sheets. You can give any combination of anything that's on your browser, and Rtrvr goes through them.

As you can see, it extracted, like, specifically what the first document talks about and then what the second document about, like, design of, like, thinking with LLMs, with memory as such. Now, not only with the existing documents that you have.

Say you want to do, like, a market research on a bunch of companies to, like, invest in also. Like, here I'm asking it to go through a couple of agentic companies, like Rtrvr, Bardine, Browse AI as such, and I'm asking it to extract certain things, like strategy of the company, like features and pricing.

And what it does is, first, based on what you asked, it generates the schema. And then, as you can see, we started with the homepage of all of these agent startups, and because we asked for, like, pricing, it navigated to, like, pricing page.

So Rtrvr is capable of going, like, basically, like, exploring the web page or going deep into multiple pages to find the data that you asked for and extracts itright into the Sheets for you. And also, we are one of the very first in implementing this deep research feature compared to operator also.

And as you can see, the best part without deep research is your ability to select all your documents that you have access to, not just on the browser. And you can specifically give what all URLs or websites the agent can go in terms of, like, extracting this data for you.

As you can see, it successfully extracted all the information you asked forright into your Google Sheets for you. Now, like, let's do this on, like, some steroids,right? Now, like, how about I want to market research, yes, but I want to market research about a couple of stocks.

And then you want to, like, buy, make a purchase choice as such. So here, all I have to do is select theright Google Sheet that has the information. So, for example, I have this Sheet with a couple of stocks lined up, and then I'm asking it, like, complex data.

Like, I'm asking it to give me, like, P/E ratio from Yahoo Finance and some revenue in the last two years, and I'm also asking it to create new data fields of computing the revenue growth based on these numbers.

So what it does, it's, like, first, goes on, opens these websites, and tries to, like, extract this data for you. Say, for example,right now I gave the websites as such. Now, even though I gave the wrong URL, instead of giving Yahoo Finance, I just gave companies' websites, our agent figured out that it was not on theright URL, and it went to, like, Yahoo and, like, kind of, like, tries to extract this data for youright into your Sheets as such.

Now, so how cool is that? Like, even, like, there's, like, an error from the user, Rtrvr was able to, like, correct this for you. And one of the cool features we have is our function calling features. Like, a lot of companies have, like, a bunch of, like, third-party integrations with different tools, say Slack, like, say Discord also.

One of the features we are very keen on implementing is, like, a dynamic function calling. Like, instead of us implementing only, like, a set of connectors as such, you can call any API, any third-party tool out there by literally providing the information about the tool here and easily integrate.

Say, for example, I have WhatsApp integrationright here to send messages to, like, WhatsApp numbers. Let's say you're a small business that you have, like, a bunch of customers and, like, a bunch of, like, customer phone numbers as such.

So I have two users here, and let's say I'm asking Rtrvr, all I have to do is, like, select the Sheets for you, select the Sheets that has the customer numbers, and then I ask Rtrvr to, like, send a message to customer phone numbers here.

So when I do that, it identifies each of the customer number provided and invokes this WhatsApp tool, essentially your third-party tool, to send a message to these users. And as you can see, it says all messages sent successfully.

Let's check it out. Let's try opening the WhatsApp. So this is my test number, as you can see. We're demoing it at 2:32. So yeah, this is a 2:32, like, successful test message that we received. So yeah, imagine the world of possibilities, like, automating your social communications across your Instagram, Facebook, WhatsApp, everything, all in, like, one click with Rtrvrright now.

Now, let's do one more use case, which is graph generation. So we have, like, a graph bot tool. It's like, almost like a mini agent within Rtrvr Agent Studio. What it does is, let's say you have, like, a bunch of data as such, like, and then you want to generate, like, a data analysis graph on top of it.

You can leverage the graph bot in doing so. As you can see, it's like a dynamic graph generated on the fly for you. So we've realized when working with LLMs, they're not only inherently good with, like, data extraction as such, but are also capable of generating or representing this data in various formats of interest to the users.

