Transcript: Autonomous AI Agents Are Changing How We Interact with the Web | Abhishek Das - Co-founder and Co-CEO of Yutori

Host: Hello everyone, welcome to startup project. Uh today we are joined by Abhishek, he's the co-founder of Yutori. Uh, Yutori has a really high pedigree, uh, co-founders, um, Abhishek, Devi, and Dhruv, all, uh, been ex Meta AI division.

I think Abhishek, uh, himself has a PhD from Georgia Tech. Um, and Yutori has been building anonymous AI agents that can handle our everyday daily tasks. Uh the first product Scouts by Yutori.

Uh deploys a team of AI agents that track anything that you want on the web uh from flight prices, restaurant reservations, sales leads, um, job interviews, job openings, um, and uh basically help you save time in terms of browsing the web.

Um they raised $50 million from an interesting group of investors including Feli and Jeptine, who are uh the icons of AI industry.

Um in this conversation, we'll explore um the opportunity you know, to go deep um in terms of what we can build with agent AI, what are the technical challenges uh behind building useful AI agents.

Uh, and how is this technology going to transform how we interact with software and more importantly, I think the web. Uh with that, uh Abhishek, welcome to the show.

Guest: Thanks for having me.

Host: Uh so give us a little bit background, you know, how did uh Yutori got started? How did this team uh of uh ex Meta folks uh form and form this company?

Guest: Yeah, um, so I um, just for the background, I I grew up in India. I moved to the US in 2016 uh for my PhD. That was at at Georgia Tech.

That's where I met my co-founders, like now co-founders, uh Dhruv, Dhruv Batra, who who's who's one of my co-founders at Yutori. He was my PhD advisor at Georgia Tech. Um and Devi and Dhruv, they were both faculty at Georgia Tech.

So the three of us worked together a whole bunch as part of my PhD. At the time, so this was in like 2016, 17 as part of my PhD. I was working on problems at the intersection of computer vision, language, and reinforcement learning.

So if you think of um um a home robot and asking it the question, where did I leave my keys?

In order to answer that question, the robot would need to understand language, it would need to know how to navigate an indoor environment, and then come back to you with the answer or with the object that you asked about.

And at the time, I was looking at, okay, how do we train deep learning models for these kinds of tasks? Um again, this is yeah, back in 2016, 17. So LSs weren't anywhere close to as capable as they are today.

It was a fairly exploratory space um to be to be working in. Uh like the first set of sort of deep learning image classification models around like AlexNet, VGG, etc. Those had happened.

And then 2015, 2016 was around the time when um image captioning was uh a pretty big deal. So generating a one sentence description about an image. Uh, so that was sort of state of the art at the at the time.

And I was kind of um pushing it towards being more embodied that, okay, like, if this agent not only does it see and talk, but if it uh what would it take to make these agents act in the physical world?

Um, and initially, I was of course like trying to develop these methods entirely in simulation, uh, but I did some, I did collaborate on some projects where we put these on actual physical robots as as well. Um, I had a lot of fun as part of my PhD.

I spent time at like uh uh on internship, there are a bunch of places like Deep Mind, Fair, uh Tesla, Artel, AutoPilot and and so on. Uh, worked with a phenomenal set of people. Um, I finished my PhD in 2020.

After that, I I joined Fair at Meta as a as a research scientist. Um, yeah, I worked there for like about three and a half years uh till early 2024 when Devi, Dhruv and I also of us decided to leave Meta uh to start to start Yutori.

Uh, Devi and Dhruv, they've they have actually held joint positions as faculty at Georgia Tech, but as well as uh research scientist at Meta since 2016. Um, yeah, so uh like we've known each other for almost 10 years at this at this point.

Uh, I think since we we started having a weekly um sort of uh brainstorming session back in like 2017 or 18, when we would just meet once a week uh to discuss that, okay, like, if we were to start a company at some point, what would it be about?

