How We Raised $4.2 Million With No Product

Rabi Guha: Whether you are solving a problem that is hard enough, that is deep enough that actually creates value, and whether you have the right people to do it, that would be one half of the problem statement. Another half is obviously like, like in all exchanges and transactions between two sets of people. You need to find people on the other side who believe in you, who want this problem to be solved deeply enough and who are willing to take a bet on you, honestly. So it's a combination of both.

Sean Weisbrot: Is the co-founder of Thesis, a company that uses LLMs to generate UI and UX experiences for companies. I tested it out after we finished recording and it was really cool. In this episode, we talk about the company as well as their experience fundraising millions of dollars and living and working between teams and offices in the US in India. This is a very interesting conversation about fundraising and AI and ui ux. So if that's your thing, then you're in for a treat. Let's get to it. How does it feel to have successfully completed a fundraising round?

Rabi Guha: It just feels great. I mean, the time that you take two days, it can range anywhere from like, let's say four weeks to 12 weeks. It just feels way longer. But once you have done that, you turn back and think, okay, that was not so hard.

Sean Weisbrot: And how much did you raise this time?

Rabi Guha: We raised 4.2 million.

Sean Weisbrot: How long did it take you from I wanna raise to monies in the bank? I want to raise and we have to sign the term seat and everything is in place.

Rabi Guha: That took us close to like three months, but then it took another couple of months for the money to actually come in the bank. Due diligence and everything.

Sean Weisbrot: Fundraising is very hard. What can you attribute to your success in making it happen so fast?

Rabi Guha: Especially at our stage, the seed stage When we raised, we didn't have a product. We just had an idea and we wanted to bring this in front of people. It's a combination of two factors generally. One is your team, your founder, market fit, as they call it. Whether you are solving a problem that is hard enough, that is deep enough that actually creates value, and whether you have the right people to do it. That would be one half of the problem statement. Another half is obviously like, like in all exchanges and transactions between two sets of people. You need to find people on the other side who believe in you, who want this problem to be solved deeply enough. And who are willing to take a bet on you, honestly. So it's a combination of both.

Sean Weisbrot: Did you have any previous experience in startups that made it, they were more likely to trust you or did you have a personal network of investors or someone that knew investors that introduced you? Because three months is typically like a very, very short time to raise funds.

Rabi Guha: Almost all my life I've worked in startups, so I have that network. Even before this, I was the one of the earliest members in a. Uh, up and coming startup and before that I had a small, small startup of my own and we had friends and mentors. Back then, we didn't really do any fundraising for that, at least did not do any fundraising successfully, so that at least gave us a network. And going into fundraising this time, I kind of had a, a leverage that I knew what not to do. But on the other hand, like you said, uh, I have a good network. I have, thankfully, I have friends and mentors who knew a lot of vc. Warm intros work great. That's all I can tell you.

Sean Weisbrot: Having an idea and successfully raising money without building the product is, is very rare. Um, I am constantly on LinkedIn and Reddit and talking to startup founders. A lot of them are pre-seed. They maybe have an MVP, they may have people using the product already, and nobody's willing to touch their their product yet. Like the investors won't give them any time and. But they're first time founders. So I think that that might be it, is that you have that network and the network's like, oh, we know who you are. Yeah, we'll give you the money. That's fine.

Rabi Guha: Like both me and my co-founder have built multiple products over our career and, you know, the problem that we are tackling, uh, which is what thesis is about, it's something that we have worked very deeply for most of her career, if not all of her career. So yeah, we had that. Kind of trust that you can say within our network that, okay, these guys know what they're doing.

Sean Weisbrot: So if someone wants to successfully run a startup, do you think they would be better off holding off on their idea and just go work for a startup first to learn a little bit more about how they work?

Rabi Guha: Advisor? I generally give to a lot of juniors of mine from, uh, university that it's not a bad idea to go and work for a corporate for a few years because otherwise you don't really know how the world works. Like for. A lot of people jump into B2C products because they, they're easy to understand. Uh, you have been a consumer at some point of those kind of things. So you know what, you have a, you have a notion of what people are looking for, right? But at the same time, if you work in a industry, for example, even healthcare or med tech, et cetera, right? You will have a much deeper understanding of how the world works in their realm. So even when you are starting to think of what to build, how to build that, how problem, how to. Find your first, uh, few customers and drive value for them, you will be at a much more, uh, advantage compared to somebody like me, for example, who has never worked in healthcare. So yeah, I would definitely say it's not a bad idea to go work for a startup or even a big company in the early part of your career. But at the same time, uh, just like at some point you have to jump out and jumping out is going to be the hard part. You might as well do it sooner than.

