Inside Genspark: $0 to $250M ARR in 12 Months with Wen Sang

19:14Grant:And you also have to figure out all of this engineering and harnessing yourself. Exactly. Yeah. One of the best benefits you have here, because one of the things I've been struggling with with the frontier models is trying to figure out exactly the best way to build essentially something like what you've built here, where for myself, where I'm just saying, hey, I have a lot of recurring tasks that I need to do in any given week. and how do I build skill systems that help me, you know, essentially solve all these problems for myself, but I have to connect it to all of these different applications.

19:47Grant:So it looks like here you have connectors as well. Is that correct? Yeah, let's take a look. So from a connector standpoint, we totally understand a lot of our data sits in different systems. And look, here's the thing. The world used to be one that is tool-centric. Each of us would have to learn how to use 20, 30, 40 tools to get our work done. And people used to get certifications for being good at Excel. That was a thing. So those days are long gone. You no longer have to get certified for being good at Excel anymore. What happens is now you can just connect your JazzBark. And then JazzBark will learn how to use all of these tools and move, edit, delete, and then just make changes to all of these data for you.

20:31So, yeah, to that, like connectors to all mainstream stuff, Google Suite, Microsoft Suite, Notion, Slack, HopSpot, Pipedrive, GitHub, Box, anything, everything, Zoom, MailChimp, all of these integrations. Beyond that, we also support different super base and so on for the AI developers and so on, financial data. And we also support over 650 MCP servers today. So you can even set up –

21:04Corey:Even the Reddit MCP. What's that? Even the Reddit MCP. I like it. Yes, yes. And you can have your own MCP server set up for you. So we understand the world is entering into a human-centric world. So none of us – the boundaries among all of these tools are disappearing, and you probably saw the ex-post by Mark Vinioff saying that Salesforce has gone headless. So we see the world exactly the same way. Basically, the legacy software is taking a step back to become the infrastructure for agents, and we're building the agent to work for human beings to use the software tools on their behalf. so we're getting elevated everybody's a boss now you just have a fleet of goldman sachs and alice running in your pocket doing things for you is basically the case how would you um launch your

21:56Grant:own aid like can you create your own agent on here i know you have a lot of built-in ones yeah like you could you could build a creative designer coding assistant and then you can configure configure it basically for example say um we can say hey build me a executive assistant for Grant at the Neuron podcast. And he loves chocolate cookies. I do. It's true. So, yeah, basically, this is where you get into the custom agents. Yeah. Okay, perfect. And then once you tap onto that, you come to this page. And then once you click on create new, there are a bunch of other agents you could use. But once you click on create new, you could just get in here.

22:50The process of working is similar in a way that now it's building the grant's executive assistant agent for you. Yeah, amazing. So, okay. It's like, oh, yeah, it is. Yeah, we can publish it now. Put this to the test. Yeah. Yeah. Okay. That was fast. It was fast. Okay. Let's try it out. Who are you? What are you here for? Hi there. And yeah, what's your purpose?

23:26Grant:I'm Genspark, Grant's executive assistant. I'm here to keep track of the latest AI companies and agentic systems, deliver Grant's morning briefings, and prepare board-ready reports based on the Neuron podcast. If you need anything related to Grant's work, just let me know.

23:46Grant:Grant's big on AI and agentic systems, loves chocolate cookies, and has a soft spot for the show. Yeah, thank you. Anything I can help you with along those lines? That's cool. And then, so, would I have to do any sort of extra work to connect it? Like, let's say I wanted it to look at Reddit. Let's say I wanted it to look at, you know, the websites that I check every day. what could I do from this point now when this is set up? You just tell, hey, moving forward, please remember to check Reddit. And what else you said it would be? Whatever other words. Let's say, like, let's check Reddit. Let's check TechCrunch and all the tech blogs that publish about AI.

24:24No.

24:26Grant:TechCrunch as well. I'll keep checking Reddit and any other sources you find helpful. Great. Just let me know what else you'd like included.

24:34Corey:This is exactly how this should work.

24:36Grant:I'll make sure to keep checking TechCrunch, too. If anything else comes to mind, just let me know. So, look, yeah. So, this is a custom agent that lives on our cloud in front, which is Microsoft Azure. So, you can give your custom agent more and more and more context whenever you feel you want it to. and on top of that you can also set up your claw jensbark claw which works i'm gonna ask about jensbark claw yes it's even cooler for people who don't know so what is open claw we we've talked about it on our show and we've talked about it you know in our newsletter but you know how how do you understand it and if you could just give people a quick review of what it is so basically um my My understanding is this, Grant.