What would we go after and so on? So it's not like we've been just brainstorming for like 10 years, seven, eight years. Uh over like over time, it just became a casual social hang.

Uh, but that's uh sort of how early we had started uh thinking about starting a company potentially together.

Host: So how how did the idea for like um Scouts uh came about?

Guest: Yeah, so with with Yutori, um when we left Meta, it wasn't the case that we had a clear cut idea of what we were going after.

Um the the plan was that, okay, let's start prototyping, brainstorming, building a few things, see what sticks and take it, take it from there. And that's basically what we did. So we left Meta, Devi and I left Meta April 2024.

Um and we started writing up a few ideas, discussing it, we built a few prototypes. And very early on, we had written down this idea of uh of a an AI concierge um that could handle a few like digital errands or a few chores on your on your behalf.

And we had kind of moved on from it to other ideas. Um but we kept coming back to that one um from different angles.

Um and initial bit of context here that might be useful is that like um all three of us in some in some capacities care quite a bit about like are very into like productivity tools.

Um and so like anything that can take um uh mundane work off our plates or like any tools that helps increase our productivity. Um, yeah, we've been sort of experimenting with different kinds of tools and methods over the over the years.

Like Devi has a popular blog post on a very opinionated blog post on how one should use their calendar. Um, that went viral and she had written it out, I think several years back.

Uh, I had built a uh a very simple sort of note taking app, which was also like pretty opinionated on like the the sort of affordances it offers versus not. Um, that that was it's almost 10 years old at this point.

Uh, yeah, so we had uh to varying degrees, we've we've delved into like productivity tools. So that intersecting with our background in agents, um sort of eventually evolved into what we're going after with with Yutori.

Um, so with Yutori, what we're building are basically uh like you said, AI agents that can do everyday digital tasks on the web. Um, so anything that we currently open a web browser for that one would broadly categorize as um digital labor.

Um, so if one category of things that we open a browser for are entertainment, right? like to watch a show on Netflix, let's say, right? Uh, but this is not that.

This is if you are trying to get a task done um on the web, um it feels like in the future, five to 10 years from now, how we get that done is going to look very different from how what it looks like today.

We may no longer be manually clicking buttons, navigating web pages, scrolling on web web pages, have like 100 tabs open.

We may be operating at a slightly higher level of abstraction where we're talking to AI agents. um, and agents are the primary drivers of action on the on the web.

And so our North Star is to push on that frontier and make that vision sort of a sort of a reality.

Host: So, in 20, um, 19 or so, like, uh, I wrote this blog post called like the push versus pull model. Um, where you're sort of like, your brain is hijacked by push notifications. like everything is pushed to you.

You know, which song you want to listen to, Spotify recommends to you. Like YouTube will recommend to you which thing to watch next. Um, and you're you're sort of like browsing for anything. Everything is pushed to you.

Uh, so you're constantly juggling with uh content that is being pushed to you. And people and that's what like you see like the choices people make are actually being reduced.

Uh in some form like personalities are also becoming, if you had like thousands of different personalities as human beings, now we have hundreds of them because Yeah.

Everyone is sort of like mimicking each other because this they're fed by same content, right?

Um, and that's why you also see increasing in polarization also because like it's easy to put in three buckets as a person rather than in 300 or 400 because people used to consume things differently based on that curiosity.

But now people consume based on what is being pushed to you. Uh, so my whole point was like, how do you have a pull system? How do you like uh go from, you know, someone else telling you versus like you follow your intuition, your curiosity?

Uh, but the tooling was only there for if you can build things. If you can write scripts for it, if you can, you know, hijack, you know, or put together a bunch of tools together if you're a software developer, you could do that.

Um, or it's also like um there's uh element of uh can you pay for stuff? Can you avoid the ads? Can you avoid the recommendation systems? Right? Like uh can you get like YouTube subscription where you don't see any ads?

Um, so there's almost a certain segment of population who has evolved above that thing where they'll never see an ad uh because they have all the subscriptions. Uh, right?

