Zero to 36 Million in 45 Days: The AI Agent Revolution

Speaker A: Welcome everybody. This is Ryan Staley. And I am back. I have a very special guest with me today. I have Wen Tsang, who is co founder of, uh, genspark AI. Something really cool about genspark and how they kind of came on my radar as you know, I test tools and evaluate them for my clients for a living. And I was blown away once I started using genspark. So I'm like, I gotta have. When on, um, the, on the, on the podcast. So when. Happy to have you on the show, man. I'm excited to have you here.

Speaker B: So. Thank you, Ryan. Excited to be here.

Speaker A: Yeah. I mean, like, here's what I would say is like I test tools on a continuous basis and I always evaluate, like, does the marketing match the outcomes? And what I'm saying it with your solution is it does. Right? And it's a multi agent, I guess I could say, and you correct me on this, man, but a multi agentic tool ranging all the way across from like your super agent to your AI podcast agent, to your sheets agent, your docs agent, your presentations agent. Right. I don't want to, I don't want to skip anybody. And then you have a general browser agent with all included within the same license. Is that accurate? Did I miss any of them?

Speaker B: Yeah, I mean, that's pretty much it. We launched our super agent suite, it's truly a suite, back in April. So it is built for knowledge workers and exactly as you said, Ryan. We have, uh, AI slides. You tell AI, what do you want? It'll build you a full presentation, um, AI sheets. You could, you know, ask for a waterfall financial model. It'll just get it done for you with formulas and everything. You can just populate the, uh, you can just, yeah, do, do the numbers and it'll just, uh, get you everything. Yeah. And we're, we're still early in this journey. We're still rolling out new stuff on daily basis. Yeah. I don't know if you saw, uh, we just released this short preview of the world's first AI meeting Notes on your Apple watch. It's not built by Apple, it's built by us. Yeah, yeah, yeah. It's coming out in a week or so. So this is like, uh.

Speaker A: Yeah, yeah, I gotta get plugged in, man. Like Eric. So Eric's your, your, your CEO co founder as well. Like, I really appreciate, like how fast you guys ship, different features and capabilities. So it's not a surprise to me that I, I missed one of them, you know, but that's pretty cool. With the The Apple meeting recording on the watch and is that LinkedIn to the actually give. Give us a quick, well, quick understanding of that real quick. And then um, I want to give a little bit more background on, on your organization and some details higher. I mean just for context, everyone like they've grown from 0 to 36 million in 45 days and this was as of mid May so they continue to grow at an insane rate. But tell us about the Apple Watch meeting function because I'm, I'm curious now for sure.

Speaker B: For example, for someone like me, I love meeting with people, right? But it's always a pain for me to you know, take notes. I always have my little notebook but now I don't have to do that anymore. Essentially with the uh, power of AI, you know, you can do speech to text, text to speech. So, so what we did is while a lot of the other folks may have figured this out for you know, online meetings, like you know what we're doing here, the uh, AI would just transcript everything. What we did is we basically put that power into your Apple Watch. So wherever you go you just do this and then it'll just start transcribing for you. That's just. Yeah, I mean after your meeting you just get your meeting us from jenspark. That's as simple as that. Yeah.

Speaker A: Wow, that's awesome. So is it like a mobile app then? Like that you got to download on your phone or your watch or combo? Both I assume. Yes.

Speaker B: And uh, you could just use it uh, from our uh, JetSpark application. Uh, you can download that today from the uh, App Store.

Speaker A: Nice.

Speaker B: In the back. It is a ah, browser technology. It supports all our agents. But this is one additional capability you can have. And the beautiful thing is it works not only uh, for in person meetings, it also works for online meetings just like this. Because my watch is here, right? It'll just you know. Yeah. Be there.

Speaker A: That's awesome man. Well, so why don't you, why don't we do this. So Wen's going to show us some things and I'm really excited to share with you because I'd like. I always recommend Genspark when I'm working with folks that did a 900% cohort mentioned it there a couple times and like folks just like to give you context. Like I've compared it side by side versus manuscript versus agent mobile with OpenAI and Genspark does an amazing job. So Wen, like big picture I guess. What's the vision for the company and where do you see this vibe Working, uh, kind of vision that I've heard Eric mentioned. I haven't really talked to you about you, but for the company, where do you see it going and what are you most excited about?