25:30When Chagibiti just came out, the way human beings interact with AI is we go to AI. We go to Chagibiti.com or we go to Jenspark.ai and we do single sessions. Maybe a talking session, maybe a project session. There's no true memory. There may be preferences saved somewhere, but it's like session by session. Every time it's a complete new session. Today, with the open-claw framework, what happens is, one, the AI follows us wherever we go. Let's take a look.

26:03Corey:I'm an open-claw junkie, so you have my curiosity. So on a higher level, the technology or the framework is similar. But we know that not everybody is as technical as you gentlemen here. Buying a Mac mini, setting up Node.js scripts, and then getting the API keys is no joke for a normal, non-technical person like myself. Absolutely. It scares me. And trying to make sure it's all safe and secure, it's a lot. So what we did was, you know what? We see the tremendous benefits of running long memories and then allowing your agent to do long horizon tasks. but how to save that trouble for our 1 billion plus non-technical target audience.

26:54We say, yo, we're just going to rent a virtual machine on Microsoft Azure for you and we'll set up OpenClaw on that machine for you. And all you need to do is click a button saying that I want access to this and that it's up and running. So if we click on Gensbart Claw here, it's the same thing, right? You can connect all kinds of channels. Is your claw in the cloud too or running locally? So we have both now. Okay. Yeah. So that's such a great question, Corey. So this claw you're looking at, I'm accessing it through a browser, and this is behind it, I'll show you. It's a virtual machine. So you can see open remote desktop.

27:35So this is a virtual machine. Sharing this tab.

27:44and so basically, yeah, this is my LinkedIn. It opened up. I asked it to add everybody who I met at this event on LinkedIn for me. Nice. I don't have the click of the button. But, yeah, it's a virtual machine. But if you – yeah, this is like – this is a Microsoft Azure computer at the same time.

28:08Grant:So basically what we're looking at is the computer that the agent is running on right now. Correct. So if you were to click on that randomly at any point in the day, would you be able to see what your agent is doing technically? Yeah, let's see it. So let's say, okay, buy a cup of coffee for Corey, who is in St. Louis. Ready to continue. Who is in L.A. for me. I'll give you their addresses shortly. Feel free to use my DoorDash account. That's a good callback. We just tried to set up a DoorDash agent last week. Sure. Okay, so Corey is at the 123 Main Street. Grant is in Hollywood. I don't know.

28:59Sunset Boulevard 1234.

29:06Let's see what happens here.

29:08Grant:I'm tempted to give you my actual address.

29:10Corey:Let me set up those coffee orders now.

29:12Grant:What's that? I'm tempted to give you my actual address. I'm on it. I'll let you know as soon as the orders are confirmed. Let's get it. I don't know if I want to dox myself. Maybe we'll take that part out.

29:29Grant:It looks like we need to log in to DoorDash first, and we'll need to place separate orders with the coffee details. Could you log in and let me know what type of coffee you'd like for each? So in the very beginning, it does require, yeah, log in. So let's see what happens here. Yeah. Let me check on that for you.

29:44Corey:DoorDash stumped us the other day. We tried to do that last week on another tool. Oh, really? Yeah. And had DoorDash problems, it felt like. Okay. Yeah. I see. It was on their side. Yeah, let's see if Desperate Cloud could come through. um you know something i think is is really cool here that you've done and especially with the claw but not just with that and and you mentioned it was this idea that for normal people a lot of this stuff has been inaccessible like like open claw is awesome if you can do it and if you get it and same with these other edges but the fact is most people are not technical enough for for that approach or that amount of troubleshooting that comes with it sometimes.

30:30Corey:To have simplified that in an easy way is really cool.

30:37Grant:So this is all the agent operating right now? Is that what's happening? So let's see what's happening. It says I'm not logged in. This is me. I'll just log in for the agent. Is there any sort of indicator? Because I know when you use let's say like GPT agent, for example, you know, it's all in that sandbox and there's the window and it gives you the option to click and take over. Is there any sort of indicator, you know, perhaps that we can't see that says like when you're controlling it versus when the agent's controlling it? Or do you both control at the same time and you compete for the mouse?

31:14That's a great question, Grant. So for the computer use, I could use the computer. The agent could use the computer. We could compete. We could do it separately on separate tabs at the same time now for the browser use we would basically it'll have the the blue thing that that'll it'll tell you it has taken over um so that's browser use yeah but here i'll just uh tell okay so yeah uh let's just get latte for uh both of them from the closest the starbucks or not or so um and yeah feel free to uh operate now typically i do this through my whatsapp i I don't even come into the –

31:50Corey:I use Telegram for my claw work. All of these channels, WhatsApp, Slack, Line, Telegram, Microsoft Teams, all of these, you can basically kind of in and you can talk to your claw, your dedicated agent like 24-7, and it'll just do things for you, not only using the agentic capabilities, for example, but also it could open up a browser. It could operate the browser. It could anything that human beings could do, the agent would be able to do for you. So the AI follows you, and it's not just a single set. It runs a long memory of you. I've told my priorities, and whenever it does anything for me, like scheduling meetings and or just setting up new conversations with the clients and so on.