Sean Weisbrot: Your team is based in India, I believe. Right. But you've also lived in the US too.

Rabi Guha: Yes, yes. So currently our team is mostly based out of India. We are starting our San Francisco office very soon, actually. We are hiring for that. And uh, yeah, uh, I have lived here for a part of my career. Actually, the earlier part of my career, I was with this company and I started working for them in Bangalore, but then I moved to US and lived here for one and a half years or so. That definitely helps. I'm more familiar with the area, I'm more familiar with how people think and how they'll approach problems and obviously this is the place where like hundreds of billion dollars company and like few trillion dollar companies are made. It definitely helps.

Sean Weisbrot: Were you trying to fundraise in India or in the US or from both sides? How did that work?

Rabi Guha: From both sides actually. Uh, the way we thought about it is, see, at the end of the day, even Indian investors right now have a very global portfolio. So their companies might have started in the, might have started in India, but they have a big US presence right now. Or vice versa too. Like a lot of the portfolio companies of our US investors have India offices now and, uh, if not, uh, India Market, but definitely India, r and d office. So this like global corridor has become very flexible and that definitely helps. And that's why when we set out, uh, to raise funds, we, like I said, we got a lot of warm intros and we went with that more than the geolocation of where they're actually based.

Sean Weisbrot: That's good. 'cause I've heard people complain that they felt as a, an underrepresented founder, someone who is not basically a white man. Um, that they're, they're from India or from other countries that, that, like Silicon Valley VCs look down on them. And again, it's probably because they're first time founders. If anything else,

Rabi Guha: I didn't personally feel this way. Uh, but there is obviously, I will be honest that there is a hint of hesitation among the Silicon Valley VCs for founders who are based out of India, but not so much like Indians. But uh, having said that, I mean we definitely got, uh, a first call everywhere. We had a great conversation. We learned even from people who did not end up investing or participating in our round, we learned a lot from them. They gave us honest feedback, so that part I definitely loved.

Sean Weisbrot: Did you feel pressure to open the San Francisco office because of that hesitation? Was that part of your deal?

Rabi Guha: There are no terms in our deal as such, but we were very sure that. San Francisco is the place where, you know, people work on cutting edge stuff, especially given that we are working with LLMs and we are working on making AI applications much more interactive and much more user friendly. And a lot of these AI applications are actually getting made here than anywhere else in the world. So it makes all the sense for us to open RSF office and start working closely with the people, our design partners, our early customers, our early adopters. But no pressure from either of our investors. Just that's something we wanted to do always. And yeah, it's the right time for us to start doing that now that we have a product out there.

Sean Weisbrot: The Reddit forums are very interesting with these colored stories of people like having these experiences. So I'm, I'm trying to, I guess, validate other people's hypotheses. Um, 'cause a lot of them are like, yeah, I just feel like these investors don't care. They're not interested. I've had people say, I had a product with users and revenue and VCs ignored me. I mean, that happened to me too. You know, I got ignored by VCs. They're like, come back when you have a million in revenue. And I'm like, yeah, but I, I just don't have enough cash to get there. Like, with your money, I can get way beyond that. Yep. Not my problem. Sorry. Come back when you have it solved. You know, so we, we get this, I, I see that kind of stuff all the time. So, uh, just, you know, I, I've, I've done my own angel investing, but VC is a different world. I've, I've raised from a, a, a VC for myself and I've raised from VCs for a few other companies, but. Um, I'm always fascinated by the way that they think, because I think it's just so interesting and, uh, and, and alien in a way.

Rabi Guha: A few factors definitely play a role. One of them being that you being able to convince them that this is a VC business since that this is, this has the potential to become a billion dollar company at some point. Even if it is like a 10% chance, that's what they're probably looking for, that a 10% chance of becoming a billion dollar company versus let's say a hundred percent chance of becoming a a hundred million dollar company. So, and I've, and though I have never really worked in this circle, from what I understand talking to VCs is that they are also not trying to burn bridges. So even if they have some feedback for you or they don't think you'll be able to take it positively, they'll hold it back. And so yeah, sometimes you come away from eating thinking that went great, but then something else is going on in their heads.