Speaker B: Yeah, absolutely, Ryan. So look, everybody's talking about how AI is going to take people's jobs, right? We simply don't see the world that way. Our view is very simple. AI is happening. Yes, we wanted to help everybody, most importantly, the knowledge workers harness the power of AI and really elevate each of us to unleash our potential. More specifically, what we mean by that is there's a lot of grunt work in our daily life. Right back in the days when I was running my last company. When I go to pitch, a, um, investor always have to craft that pitch deck, right? Sequoia style, whatever. But it's like word by word, it's just a lot of work. Imagine I could just ask my AI super agent, I say, hey, this is what I've had so far. This is our business progress. Just turn this into a beautiful deck for me so I don't have to spend hours just nitpicking all the shapes of the pictures and so on. Yeah, JamSpark AI does that. It does that by writing HTML. And when it comes to, you know, building charts, we've built internal tools. Um, it'll use in Python. It'll be using Python to essentially draw you all the beautiful charts with numbers. So, yeah, that's just a very simple example. How we, uh, see the future of the world is that all of us, you know, becomes kind of, if we use the movie analogy, we're now actors, all of us are going to become directors. So we're going to tell AI how to, how to act, and then we're going to essentially make our work happen in a much more efficient way. 50 to 70% of the work is, all the grunt work is done by AI now. But still we got to come back and rely on our, um, human judgment to make sure everything is done right. Now, from a company perspective, what we're seeing is because of this shift of how human beings work is happening. Um, and yeah, we want to write real wave. We want to be the pioneer to help everybody to achieve that. That's why we coined the term Vibe working back in April. So everybody knows Cursor, the Vibe coding tool for software developers, using which you just prompt the AI say, hey, I wanted to build the application, and then it'll just write lines of codes for you, right? In our mind, we have 4 million software developers in the states, probably 30 million across the globe. But when it comes to knowledge workers, we have hundreds of millions of knowledge workers. Even, Even here. Yeah, in the States and uh, across the world, billions. We wanted to help a much bigger crowd to really take advantage of what's happening in AI is our vision. We believe this uh, company, if we do this right, there is a shot for us to become a trillion dollar company. I mean imagine how many people using like Google Workspace, Microsoft. 365. All amazing stuff.

Speaker A: Awesome. Well, super excited for you and excited on the journey that you're on. And I've been following pretty much since you've launched and I'm really excited. So why don't we uh, share, man. Let's, let's uh, let's not tease anymore. Let's, let's give them a little something to chew on. You know, I want you to be able to chew on this and see, uh, you know, and I have some ideas on some things I might want to do. But yeah, share your screen. Walk us through it. And we'd love to show how simple it is. And like folks, I've told you this too. Like I've used it in a lot of different ways and um, I've been pretty impressed with the output.

Speaker B: So let's, let's do it. Ryan. All right, so let's say because we're on the show, just say, okay, conduct research on Ryan's. On Ryan, uh, Staley and his show built me a, um, three page because we have limited time, so we'll just do a three page slide deck. Ah, about the show and Ryan, let's say use a dark, uh, background, uh, clean design, um, do it fast. How about that?

Speaker A: Do it fast. I like that. Uh, so how fast does it speed it up when you ask it to do it fast? Is it?