32:47It always check on that priority list to see if this is a good use of time. So basically the progress we're seeing is the feeling of human AI interaction is getting closer and closer to human and human interaction. Yeah. That is kind of, yeah. I would say though, this may not be the last innovation in the year. We probably would see even more powerful things coming up because even with OpenClaw kind of a framework, even with Jens Bar Claw, sometimes it's fast, it's amazing, it does a lot of good work for us. You still see kind of certain gaps like, oh, that's a common sense thing. Why don't you just get it?

33:32There are still like, of course, we have a high bar. But what I'm trying to say is, yeah, we're in the beginning of the beginning, and it is just amazing to see how quickly AI could become something. Not only it talks to you like a human, it actually does things for you like a human, and it remembers things now for you like a human. It's just, yeah.

33:56Corey:Yeah, and we're only at this moment, what, less than four months removed from the Open Claw moment when it first released? So, I mean, the truth is we're, you know, this is still brand new. That there are already, there's already solutions that work in a way that is more user friendly is really encouraging. I can't imagine by the end of this year that everyone doesn't have one of these. I couldn't agree more. I think. That probably works easier than we could dream of right now, if I had to guess. Yeah, yeah. Look, I would say the exciting thing about, you know, this is really an exciting time to be alive.

34:36It's things are moving so quickly, right? Like new things are coming out every day. And I would say, yeah, any product on the market, it may not be perfect, but you're seeing magical moments just popping up constantly. That is kind of what we're experiencing. I'm sure there will be more polishing work falling in the future days, whatnot. But right now, it's like an explosion of innovations. That is just, yeah, insane. Like, yeah, I mean, you asked me a question earlier.

35:14What Jetsbark initially did and then what is 4.0? I mean, at this point, yeah, 4.0, we launched the club. but also like we have the creative suite, like beyond the, you know, dark slides, whatnot, you have image video, you can do developments. Basically, we can build a song about your podcast right now in a couple of minutes, a theme song. Oh, wow. To your preference that you own the IP. So. I love that. Let's, yeah. On that subject,

35:42Corey:I have a quick question for you. We're getting, I noticed we're just a few minutes from time here and I don't want to, I want to be respectful of that for you. But as a business owner, going through what you all are going through right now, how do you keep up with the pace and ensure that, like, you have things in here that are extremely current, in my opinion. And I'm wondering, how do you keep up with that? It's a whirlwind, Corey. I'll be honest. When we first launched Jazzburg, AI Workspace 1.0, we did not have the expectation that it's going to go like this. But it's amazing. It's overwhelming at the same time.

36:22It's just so exciting. I mean, yeah, I'm having fun. It's also, you know, there are new challenges constantly. Like when we first launched, it was really a product for individuals. And by summertime, a lot of corporations started reaching out, you know, consulting firms in Chicago, auditing firms in Atlanta, you know, insurance companies in Hartford and San Francisco, you know, PD firms in Japan, management banks from London. and oil gas companies from Bogota, Colombia, government agency from Dubai. They were like, oh, our employees used their solution. They loved it. They wanted us to pay for it.

36:56Do you have a business offering? We're like, what do you mean? You can just swipe the card. They were like, no, no, no. We need to centralize the billing. We need SSO. We need SOC 2 Type 2. We need all the enterprise stuff. That is when we got into the motion of, yo, what? We need to build this out because eventually we wanted to serve prosumers, power users, but also the Fortune 2000s. And we launched Jenspark for business back in November, right before Thanksgiving. And just last week or two weeks ago, I checked, we have now over 5 ,000 business customers. So that sort of change, that sort of change is like, right?

37:34It's insane. Yeah. This is absolutely, I'd say, on the one hand, unfathomable, and on the other hand, impossible in the old days. But today it's also possible because at GenSpark, today near 100 % of code is written by AI.

37:54Corey:Yeah. Wow.

37:55Grant:Do you use GenSpark to build GenSpark? Oh, sorry to cut you off, but do you use GenSpark to build GenSpark? So not exclusively. We absolutely use Jenspark to build Jenspark, but we also use – so, for example, our landing pages, it's like Jenspark. But we also use other amazing tools on market. Like, for example, earlier it was Cursor and then later on ClockCode, CodeDax. So, yeah, I mean, we're a believer of this amazing, this great time. And we are absolutely in it to basically just go all in. We're absolutely fully native because we see some of the really successful tech giants struggle with building new things because they have too much burden to kind of patch on.

