Prajwal Prakash, Clueso: YC, Clueso and gold medals in math

Welcome to another edition of Founder Wisdom with Rajal Praash, aka PJ Praash. He is the co-founder of Cluso, previously co-founder of DeskLamp . Cluso is a SaaS project from YC that we will talk about today.

This podcast is brought to you by podpire.com. If you want to start, scale, or be a guest on a podcast like this, head over to podpire.com. We help monetize podcasting.

PJ, welcome to the podcast. Tell us a little more about yourself and what you do now. Thank you very much for inviting me, Charles.

Yes, it's nice to be here. I think you gave a pretty good presentation . The only thing that was missed was that I am actually very young.

I started my first company, DeskLamp, back in college, and it was essentially a PDF reader. It started because I had nothing to do during the COVID pandemic . The three of us got together and simply created something that would help us learn in the new environment of remote work.

This was the birth of DeskLamp. We created it for ourselves, and then the project began to scale at our college. We connected neighboring colleges, expanded it there, and it became popular in India—that's how we got into YC.

I don't think we seriously thought about monetization back then. How to turn this into a business? We were just a group of students writing code and came across a product that was evolving on its own.

We were very excited at that moment. I decided, " Man, I really like creating products ." I like working with friends. I like to communicate with users.

Let's handle this professionally. Then , when we got to YC, the partners said, " You're cool, but your idea sucks because no one in the world will pay for a PDF reader." That was the beginning and end of DeskLamp , because we were pretty stubborn. We thought we would be able to earn at least some income.

But what YC helped me realize and what I want to talk about in this podcast is this: if you're going to spend your whole day, night, and whole soul on a business, it has to be something that can scale. Because if you're going to dedicate so much time to it anyway, you might as well at least try to achieve a great result. So they really hammered this mentality into our heads : play big or go home.

This is how the transition to Cluso began. So, Cluso is a company that I currently co-founded and work for full- time. We create product videos and documentation using AI.

Have you ever used Loom, Charles? Of course. If you're into podcasts, then Vidyard, Loom—something like that.

Yes. Yes, of course. These tools are great for recording your screen, yourself, and sharing that content.

Well, yes. We decided to occupy a niche when we realized that many of these recordings were simply from the screen. You yourself are not very important there, and it was really difficult to make these recordings of high quality .

We discovered this when we were scaling Desklamp and wanted to create demo videos . We wanted to tell customers how to use the product, but creating such content took a lot of time—it took a whole week to make one good video. I'm sure you understand how much goes into making a video that's truly high-quality, right?

This is where Cluso began. We set a task: can we turn a rough screen recording into a high-quality result that looks like a professional designer worked on it for a week and the voice was recorded by a real pro? This is actually where Cluso started.

So, to summarize: Cluso takes as input a screen recording, such as a Loom video, that you make as an account manager or marketer at a SaaS company, or as a trainer for internal tools. You just write a draft in Loom, as simple as possible, just explain everything as if to a friend. And Cluso uses AI to turn this into a cool studio-quality video, replaces the audio track, and even creates written documentation.

This, in fact, is the essence of what Cluso does. I would like to discuss with you what you think might be interesting for people on this journey. Well, you started with an interesting point.

I think this is a question I often ask myself. You can work in the same niche for decades, offering a very specific product that will never pay off. I don't know, for example, about being a virtual assistant or, say, having a VA agency, right?

How to understand, for example, how little money they have, or, um, how small their TAM is, or how strong the competition is. How do you determine the best product-market fit that can ultimately lead to the creation of a decacorn? Yes, that's a great question.

So, I think it gets a little easier in the end, right? It's easy when you realize you've hit the target, because your product is flying away faster than you can produce it. You are scaling at an incredible growth rate.

The most popular example is how Facebook spread at one time. So it's easy, and I think even the downside is pretty clear, when your statement of the problem —I mean, just at the level of the problem—does it really elicit a strong response? I think there are some exceptions to this rule in terms of how responsive the problem is, especially for companies that create markets.

But for the most part, I think we realized, for example, in our story with Deskclamp, that the statement about the problem of students who really have learning difficulties was not entirely true . Our tool was a better way to do it, it increased their productivity, it gave better results, for sure, but that resonance of the problem with your customer segment—that, I think, is really the key. Because your solution will change, it will iterate.

But I believe that intuition and confirmation from customers that the problem is real is a good indicator in the early stages. Of course , this changes as your solution grows and evolves. You will understand that there may be some barriers to solving this problem, or gain a deeper understanding that there may be a reason why there is no money in this problem .