Speaker B: That's a great question. So essentially, um, let me explain what uh, may be happening in the background if I say do it fast. Because our tech architecture is three layers. In the bottom we have a mix of all the latest and greatest, um, large language models from OpenAI, from Google, from Anthropic, from Everybody Grok. So all of these models are capable of digesting natural language and then, uh, you know, provide a response. So if I say do it fast, what? So but then we built this um, uh, routing mechanism. So depending on the task, as you know, OpenAI models are good at reasoning. Anthropic models are good at coding. Google, uh, models are good at creative stuff. So um, our little model committee, they're Going to vote to decide which model to take on what task. Um, and with that when I say do it fast, it'll basically take that into account and basically say okay, um, when we vote which model to take on what it'll calculate, uh, or estimate, uh, which model may be the best to go for this task, um, maybe it's a lighter one, uh, maybe it's a more efficient one versus a uh,03 pro reasoning it takes some minutes. So then on top of that, uh, what makes um, AI agents different from LLMs is you can see the LLMs as kind of brains. They're really good at processing information, spit out the response. But um, what we do is we build arms and legs for the brains. It's a set of 80 plus in house tools. So basically with these tools you could uh, write Python code, you could digest PDF documents, uh, you could search on the web, you could build indexes. It does all kinds of things to allow the large, large language models to touch the real world. So that is where agentic AI comes in. And then on top of that, as we're you know, asking an AI agent to do research on Ryan's Daily and your show, um, we also pay for access to certain private um, databases such as Crunchbase and so on to make sure the output of the um, work is not just from fast web search. People say all kinds of things on the web. Right. So you wanted to ground the results uh, to the uh, uh, very best truth. So that's kind of the three layers of our tech architecture. And with that, um, the AI agents basically takes um, on your objective. It's like you're interacting with a um, Goldman Sachs analyst in your pocket.

Speaker A: Looks like you need to reload it real quick. It looks like it's.

Speaker B: Oh wow. Okay, let me see.

Speaker A: Just to apply the changes. So, and for those of you listening like definitely check this out on YouTube because he's walking us through and what you're seeing is like the different tools um, that the model looked at, you could see. He just showed the code that it created, um, the sources it looked at and like this was all happened. I looked at when you started, it was around seven minutes after and we're now nine minutes after. And so basically what it's done is it's built this out. So Eric, go, go, go back to page one. Would love to see this from page one.

Speaker B: Yeah, I mean look at how many articles, uh, and videos the AI agent digest in such a short time. Right. And then based off that this is. Yeah, exactly what I think uh, looks

Speaker A: up my podcast, looks up my YouTube, looks up Spotify, Instagram.

Speaker B: Research. Research. Research, yeah, research. Uh, and then uh, yeah, it's not building the uh, presentation. Yeah, building the presentation. And uh, uh, the way we build presentations is our AI agent writes HTML code instead. Obviously it thinks first, okay, how do I tell the story? And then it'll just build. And the interesting thing is you see page three is still coding. Uh, we're looking at page one, uh, in the uh, thinking part. But yeah, um, you know it's, it's right here.

Speaker A: That's sharp too.

Speaker B: Yeah,

Speaker A: I'm looking at the facts on here and these are, I mean this is, this is stuff like that's really accurate. Like it's got like some of my work history, like some of the unique things I've done. Can you go down a slide too? It's got notable guests on the scale up show. So some pretty. I had the founder of Travelocity on. Now I have one of the founders of genspark on. So when you're as big as Travelocity, someday you'll be recognized. Okay, let's look at slide three.

Speaker B: Um, okay.

Speaker A: Yeah. Then it's got methodologies. Wow, that's cool man. So keep going. Go ahead. What were you going to say? I had some questions.

Speaker B: The beautiful thing is just like you working with your um, associate, sometimes you're happy with everything, you're satisfied. Sometimes you feel like, oh, maybe I need to tweak this here and there and so on and so forth. Um, you could do that, right? Once the first version of the job is done, you could either edit in AI slides, you could just go to AI slides. And now we allow you to do advanced editing. So this is very different from everybody else, uh, on the market. Right? Like uh, these are real objects you can manipulate. Right. You could just uh, change, say recognize sales, uh, and business leader. We want to say recognize sales and business. Let's just say executive. You could just do that. Um, many other, let's just say uh, uh, so called slides, uh, agent or AI slides out there. They don't allow you to do that. Uh, right. Uh, it's just, you know, you can move some pictures, whatnot, but you can really do this level of granular kind of editing. Um, and then besides, you know, this is almost like Google Slides here. Right. We're obviously still, you know, making it better, crafting it day by day. But besides, you know, the um, manual, uh, editing, you could also continue to prompt the AI, basically say hey Jenspark, um, You know, uh, let's say add, um, hat shot picture of Ryan, um, to the, uh, let's say first page. M. Find. Find the best picture, uh, on his website or something.