So right now what happens is uh, I don't have to be on Twitter to know what is happening. I can just write an agent that is going and telling me, hey, finally top 10 uh important AI tweets, um, that I want to know from, you know, XYZ researchers.

And I can have a scout or I can have deploy another type of agent. I can, you know, write a Python code. I I can just code this up and or ask Replet to write, do something. Uh, right?

So I think the interesting thing, the point I really trying to make is you go from that someone else telling you to what to consume, uh because you're in that web interaction system to you're going towards where you can, you have more control about what you want to consume and what you want to stay on top of.

Um, and I think Scouts really, you know, helps us do that. So, I think before we jump ahead of the conversation, I think it'll be good to sort of like um give a small demo of how Scouts work and then uh we can continue the discussion from there.

Guest: Sounds sounds great.

Host: And the push versus pull model, yeah, definitely resonates.

Guest: Okay. So hopefully you can you can see my screen.

Host: Yes.

Guest: Um This is what the the homepage of uh Scouts looks like. Um, as context, Scouts is basically agents that can monitor the web um for anything you care about so that you don't have to.

So that you don't have to sit there refreshing tabs and monitoring various things. Um, where like you can you set up agents. Um, they'll monitor and they'll notify you whenever they find something, something relevant.

So I'm just going to go ahead and um set up a few few scouts. Um, so let's say, anytime someone mentions or reviews Scouts by Yutori on Twitter, LinkedIn, Reddit, etc.

Host: That's actually one of the most useful things I found while using because you've some you frame something and then realize, oh, this is actually better framing. So it's actually a very good uh product feature.

Guest: Glad glad to hear that. Yeah. Um and then the second thing it asks for is, okay, how how frequently do you want to be notified about it? Do you want a daily digest, a weekly digest, or whenever there's a new mention or review of scouts?

Um so for now, I'll I'll just pick whenever. Um I'm actually going to go ahead and set up a few few more scouts, just so we have we have all of those running.

Um so let's say whenever there's a new um browser or computer use model, um especially open source releases.

Host: Um, that's another thing that's top of mind. So I set that up. Um It can also um do a good job of like monitoring um like product prices or reservations. So let's say a lunch slot for two at this restaurant called Copro in SF, which is an Indian restaurant that I really like. Um, for Saturday or Sunday.

Guest: And there's a bunch of examples here. Um, just to sort of solve solve that cold start uh problem uh that gives you gives you ideas for what to scout for.

And there's a bunch of like existing scouts uh that we have kind of curated uh that give you a flavor for what the product can do. So once we have these scouts set up, um that's it. You don't have to do anything else um uh manually.

It's basically set up, it's it's on. Um, one thing the UI lets you do is you can see behind the scenes what it's doing um to to generate sort of the first report for that. So here, this was my original query.

Anytime someone mentions or review scouts by Yutori. Um, you can look at, okay, there's a few sub agents, there's a social media sub agent that's looking at right now, Twitter, um, YouTube, Reddit.

Um, there's a there's another sub agent here that's looking at that's doing a Google search to look for mentions of scouts. Um, so it's going to do its thing. It takes a little bit of time.

Um, whenever it's done, it it it's going to send a report over email.

Host: And I just want when I started using uh the Scouts, uh, there was no this UI. I think all it was doing was, uh, it was just uh deployed and you see progress or it's done and you get an email when it's done.

But this view actually is very interesting because we also get that report of like how many agents were deployed, how much time it was saved.

And that was also very really useful uh in terms of like quantifying how good of a scout you deployed uh or not because like, okay, is it worth my time to have this scout or not? Because that time saving aspect of it was also an interesting feature.

Guest: Yeah, exactly. Uh, in this UI, this this live stream UI, we shipped recently, um, so it makes sense that at the last time you used it uh you mean I.

Host: So behind the scenes, let's talk a little bit about, you know, what's happening behind the scenes before like the scout does the job, right? So how does the technology work behind the scenes?