38:55Like if you have – yeah, so for us, clean slate. We architected our technologies three layers as we talk about. And for us to build new AI agents, it's like just the speed of – our company is basically speed of an individual. We don't have like front-end, back-end, design, DevOps teams. No, it's like every one or one of two engineers lead one full project or product, and they can just run with it.

39:20Corey:Say three years from now, where do you see GenSpark? We believe deeply that with this wave of AI breakthroughs, humanity finally arrives at this historical point where we no longer have to work. We can choose to work. And at least we no longer have to do the busy work in order to achieve the real business results. And we're in for that mission. We wanted to free all human beings from the mundane, busy work so that we can decide if we wanted to work harder and earn more, do more, or we want to just take a step back and chill and then just enjoy life, spend time with families and so on. That's completely up to you, but it's about more freedom.

40:02It's about more freedom. It's about creating a lot more freedom for each and every of us. We're all different. Yes, some people like to hustle. Some people like to chill. That is all good. But the important thing is to give all of us a lot more freedom. And how to do that? You just autopilot all the busy work. And you kill the busy work, but you create a lot more new jobs. Yeah, because we're seeing a lot of these companies, if you're good at AI, you get promoted these days. We see new job titles popping up, AI specialists. I've never seen that before until someone told me, you should search on LinkedIn.

40:40I searched it. Oh, my God.

40:42Corey:Yeah, there's about 40 names it might go by too.

40:45Grant:Yeah, yeah, yeah, exactly. The two lightning round questions I have. So the first one is we mentioned the Git server. Could you just very briefly explain why that's needed? And then the second question, which you can answer right after, is if people are using, you know, you have so many capabilities in here, there's a lot of power. If people are using GenSpark for the first time, what do you recommend they do? What's their, what flow should they start with? So those are the two questions. Yeah, yeah, definitely. So on the Git, basically, Grant, what we're seeing is we're seeing a lot more teams and organizations started using Genspark.

41:21And there are possibly years of experience insights within these companies. and it has always been a struggle that there was no like organizational or even for a single person kind of that memory or that knowledge base. It's just so difficult to do the kind of the search thing and that really truly synthesize all the information and make good use of it. And the world is moving toward one that, you know, human beings just interact with AI agents and AI agents does a lot of these work. So we basically – this probably got sparked by Andrew Kparthy's one expo saying that, hey, it would be amazing if there's a wiki or something for the agents.

42:02Well, that makes perfect sense. And the Git technology by itself is born to do this, right? So we're like, okay, why don't we just build this out to benefit all the people out there who want it to have – because memory is a huge thing for us to approach the next level of agentic capabilities. and we have Michael Von der Jesting, he scaled technologies like Facebook Messenger from zero to billions of users, also founding member PyTorch he's like, I know exactly how to do this so that was just it was, yeah

42:32Grant:and then 30 seconds if people want to use GenSpark where do they go and what's the first thing they should build with it? You go to GenSpark.ai like I'll show you quickly here on the screen yeah, GenSpark.ai GenSpark.ai And then first thing is anything that you want, truly, because I've seen, you know, I've seen, you know, dev shops asking Jenspark to build a full website on Shopify for them and then just deploy and then, you know, just charging their clients for six, seven thousand dollars. And they pay it past twenty five dollars a month. I've seen business, you know, brokers, consultants securing bank loans with Jenspark by building up the full market analysis, financial models and the pitch deck to the banks, securing millions of dollars.

43:16Wow. I've seen, yeah, marketers building on marketing materials and then packaging up a whole project, you know, just to deliver it to their clients and that really make their lives a lot easier. All kinds of things. As long as it is leveraging AI to get the work done and to earn you more money, James Bart got this.

43:35Corey:Well, Wynn, thank you so much for joining us today, man. This has been a lot of fun. It's been really cool to check out your tool with you. Absolutely, Corey. Thank you. Thank you, Grant. This is amazing. and hope we can launch more new agents and then come up onto the show again soon. Excellent. Yeah, absolutely. And gensmart.ai? Yes, that's it. Cool. Go check it out, everyone. One quick note before we sign off. First, please remember to hit like and subscribe above so you don't miss any future or past conversations as well as our weekly live sessions. Also, if you're trying to keep up with this space, likely all are, and it feels like a lot because it is, that's basically why we write the neuron every day.

44:13Corey:It's a short daily breakdown of what actually matters in the AI space. Pop by and subscribe today at the neuron.ai. And that's all we have for today, folks. Farewell for now, humans.