Speaker A: It's not. Yeah, I don't think you click the box because it's not typing in the box.

Speaker B: Oh, uh, uh. Oh, yeah. Oh, the prompt is. Can you, can you still see.

Speaker A: Uh, yeah, I see the presentation. Yeah, I just don't see what you're typing.

Speaker B: Okay.

Speaker A: Unless you open a new window, maybe that's what.

Speaker B: Oh, yeah, yeah, I'm on the new window. Okay, let me, let me, let me switch that. So when I add, when I clicked the edit in AI slides, it immediately opened a, uh, new window.

Speaker A: Oh, yeah, yeah, yeah.

Speaker B: Okay, so this is the second tab. So when I tapped on, uh, edit in AI, um, so it just switched over. It took over all the context. It switched over. Now I told it to, uh, headshot picture of Ryan. It started to search all the beautiful pictures of you.

Speaker A: Some of those are not so beautiful though, too. Just throwing that out there. Some of them, some of them look like I'm a real estate agent from like the 80s. It's that, it's that one with the jacket. I gotta get that updated, man. It does. It totally looks like my 80s. Like my dad was in real estate in the 80s and he had a little like envelopes and stuff. That's what it looked like. So.

Speaker B: Oh, uh, that's how you make money back then, right?

Speaker A: Door to door, baby. Door to door.

Speaker B: Okay, Exactly.

Speaker A: Now what it's doing is spinning it up. And just while it's building this, I'll give you my preference first hand experience that I saw. So you could fact check it, um, as well. Like you're looking at docs or you have a pull report and you're like, that's a lot of data. I'm not sure. It could fact check things. Um, you could export it to PDF or PowerPoint and then edit it there as well. I've also, um, he's adjusting like a slide. I've also done like, hey, add an additional slide, but include this data. And it does a really good job of adding it, keeping it in the same style, structure things along those lines. So. Yeah, uh, question for you though, when. Well, this is. Oh yeah, it's got the, it's got. Oh, it's got the good headshot. It's. It's got the good headshot. I love that, man. The colors are aligned with my brand as well. Which is, which is pretty interesting because you didn't even tell it to do that. So let me ask you that. So how do you do that to keep it um, like control presentation types, um, like fonts, brand. Is there anything that you could do to do that on the presentation side? Because I know that's what a lot of our B2B customers are thinking. Especially um, for those that build it.

Speaker B: Yeah, 100%. There are multiple ways of doing this, Ryan. So one as I mentioned. So I'm going to click on this advanced editing and uh, it's going to open up a new tab, but I'll share that one once it's open.

Speaker A: Go for it man.

Speaker B: Uh, actually we're in the uh. Can you see the open up? Yeah. So if you want different fonts you can just change it easy right? What's your favorite font? This is one way of doing it right. Like you can, you can, you can change everything. If you don't want it to do it manually, you could just go back and then um, tell. Uh, so let's say right now it's, let's say we wanted to change all the titles uh, to Roboto. So we just go back and say okay, great work and change all the uh, uh, title um words to uh, slot Robo. Nice.

Speaker A: Okay, so yeah, then I like that, like if I can I feed it an example of a presentation I've done before and have it mimic that style and structure 100%.

Speaker B: So not uh, only you could do that. We also have ah, hundreds of beautiful templates uh, at your um, at your disposal. So let me just jump out here and uh, it's doing its thing. Let me uh, jump out and Actually one thing I wanted to quickly show you is um, once this is completed, um, you will have an option to save this as your future template. Uh, you could do that also at the same time if you have um, you know, existing slides you wanted to throw in and then just say hey, use this as template and then build me new new ones. No, it's, it's, it's done. Right. Nice. Yeah, it's just, yeah. Changing the future.

Speaker A: So yeah, I love this man. I've used other tools like Gamma and I think that's just the, the versioning you have is much, much tighter and looks more like uh, real presentations that folks use. So real quick because we're just about up on time man, and we're going to have you back obviously. Um, what are your like, like? I guess any favorite presentation template styles that you like and then any like, tips for to be top 1% when using this.