Is like, do you have your own version of the web or you're just plainly deploying virtual agents, uh, you know, in the cloud that are going and literally browsing it? Or what is a type of approaches that you're taking here?

Guest: Yeah, exactly. Um, so there's a there's a um there's a few things, few few different things that are happening under the hood. And this view kind of already gives you a glimpse into it.

But there's more than 100 MCP tools that the agents uh are using under the hood. Um, many of these are happening in parallel. Like you can see here, there's three sub agents working in parallel.

Um, and for the long tail of the web where uh so the so the long tail of websites that don't have APIs or MCPs where you like and you need an agent or a person to actually uh browse or navigate that website, uh click through pages to get to the right piece of information, we have our own in-house navigator agent, uh, which is basically doing that.

So if you if I if I skip to the back here and I look at the look at what it's doing, it's clicking through various web pages to get to the right piece of information, um, so that it could it can report it back to me.

Uh, all of this agent orchestration um is built in house. The the navigator model that powers this sub agent is also built in house.

Uh, we are in fact currently state of the art on uh several browser use benchmarks compared to like Anthropic, Gemini, etc. Uh, but that's basically how it works uh under the hood.

Host: Yeah, I was trying to, I don't know if you know, uh Prarag Agrawal, um, he's the ex Twitter CEO and uh if he founded a company called Parlan.

And I was having a conversation with him at an event uh in Seattle and he was mentioning that they basically recreated the search indexes.

Um, sort of like what Google has or what Bing has, uh, for deploying their agent for for their agents to work on. And they continuously update their indexes so that they always have a snapshot of the web.

So I was curious if you guys are having a similar approach or is it more is it not the direction that you're going in?

Guest: Yeah, um, so a couple of thoughts on that. One is that for monitoring specifically where you're looking for new information, um, an index is less useful.

Um, for retrieving historical data or uh like answers to questions about historical data, um, an index is super, super useful uh, because it can make things very more efficient and you don't need to start from first principles every single time.

But for monitoring, it's less useful. And second is that like that's not necessarily where we're looking to differentiate on. So like I said, this uses a bunch of like MCPs and APIs under the hood.

We're we all as you can see here, it's doing a Google, it's it's doing a call to Google search API. Um, we use a bunch of other search APIs as well. Um, so we're happy to use whichever APIs return the best results.

The the important bit here is to make this agentic so that it retrieves a set of results, um, it reasons about it, decides if it if it found the relevant piece of information or not and then decided, okay, what to do next.

And this is happening in sort of a for loop till it gets to the right piece of information, including actually navigating websites in case search APIs are insufficient for certain queries.

Host: So if so it is it like a scout can be run per day or per week, right? Or like whenever a change happens. Uh like navigator is one way you're doing that. And then you're this is researcher. What does the researcher do?

And how is it different from navigator? Navigator is like a little browser instance where you're going and looking at it, just like a human would. How is how is the researcher different?

Guest: Yeah, these are just uh a few different sub agents um sub agents that we've we've kind of um uh set up.

So I yeah, the researcher sub agent, for example, is calling a bunch of different search APIs um to to see if it can find relevant information about the about the particular query. Um, you'll see yeah, depending on the query, you may see more.

I I think we saw the social media manager sub agent as well. Um, yeah, so that's scoped to look at um social media websites in in particular.

Um, so each sub agent here has a very well scoped narrow set of tasks and websites and parts of the web that it needs to look at, um, including a mix of like APIs or or navigators, right?

Host: I think let's let's go into like one of the Scouts that you have that already worked and uh we can just This is yeah, this is another account where I have a bunch of scouts set up.

Um, we can look at what the what the findings view looks like. Uh so the findings view is where all the reports from all my scouts are aggregated in one place. Um, so here's an example report.

So this one is for like top rated PS5 games under 20 bucks. Um, it found that, okay, there's currently a discount running on this game Saints Road 4. Uh, there's links here that we can we can click at it, click on and look at the look at the listing.