But at a basic level, that's where it all starts. But how did you decide in your case that Cluso was a billion- dollar problem worth solving? Yes, that's a great question.

For us, we were actually stuck in "pivot hell" for about a month when, you know, we decided, "Okay, we're moving on from Deskclamp," how do we figure out what the next idea is going to be? And we tried; again , as I said, we are a very young team, so we don't have much experience in solving specific industry problems. So first we tried, you know, things in the software realm, like code documentation.

So we talked to a bunch of engineers. We experimented with a few of these ideas, and then I think we talked to Michael Sell from YC, and he gave us some really good advice: when you're in the early stages and you don't know what to do, find one client who will pay you for something , and start with that. This is exactly the indicator that you should pay attention to at the start.

In our case, we found another SaaS company in our set that said, “ Hey, if you solve this, I’ll pay you $200 a month, no questions asked.” So we actually created it over a weekend. I stayed up all night and created the first version, which could already be of some use. It was just a Chrome extension, without even a full-fledged web page, but it was valuable to them because they really needed to publish content.

So, you start with this. You try to find signals that support your idea. But I'm sure that in the beginning you won't be able to learn anything without having something to show your clients.

The problem is important, but it is very difficult to assess in theory alone. You need to have some kind of solution, at least a hint of something that you can show the client to understand the situation. Are we on the right path?

Is there any relevance of the product to the problem? Honestly, we never estimated whether it would become a " decacorn." Everything evolved over time: we just kept working, and the use cases naturally expanded. We started with SaaS documentation, then we started reaching out to non-SaaS companies and asking, “ Can we use this for internal training?” and that expanded the market significantly.

So, to be honest , even now I don't know how big a market it is, and if you talk to me in a year, I might have completely different views. But this is roughly how our path went. And again, once you start building a product, if you're open to new things, you'll experience expanding markets, you'll experience adjacent niches, small changes in the product, or the emergence of a different user or a new problem for the same user— all of these things evolve.

Of course, how many clients do you have now and do you plan to attract investments this year? No. We raised investment through YC.

We are backed by YC. We have a very compact team. We are based in India.

So we don't have an acute shortage of cash. Honestly, that was a big motivation for being based in India. So no pressure.

We now have over 70 clients. The last three or four months have been just crazy. July to November was a bit difficult because we were still building the product.

We tried to determine which aspect of the customer education problem for SaaS companies really works. By November, the video format had become key—it was the issue that resonated the most because it took up a lot of time. So we focused on that and started adding more features.

Now we have a full-fledged video editor in the product. This helped a lot. This is another one how many customers We have 74, maybe 75 as we speak.

There's just one conversation going on at the same time. Good. That's pretty good.

Ovva. OK. So, 75.

And what is your goal for this year? At the end of the year, let's say 150. Honestly, I don't even look at the number of clients.

I want to increase the size of the contracts. This is a big question. This is where the question of market valuation arises, right?

At least now most of our clients are B2B SaaS companies, and there aren't that many of them. At least those who are big enough and willing to pay. So I'm trying to figure out how to add value and increase the size of the contracts.

For me, the goal is whether I can sign significantly larger contracts with these 75 clients. Now our average check is about 3 thousand. I want it to be much taller.

3 thousand a year or so. Yes. Per year.

Of course. And how do you plan to sell it additionally? Is this a licensing model?

How does it work? Yes. These are licenses and fees for use.

The number of videos exported becomes the metric by which fees are charged . This is quite understandable. Think of it as API tokens or credits for export minutes.

But we are also looking at ways to bundle additional features and charge for them separately. For example, something we are very passionate about is language translation. Let's say your team creates content in English.

It is very difficult to adapt it into, say, German or Spanish. This requires a separate team, and it is almost impossible to make a video on your own . So , now with Cluso you get exactly the same thing in German with one click.

So we're looking at that, and I think there's still a lot of work to be done on the business model. How do you make it fair for both parties and scale naturally? Well, it's a little complicated, but we're trying to figure it out.

And who are you targeting? Positions, industries, verticals, incomes. Yes, our main bread and butter now is B2B, SaaS, after Round A.

That's usually when they hire a customer success manager. This is a great fit because when a customer success manager comes in, the first thing they ask is: how are we supporting our customers right now? This is just terrible.

This usually happens between the seed round and round A stage. This is the best target audience. But we see what happens when you become an even bigger SaaS company, like in a D or E round: you have separate customer education teams that just do videos.