Speaker B: Uh, absolutely. Let me, um, just, uh, um, close this one and open up a new one to show you some of the templates we have and we'll talk about. Let's see, one sec.

Speaker A: And as you can see, guys, we're doing this live, so I mean, I

Speaker B: should have prepared, but it's okay, man, this is good. So, yeah, uh, can you see my screen now?

Speaker A: Yeah.

Speaker B: So this, this, yeah, this is our user interface. This is our AI slight, uh, super agent. Uh, you can explore, right? All kinds of, you can filter it, all styles, minimalist or whatever themes, you know, education, business, popularity. Newest. Uh, yeah. So you can, you can build a, uh, slide deck in Japanese if you want it. Nice. But yeah, I personally like dark backgrounds. But let's just see this one. Business Executive solutions. So these are all templates you could just tap and then just directly use. And then you just say, hey, uh, build me, uh, an intro deck, rewrite daily. And his, ah, show. Um, basically it'll just fit all the content into this template. This is one way of doing it. Right. And then another way of doing it is my template. You could, um, upload your previous slides because we all have preferences. You, uh, could just throw it in and the AI agent would just work exactly like your own staff. It'll just build on top of that.

Speaker A: Nice, man.

Speaker B: Yeah.

Speaker A: This is fantastic. Um, all right, top tip. Outside of the template style and you're building this out. We might have to show this in the next one.

Speaker B: Yeah.

Speaker A: Any other great tips for presentation creation? Cause, like, what you showed, I think was like insanely simple, which I think is great for adoption, and then the quality is really good.

Speaker B: Yeah. Um, top tip, I learned this kind uh, of the. So top tip, really two things. One, um, you gotta really just want that. You gotta ask the AI to do everything before you even do it. That's it. That's it. Just the allocate. Yeah, that's it. Just, yeah, assign it out. Don't try doing it yourself. Um, that's the number one thing. Uh, and I'm serious, like before, I'm a hustler. Uh, I don't know if you can tell, but I always jump on things and sure, I benefited from it, but, uh, now what Skills, Stilton, all that. Um, you know, it's just, I'm just a lot more efficient. I can get so much more done if I just, you know, ask the AI to, you know, get going with, you know, research this client for me, built this deck for me, build a financial model for me. Write me this investment memo, this investor being nagging me for, um, and all the good stuff. And then, um, you know, you just come back and then talk to the AI again, say, hey, I'm happy with this. I'm not happy with that. Can you change this and that? Um, that's it. Yeah, just. Just don't try applying yourself, uh, to everything. Ask the AI to do something, go grab a coffee, come back and see what happens. It's magical.

Speaker A: Yeah, that's great. Well, and I've used that, man. I've opened multiple tabs and the solution have done that. Exactly that. So I think it's great. Just. And what I would say is a tip on my end is identify the outcome you want. And if you need multiple steps, sometimes it's. It never hurts to do that if you have it really, like, specific in your mind. Kind of what he said, like, you know, research this, create a presentation here, create a financial report afterwards. Um, I even created a, um, a whole brand deck for Apple based on Apple's like, iconography and everything that they do, and then turn that into a webpage relatively simply as well. So. Awesome, awesome, awesome. Did not disappoint. And we will have back when. But where can people find you? Where can they find out more about genspark and then we'll wrap things up for today.

Speaker B: Absolutely. Yeah, just go to genspark AI and you can just. Yeah, look at that. You can just get going with, um, you know, everything we have to offer. Um, and we're on LinkedIn, we're on Axe. Uh, we even have a Discord group, so if you wanted to join the community and stay tuned on all the latest, feel free to do so. But, uh, yeah, just go to JetSpark AI more than.

Speaker A: Well, I appreciate you being on the show, man. Appreciate you taking the time. I know things are blowing up for you right now, and, uh, it's still building out the new docs right now as we're speaking. So. Anyway, yeah, thank you for joining us today and we will see you all on the next episode.

Speaker B: Sounds good. Thank you, man.