And we leverage that capability to build such cool use case and present it to you. Now, like, bringing it back home, now let's look at the overall agentic landscape,right?

Like, let's look at some of, like, the OpenAI's operator, like, Anthropic Claude, and, like, obviously, Rtrvr, and so on. So most of the agents that are out there use, like, vision-based approach. Like, your Anthropic Claude or OpenAI, even, like, Google Mariner use a hybrid approach, also including vision approach.

As such, what do they do is they take screenshots of the pages and extract the data that you're asking for. And what's the inherent problem with this approach,right? The vision-based models are more prone to hallucination compared to what Rtrvr is doing, which is text-based approach of leveraging the web page stuff.

And vision-based approach is also highly expensive, taking multiple screenshots for just, like, one single action to do, or, like, for one single page scrolling as such. And most of these companies, such as these browser-based or so on, they use browser on the cloud, unlike Rtrvr, which is an extensionright inside your browser.

So what are the problems with browser on the cloud? Like, number one, non-personalized results,right? It's like a generic page that's opened on the browser. It's not like the content might be totally different from what you're seeing in your browser.

And also, because to support browsers on the cloud, these companies should implement a lot of proxies to funnel the network or network requests to theright IPs as such, which is way more expensive than the whole agentic setup as such.

So with this text-based approach, as well as being, like, an extension in your browser, Rtrvr is capable of processing not only the active tabs but also your background tabs or, like, multiple tabs at once. And it can even go beyond your subscribed content.

So with Rtrvr, you don't have to, like, share any of your passwords. We do not store any of the passwords. Whatever you're logged in and you're seeing, Rtrvr sees the same thing. Unlike on the cloud, either you have to, like, store or give the passwords, which are prone to, like, various security risks, or you cannot get behind paywalls, or even, like, Cloudflare website protections as such,right?

Now, like, Arjun will dive deeper into some of these and, like, yeah, bring it back home.

Yeah, so just to recap what are the advantages for Rtrvr, it's that

yeah, maybe I can share this. Actually, it's fine. So to highlight the advantages for Rtrvr, since we're using a text-based approach, for creating output, there's much less hallucination because the text is

because the text isright there in context for the model. And because we are using this text-based approach, we can also take actions on multiple tabs because these background tabs don't get rendered. So you actually can't use a vision-based approach to take multiple actions in parallel.

So when we can leverage this background processing to do actions, to do multi-tab context-ed actions, as well as just speed up performance by taking actions in parallel. And compared to our competitors of, like, trying to do consumer applications of booking flights or booking restaurant reservations, we're all in for productivity and automation use cases because we see that AI is perfect to automate these manual repetitive tasks that is a burden to a lot of people.

And compared to everyone else, we are setting up a client-side Chrome extension that is not only cheaper infrastructure-wise, but it leverages us to use new sources of content for the agent to work on. So you can access the local vault sites or login vault sites, and it is much more secure than storing your passwords in a cloud-hosted browser.

And compared to other providers of doing one single long actions on a single tab with long horizon tasks, we distribute subtasks as new tabs to take actions on. So our failure rate is much less than these competitors, so we don't have to deal with exponential failure rates.

And our approach to third-party integrations is that the user can define and set up function calling that function calls that they can share, and this is much more extensible and scalable than setting up custom third-party integrations that these third parties, that these competitors are doing.

Yeah, so this is our mission is to basically revolutionize data extraction with a transparent and efficient AI-powered exchange is our long-term goal of being allowing people to collaborate across their own local laptops to collaboratively construct data sets and set up, like, very cost-efficient and cheap data sets.

So, for example, it could be, like, all the local government events happening in the SFPA area that would involve, like, extracting data from hundreds, thousands of websites. And this is not something that is feasibleright now. It's, like, not cost-effective to do.

But if people can use it, leverage our extension to volunteer and collaboratively construct data sets, we believe that's an exciting new future and use case that is just on the horizon.

Thank you, and please feel free to go to rtrvr.ai to download the extension. That's rtrvr without any vowels. Or you can scan this QR code and try it out, and hopefully you can get a glimpse into what the future of AI agents and the browser will be.