So we are trying to reach this level as well. The tools they use are a little different there, their priorities are different, and that's where the big contracts come in. In addition, there are many interesting incoming requests, for example from large companies regarding internal training.

We are investigating this in parallel, but we are not paying special attention to it and are not actively looking for such clients. Yes, I see potential in Cluso. Just curious, do people record videos without voice?

I thought when you recorded the video you added your voice and instructions to it. So how does it work? Essentially, any voice I use, I simply provide as input to the AI.

So, I record a video, then a big language model corrects the transcript , removes grammatical errors, and then it's all generated through TTS. So the end result is a cleaner version of everything I'm saying. At the same time, it retains the same content that I was talking about.

Of course, I can edit it later, but with AI...I can show you the video. Yes, let's do it a little later, I've already seen it on my screen. I see potential in this , you know, for example, for me— I know this is not your niche, but it could potentially be, for example, hosting courses, right?

For example, if I'm recording myself, let's say, to sell a cold email course, I would record everything in one go, and this AI could just correct everything, well, when I say "uh-huh" or make a mistake, the AI ​​will detect it. He will tell me, “Oh, there is a potential error here. Do you want to delete it?

Then I click "yes." But yeah, let's show the audience a quick demonstration. Of course. Let me share my screen.

One second . Yes, let me set everything up. By the way, what are the plans for new features?

That's a great question. I think one thing that I'm really excited about is automatic content updates, because a lot of the requests we've heard are around content updates; When I record a video, it's very difficult to change one small part of it later. So, that's what we're working on.

OK. So for this quick demonstration, I'll take the example of writing, say, a function in Google Docs. The experience at the beginning is very similar to using Loom.

I'll just switch to my account. We've done this before. Oh, what a mess.

OK. Yes. The experience is very similar to using Loom.

You just share a tab and get started. And again, as I said, we'll do a very informal quick overview of what this feature is. So , let me begin.

OK. In this video, we'll go over how to enable notification settings for your Google Doc. So, first click on Tools and then go to Notifications settings.

Here, in the pop-up window, simply select the option "added or removed content". Now you will receive notifications when something changes. Finally, click "OK." That's all.

You are finished. So, I just wrote this down very roughly and informally. I stuttered in mid- sentence.

This isn't something I'd want to share with customers, but let's see what Cluso does with it in a few seconds, shall we? The first thing you get is a complete written document that describes all these steps. So, we capture screenshots for each individual step.

These screenshots are automatically zoomed in. We also try to divide it into steps intelligently right away, but as you can see, the content is there, but it's not yet in a format that I can show it to the client. So just click on "AI enhancement".

This feature will rewrite the article just like a professional technical writer would . There are certain stylistic rules for how to write instructions, such as using verbs and bolding text. So, it's already done.

The text became clear. The goal of the article is to make it easy to scan so that the user can quickly understand what exactly needs to be done. This removes excess "water" and focuses on the essence.

Of course, if you want to make your article more interesting, we have many features for working with visual elements. You can create a GIF from any part of a video, edit screenshots, zoom in, zoom out, add blur, and more. In general, everything you need to create an article is right here.

If I want to add another frame from the video, I can just scroll through the video and add it. So, everything is in one place—the article is ready. Finally, we also create videos.

Let me give it a name, for example, " demo for charge". Perfectly. The system automatically adds a branded intro or ending depending on the client.

Because I do this on our account. There is the introduction of Cluo, the background, the ending, and, of course, the main feature — the sound itself, which we were talking about. Tell me, can you hear the video?

No. Oh, I do n't think it will work on Zoom, if I'm not mistaken. But try again.

Yes. Let me share my screen again. There seems to be a check mark there.

I keep forgetting to press it. Now do you see? Yes.

Let's take a look. Perfectly. If you listen to this now, it's still my "how to turn on notifications ".

My original sound, right? But on the left, as with the article, we have a full transcript of the text. So, as with the article, I'm going to improve this with AI.

But here the process works a little differently. This is not an attempt to make the text clear and formal. I'm trying to make the text more conversational so that it really fits into a video script.

So it will keep all your information but just correct the grammar. When the transcript looks good, I can choose the voice. In my case, I have clients in America and, to be honest , I wouldn't want them to know that I'm Indian.

You can choose one of these American voices, well, or, honestly, it depends on your preferences and customer base, whichever you're more comfortable with, and click "create ." So now Cluso will replace my audio track with this new inserted audio track. Yes? And if you think about it, that's one of my favorite parts of this product.