Sean Weisbrot: In your stage of running your business, what is the most important use of the money that you've raised being spent on,

Rabi Guha: since we are the seed stage right now that is split almost 50 50 between r and d and go to market. And for us, like again, different businesses are different. Not all businesses probably need a couple of million dollars to build their MVP or their first product, or their first, like even like $1 million in revenue. But for us, since we are building, uh, we are foundationally trying to build a model that can do ux and that has really like. There are some good examples out there. For example, Claude and other stuff, but there are no fundamental models that are just built on specifically doing UX and ux. Well, for us, a lot of the cost going training, gathering data, cutting tests. Obviously hiring is a big factor of it right now. It's very hard to hire good AI engineers. That's why 50% of it goes for our uh, r and d as such. And then 50% obviously GTM, because. Like that would be, I think the most, uh, that would be the best use of money at this stage. I guess once we build it, once we ship it, we need more people to hear about it. We need more people to start thinking about it. Okay. Like chat based applications. Agreed. Conversational applications. Agreed. But is that the best ux? Like is that you, is that how you want your users to use your app? Like for example, uh, let's say, let's take the case of shopping. That you have all the Amazons and Shopify ease and everything going on, right? And now suddenly people are thinking more and more of doing conversational AI based shopping experiences. But would as a customer, would I actually want to read through like thousand lines of text and figure out, oh, what are you saying? Or would I want to see a well curated display of, let's say, five products compared between them and being able to click a button that says, buy this now. So these kind of things are what we are actually, uh, what we actually want to get out in the market that, you know, it's not just enough to build a LLM based application. You really have to think about the ux. If you want your customers to love it and switch from their old SaaS based workflows, you have to bring to them the same level of engagement, the same level of, you know, user friendliness that they were used to.

Sean Weisbrot: Hey, just gimme 10 seconds of your time. I really appreciate you listening to the episode so far, and I hope you're loving it. And if you are, I would love to ask you to subscribe to the channel because what we do is a lot of work. And every week we bring you a new guest and a new story. And what we do requires so much love so that we can bring you something amazing. And every week we're trying really hard to get better guests. That have better stories and improve our ability to tell their stories. So your subscription lets the algorithm know that what we're doing is fantastic and no commitment. It's free to do. And if you don't like what we're doing later on, you can always unsubscribe. And either way, we would love a like if you don't feel like subscribing at this time. Thank you very much and we'll take you back. I was interviewing someone a few days ago. And he invested in and then joined full-time as an executive member of a CRM company. And they've been infusing a lot of AI into what they do. And he said that he believes the future is where we strip away all of the UI and the UX is all, uh, audio based. So you, you having a conversation with an AI. And AI helps you to solve your problems. Kind of like what we were seeing now with, uh, open AI's, more recent models that you can have like a voice conversation with. Um, and I don't know if you've heard of Bordy, but Bordy is another example of an AI that does a really great, uh, conversational experience through voice through You literally like call him on Skype. Well, I called him on Skype. I'm not in America, but you can like, call him on the phone and have a conversation with this AI and it feels like a damn human. There's memory in it so you can call again and it remembers who you are, if it remembers the phone number or if you send 'em an email or you send 'em a LinkedIn. Like it remembers all of those interactions. Um. It's such a fantastic experience. I mean, the, the AI ended up raising $8 million for itself without the human's feedback until it got time to like get serious about legal. And then it was like, oh, by the way, I've raised money for you. Like you need to sign the papers. And he's like, what? Like, that's how good this thing is for the conversational experience.

Rabi Guha: And voice is definitely one of those places where we will see a lot more human touch coming and it's already happening, right? Like I would much rather. Talk to a, a conversational voice-based agent, then having to go through a bulky IVR system where there are 10 steps, you enter numbers, it forgets you, it hangs up. Now you again call and you start the process all over. But having said that, I mean even, uh, like multimodal experiences beyond voice are also important, right? For example, do you actually want to shop on a phone? Would you be able to understand or would you be able to read? Uh, uh, sorry. Comprehend everything that, uh, AI is saying on voice, or would you much rather see it in a well structured tabular format? Consider your options, compare the price, look at the reviews from, let's say Amazon or Red, et cetera. So I think this whole world is also beyond voice.