Here's another example. So this scout was looking for any new news about uh health like healthcare and AI. Um, so it found a couple of reports about this new announcement from Samsung India.

So this would basically yeah, this is this is basically what it's what it's doing.

Like you can specify a bunch of different signals that you care about, um, and it will scour the web to find relevant information for that particular query and generate reports that look like this.

Host: Got it. Uh, and when um when you're running it on a daily basis or on a weekly basis, does the agent repeat what it has done? Like let's say the first time I ask a question, it it does it for the first time, right? Is it replicating the same thing or is it basically doing it from first principles again?

Guest: It it relies, it has some memory of the work it did previously. So the work it did previously helps inform subsequent runs. Um and that's useful to make sure that um the user is not getting a duplicate um or stale piece of information.

Um and it also helps inform which websites may have the right piece of information.

So like for example, for the for the lunch slot query, uh, if I'm asking about a particular restaurant, next time it doesn't need to start from scratch to find that restaurant's websites.

It can shortcut that to some extent to start on the on the restaurant reservation page. Um, yeah.

Host: Um, I think yeah, that that's a pretty good demo and I've been playing around with it and I've like, I think I tested 12 to 20 scouts.

Um, and I I can honestly say it saved me a lot of time and it also allowed me to focus on things because what happens is like, you know, this is important, um, but you search for it, you look at it, then you forget it.

And then after 10 days, you remember that, oh, this was an important thing I need to follow up. Um, and I need to again go to that website and see if anything changed.

And that could be like, you know, looking at a red fin if you're buying a house or uh, if you want to sell your house and you're looking at like, okay, what are the sales that are happening around your house, right?

That you can just set up a scout to say, hey, look at all the sales that are that are happening in your county.

Um, and I think the this the other scout I was uh trying is uh I have a Microsoft stack in terms of work, so I wanted to like get productivity tips on a weekly basis and send me an email, right?

So every Monday morning I get a productivity email uh about like, okay, how to use the latest tools to see, okay, and then then I can see and pick up, okay, what is useful and what is not. Um Yeah.

Guest: Yeah, thanks for that demo. Yeah, uh a couple of the scouts that we had just set up um got back with the reports. Um, so we quickly share that. Like this is the Copra one. It got back with a few different slots that are available. There's a link here that we can click on to go ahead with the booking. Um, and the open source release as well.

Host: Yeah, this is super cool.

So one obvious next step that I thought, um, and I I wanted you to let me know like which product direction you're going is but, okay, I have the scouts thing, but I also want to move that information into something else or do more with it.

Like, for example, okay, how let's say, um, I'm trying to grow on Twitter, for example. Uh, I got like top 10 viral tweets in AI. Uh, right? My next step is going to be either I order recote it or reply to it.

So there's a amount of action to it or just I want to do a further summary of it uh, and put it in my newsletter. Right?

Like there are now more things uh, or in some ways, you could say scouts is mostly now reading and now you took that information, you synthesized it, uh, and then you want to use that information to do some action.

Uh, and that doesn't necessarily have to be in control of Yutori or Scouts, but something that I can control. Um, so do you guys have any plans to sort of extend the product in that direction?

Guest: Yeah, so it's it's it's some version of that is coming.

Um, like we're going to uh provide integrations for um either looking at your existing context from your other apps or exporting scout reports to your other apps where you might this might be one step in a multi-step workflow or some longer longer task you're looking to get done.

Um, and there's going to be two main surfaces where we provide that. We provide that in the Scouts product in some form, some uh yeah, we provide that in the Scouts product in some form.

Uh, but we also have an API, which is optimal for like uh developers to have maximum flexibility of like how they want to, however they want to use the results of the of these scouts.

Host: And I when looking at the sub agents that you uh like once that sub agent started showing up, I also feel like the sub agents are itself a product in some form. Right? Like for developers who are trying to do like different things, like it's almost like the sub agent itself can be a product.