The idea of ​​replacing the audio track is actually very complicated because I also changed the lyrics. So I don't want to have a situation where the audio says " click on the tools" and the video says something completely different. This is the biggest problem with artificial voice acting right now.

Synchronization is such a difficult task. Cluso solves this automatically. Let me run this video for you.

I also added some background music, press "play ". In this video, we'll go over how to enable notification settings for your Google Doc. First, click on " Tools" and then go to " Notification Settings".

In this window, simply select the option for added or removed content. This way, you will receive notifications whenever something changes in your document. Finally, click " OK".

That's all. You are ready. That's pretty cool.

And who is responsible for creating all of this ? Now it's a team of three. I am the CTO, and my CEO is Akash.

Akash handles the design and interface, and I worked on the overall technology side. Did you mostly take functions from libraries or write most of the code from scratch? I would say that a lot of what makes the product cool is the AI API.

For example, the script improvement was done through OpenAI. Voice—via 11 Labs . Yes.

But the entire infrastructure around this is built from scratch. Even this video player is something we built from scratch. We also created a rendering pipeline on the server for synchronization from scratch.

All these things in articles, like screenshot management, all this rich text structure— we use ProseMirror, but everything else we built on top of it. So there's a lot of cool technology here. I mean, the way I see it is : there are a lot of cool AI technologies out there, but the actual process of creating videos from them is very complex, and that's where Cluso comes in handy .

Of course, have you ever thought about offering these services to companies, putting it all together, while spending some of your time? I wouldn't turn down 100 thousand for just 15 hours of work. This is an interesting approach.

I think, again, this is one of the philosophies that YC has really articulated very well. Don't forget that you are here for the long haul. You also do it because it's fun.

So I would only consider this option if it was a really interesting project that would benefit the entire team. But honestly, no, I think this time is better spent on the product. Of course.

Yes, I know that YC insists on this . Tell us about that gold medal you won at the best university in India. It was a crazy race.

So, what happened there? Yes, it was really cool and I feel it also reflects the changes in Indian culture. Actually, this gold medal is given to the student who has shown the best results over all four years of study.

It is usually a combination of academic success and extracurricular activities. And historically, these extracurricular activities were, how should I put it ? They are pretty clearly defined: interuniversity festivals, cultural competitions, sports, or, you know, internships—those kinds of things.

But I think I was one of the first times they really appreciated the fact that “you sat down and created something that people all over the country use; "This deserves some recognition." I really respect them for that. This was largely due to the fact that I studied computer science in India, which is incredibly difficult to get into due to competition. I don't want to brag.

If you google the acceptance rate for JEE, it's crazy , I think it's the toughest exam, except for Chinese. This is the hardest exam on planet Earth. And getting into computer science is incredibly difficult.

So there is an atmosphere in India where almost anyone is willing to hire you if you graduate from this program, and for extremely high money. I didn't want to do it because it didn't seem like something interesting, and luckily I didn't have any financial constraints that would have me in dire need of money at any moment . So I thought: let's just make the most of the opportunity we have .

And I'm really glad that the college actually recognized this, because usually public colleges aren't the best at embracing movements like entrepreneurship among students. So yeah, that's something I'm really proud of, and I'm really happy that some of my younger colleagues at that college are now starting their own projects, like one of them is running a drone company, it's just crazy. Three of them are creating a SaaS-type sales automation service.

So I'm really glad that it somehow started and it's continuing. Yes. If you want, meet some of these guys, and yes, I don't like most of the education system, but I know how hard it is to do it, and that's not bragging at all.

I understand how difficult and almost impossible it is, because I achieve high results myself. So yeah, well done, and it's a pretty unconventional decision—going to YC instead of getting 500k a year as a quant or something. Yes.

Yes. Yes. That's right .

But I'm very happy with it, and to be honest , it even confirms the correctness of the choice a little, albeit in a rather harsh way. When I talk to my classmates now , I think there's much more to life, right? Although we are in the business of entrepreneurship, and I talked about big results and all that , but honestly, it's the journey that brings the fun.

You do it because you enjoy creating a product, giving it to users, and it gives you a sense of purpose. I stay up until 3 a.m. every day because it's fun after all.

So yes , definitely. I'd be happy to introduce you to those younger students. I think you'll find it interesting to chat.

Thank you for that. And where can people learn more about you, BJ? Oh, yeah, I'm not very good at social media.

Honestly, I think my LinkedIn is the only place where anything is happening. So I would recommend taking a look at my LinkedIn. Be sure to pay attention to Loser.

If any of the viewers are founders of a SaaS company, be sure to write to us.