Sean Weisbrot: I think that's a specific example. Talking about a shopping experience. Shopping is always gonna be visual.

Rabi Guha: You have this phone where you are, uh, adding your tasks to Jira. Let's say, just for the sake of this example. Right. And when you go in a scrum, the ideal case would be that you just talk it through and it updates it in the background. But every once in a while you need to check what are the things that are pending, what are the things that are prioritized, et cetera. Do you actually want to talk to a voice agent and explain it the entire thought process, or would you much rather just pull up your phone and the phone already knows that these are the top things for you? It displays it in a very natural and. I mean, I can keep on going basically, but, uh, every, every modality in this multimodal world is important. Like obviously videos are great for a certain part of it, like even shopping for example, videos might be a even better medium than just text, but videos, audio, uh, rich ui, everything is going to be just as important to make the AI more human than it is today.

Sean Weisbrot: With thesis, you're trying to help. Companies figure out their own UX journey. Right. And you're trying to use like text-based or, or voice-based Orvis visual agents to make them do that?

Rabi Guha: We are a generative UI company. What that means is that, again, this is not something we came up with. It has been talked about and written about a lot, but what we are saying is that CLMs agreed, but text is probably not the best representation of your data all the times. For example, let's take a practical example. You want analytics use cases, right? You want to drill into the history of the US population growth, understand the peaks and troughs, slice and dice it by gender, race, immigration, stuff like that. Right now, even today, you can go into charge pt.com and enter all this, like, tell me the population of us, how did it grow over the last 10 years? And it would keep on responding with text saying, okay, this is the population, this is a table with the list, et cetera, et cetera. But a much better UX for that is just displaying as a graph, and there has been like years of research in figuring out what kind of visualization make sense in what use cases. What we do is basically train the LLM to understand these design patterns, and instead of you having to manually encode them into your application or manually hardcode them into your application, you can just use RLLM to generate it on the fly. And just engage your users much better than you could do with a text-based chat interface.

Sean Weisbrot: I think I'd have to see it in action to, to better understand, 'cause you're talking about generating UI different for each user as they go in. But you, you have a single app that seems really funky. I'm not saying it's a bad idea. I, it is just difficult for me to picture, uh, without saying it. I think it's really interesting.

Rabi Guha: For now, you can just go to thesis and try out our canvas. And our canvas is basically the first step of this journey where instead of trying to do it on the fly, we are helping product managers and designers and engineers, everyone who's working on this, uh, cat based applications to visualize how it should look. So it's a text to design application where on a canvas you can feed it data that this is the data I want to represent. Can you show me how it should represent? We help you basically prototype it, develop it, and ship it to your users. So that's the version one. And stay tuned for version two. Within a month or so, we are going to release our version two, which will do what I just described. Instead of you having to think through all the processes, all the UIs, we can just generate it on the fly.

Sean Weisbrot: I feel like that's very heavy on server and gps, GPU usage.

Rabi Guha: GP is not so much, it's a little costlier, uh, than doing text for sure, but that's what we are building. Uh, that's why it's important to have your own models that are specifically tuned to be able to do that so that you bring the cost down and down and down over, uh, the next couple of years. And I'm sure at some point this will become a non, like, non brainer that you need to have this, otherwise it won't, your application just won't have the same.

Sean Weisbrot: Can you share a secret for developers about how they can lower their cost for the servers, for ai?

Rabi Guha: Multiple things. Definitely. The world of LLMs is changing so rapidly. Keep up with it. Newer models are dropping every day. Again, like don't always go with the benchmarks because the benchmarks are a very high level. They try to, uh, test things on a very broad level, on a very variety of topics. You need to figure out your benchmarks that matter to you. For example, for us, this matters versus that matters. And then you need to start running these models, uh, like start testing these models against that. That would be definitely like, by doing that, we have been able to save time and money because just. Newer LLMs sometimes that don't do so well in benchmarks, do much better for your use case. The second thing that you should definitely keep on doing is trying to distill models, and by distilling you can, basically, what you're trying to do is take the efficiency of a bigger model and bring it down to a smaller model at the same time. That's why I said the benchmarks are going to be super important, but these benchmarks are going to help you decide whether this is good enough. Balance between being good and being fast, and being cheap, cheap is.