Guest: Yeah, and and we we get that ask uh quite often from developers especially where like they they know for certain kinds of queries, they don't want the full breadth of sub agents. There's like they want a specific kind of sub agent.

Um, so how we like it is on our minds to like um work with them, figure out, okay, what is the best way to offer that flexibility and control um to configure various various workflows while still using our sort of agent orchestration.

Host: Because at some point, it's also becomes like once you scale and once more and more users started using, like it also becomes like, okay, what is the cost of um uh the task's value to me as a user versus what is the cost that is you guys are incurring.

Are you guys spending more time and more agents when I actually don't want that much?

And that will be an interesting uh dynamic that will also play out like where like you know, if I know like you have like a fixed amount of subscription uh fee per month, but I could give a really complicated scout and, you know, exhaust my $15 incredibly two scouts.

For you as a business as all like that would be completely negative.

Guest: Yeah. Yeah, and exactly. I think that's spot on.

Uh, we actually have a blog about our agent orchestration and what that looks like and why it looks the way it does, which hits uh touches on this point exactly that like um one of the main reasons that we had to break this up or like there were multiple reasons why we had to have the sub agent architecture uh that powers scouts.

But one of them was also cost. That like if you have a single agent that has like all the context from everything that it's doing, that's expensive on like a per token basis at every step.

Um, whereas if you just split it into different sub agents where each sub agent now has a very specific role and much smaller context, uh, then just in terms of raw number of tokens you might be spending, um, it's way cheaper.

It also keeps the context clean. Like there's way less irrelevant information in the agent's context, so BLM gets confused way less. Um, so there's many reasons for why that architecture makes sense, but cost is cost is one of them.

Host: Have you uh I mean, obviously you guys are thinking about um agents much more deeply on a day-to-day basis. Like how does the future look?

Because we started with, you know, more of uh I call it like agent workflows where, you know, the Zapiers of the world. And I use heavily an app called Relay, um that we featured on the podcast.

Like Relay is amazing because it's sort of like Zapier, but on steroids. Um, like once you add LLMs on top of it and make the interactions and integrations really good.

Uh, and the product, basically it's more about how they design the product, uh, than just pure capability.

Um, so that's why I use it a lot for and so most of the productivity IC is coming right now for most people is at that workflow, agentic workflow level. Uh and deep research is obviously really good. That's one type of agent you can think of.

And then there's like the browser agents, which is sort of like deep research, I'm assuming is doing behind the scenes some kind of browser agents.

Um, what are the other form factors that we can expect to see or you know, have ideas about that might come?

Guest: Yeah, I I I think one vector where um we're pushing on and Scouts is like a baby step in that direction right now, but there's going to be more coming is productivity in particular.

Um, so most of these agent products and GPT, Cloud, etc. all included, uh are largely reactive. Like you come in, you enter a query and then they do something, right?

Um And part of the part of the reason why they've been reactive so far is because of reliability concerns.

Like if these agents aren't um accurate enough, reliable enough, then asking the user to give up control is just too big and risky of an ask, right? Um The user would want to be in the loop um if these agents aren't aren't good enough uh yet.

But we are kind of getting to the point where uh they can do more. They can oversee your um context proactively and look out for things on your behalf uh that you may not have the bandwidth or interest uh for.

Um, and they can do a pretty good job for many such tasks. So I think what I'm looking yeah, from us from others in the ecosystem, there's going to be much more in the vein of productivity. Um, that's that's yet to come.

Host: Yeah, I remember I think Arwin from Glean is also been mentioning the productivity will be the next thing probably in the next couple of years. Right?

If you like people who are doing at the cutting edge developing products versus like people who are using it have slightly different opinion.

Uh, because it's I think I am a little bit on the cutting edge in terms of because I try a lot of new products. Uh, but I think people generally are not using that many of the AI tools. I think they're not crossed like deep research.

Um, and like if I look at like enterprise adoption or even just a regular uh consumer level adoption. What do you think that is?

Guest: I think uh on the consumer front, um, this mind share is hard.

Like even even today if you um um go just in the US, for example, ignore the coasts and and go uh in in middle America and like ask which AI tools people are using, at best chat GPT is probably the only one that's on people's minds, right? Yeah.

Consumer mind share is is just hard. It takes a while for uh that stickiness to that sort of percolation um to happen. And yeah, that's that's probably the the main reason.

A part of it may also a secondary but uh important reason might also be just the ease of using some of these products.

Like if uh yeah, like it's one thing to have a universal interface where you can come in, say anything you want and get a response back, which is like a very easy to use interface.

But if it's like a somewhat complex agent builder um or workflow creation like tool, that's just too hard for a large set of consumers, right?

Um, so part of the challenge here becomes, okay, how can we offer powerful technology, uh, but through very simple interfaces?

Like, for example, if I could um shoot a text to an agent and tell it that like uh my grandmother's out at the grocery market shopping. Um, check with her whenever she's done and call and hober for code so that she can come back home.

That's not that's a very simple ask, that's not possible to do today, right? Uh, because that kind of multi-app orchestration, um, there's some parts that may be automateable, but not fully.

Uh And it the interface being as simple as shooting a text, right?

Host: Yeah. Yeah. Like it is soluble, but it doesn't exist today.

Guest: Yeah, I think competing in consumers, you know, it's a tough uh beast in some sense.

Uh I mean B2B has its own challenges, but I think it's easier to convince of value and mind share wise because you can convince a couple of people in the company and then they will adopt you and actually try and give you some audience.

I think consumer is especially challenging and that's where like even perplexity in somehow I think has this problem of like not crossing that chasm of early users.

Like if you in the course of areas, people who are tech savvy, they use perplexity, but you know, chat GPT is directly competing with them and and normal consumer who's not tech savvy cannot see the difference on why they should not they have to switch from chat GPT to perlexity and especially when the whole memory layer is now uh sort of becoming the moat uh because chat GPT knows me.

Chat GPT there are so many interactions and once the memory is enabled, it's easier to just ask.

Like I'm filling up a questionnaire, let's say and they'll ask me uh you know, hey, give me a why you're good at particular this and I'll tell chat GPT use these previous experiences that I have and write a letter because I if I'm doing it on a new app, now I have to give them that context of what projects I worked.

Uh but ChatGPT knows all the projects I've worked on.

I it has a couple of documents or even my resume that I've already worked on, uh which has the details and that memory is so important and I I think that actually is the real mode in consumer uh right now.

Guest: Yeah, for us, um like a lot of consumers who are who come from non-tech backgrounds, um who either we've been, yeah, like somehow they've been able to discover our product. um or we'd be able to reach them. They love the simplicity of it.

Like they do like how simple it is to set up uh while still being quite powerful under the hood and like once they start getting their first, second, third report, it kind of clicks that, okay, so this is how to best use the product or this is what the product does.

Host: Yeah.

Guest: Uh and that kind of starts clicking for them. We're seeing a small but decent amount of uh interest from um uh like enterprise customers as well.

Like especially sales and marketing teams, um, who basically want to, their top of the funnel work basically looks like monitoring for various signals uh happening around um to decide who to do outreach to.

Like it can be as simple as, okay, like if a company is announcing a series A, um, it's probably indicative that like they're post PMF, they're looking to scale, might be a good time to reach out to them and setting up a scout seems to sort of fit in very nicely for that particular particular use case.

So there have been a few people who've kind of organically discovered us, um, thought this was a great fit for that use case, told their team about it and then uh we ended up signing on and onboarding them as like enterprise uh contracts.

Uh, but yeah, like that's kind of what the what the spectrum of users looks like for for us.

Host: Yeah, that that's probably what I also figured out as well. Like if I want to monitor competition, if I also monitor for opportunities, right? Like even for your own career, you want to do different things. Let's say you want to write for a website or uh you want to be a contributing author to someone or uh you want to go on a podcast and see who are putting out um,

---