Transcript

The $1M+ Solo Grok Bot Business (Full Course)

From Greg Isenberg · Episode page · Watch on YouTube

· 50 min · 9,044 words · 94 timestamped sections

This is the full, timestamped transcript of "The $1M+ Solo Grok Bot Business (Full Course)", an episode of Greg Isenberg published November 2, 2026 (50 min, about 9,044 words). It mentions 32 people, works and ideas, including Grokbot, Greg Isenberg, Nick Mosovich, A.I., Bops.

What this episode is about

This episode unfolds a practical masterclass on building and selling AI agents as a scalable business, using a Grockbot-style personal agent platform as the launching pad. Nick from Orgo breaking down how to niche, price, and package a full-stack solution—covering offers, tools, and workflows to land $25k upfront plus ongoing $3k/month retainers. Listeners will come away with a clear blueprint for assembling the stack, crafting market-ready applications, and the mindset to become more AI-native, whether or not they pursue the startup idea themselves.

Key takeaways

  • Sell AI agents, not just apps—bundle them for steady monthly revenue.
  • The sweet spot: custom Grockbot stacks plus a clear upgrade path.
  • Combine agents and apps to close the entire business gap, not just automate tasks.
  • Software prices drop; monetize with human-backed intelligence on top.
  • Think niche-first: start with a proven Grockbot idea and scale from there.

Who speaks in this episode?

What is discussed, and when

  1. – 5:00: Grokbot, Twitter, Greg Isenberg, Cursor
  2. – 10:00: Grokbot, Greg Isenberg, Bops, SaaS
  3. – 15:00: Grokbot, Opus 5.5, Twitter, Agent Mail
  4. – 20:00: Opus 5.5, Grokbot, Orgo, Paper
  5. – 25:00: T3 Code, Composio, CleanShot, Grokbot
  6. – 30:00: Grokbot, Orgo, Agent Mail, Hermes
  7. – 35:00: Grokbot, Greg Isenberg
  8. – 40:00: Grokbot, A.I., Greg Isenberg, LinkedIn
  9. – 45:00: Willow Dental
  10. – 49:34: Nick Mosovich, Greg Isenberg, Bops, Startup Ideas Podcast

Notable quotes from "The $1M+ Solo Grok Bot Business (Full Course)"

  • I think the people are saying that because they don't want other people to go and build things. So yes, do I think that software is evolving? 100%.
    about A.I. · at
  • Right, plugins, templates, it's going to be out of the box like a 10x experience for them because if they go and download Grockbot, if they go and download another one of these…
    about Grokbot · at
  • It and so you can build inside of claude code you can build inside of T3 this the guy on Twitter he has his T3 code I actually built bops inside of T3 and I'll show you here this is the exact conversation that I had you can see.
    about T3 Code · at
  • Like Figma for the AI era and I designed everything inside of paper. You simply give your coding agent. So, inside of T3, I have Opus 5.5 and I installed the paper MCP. And…
    about Opus 5.5 · at
  • It simple I didn't want to build this infrastructure myself so obviously like I use orgo I'm a co-founder of orgo we build the computers so you don't have to build them…
    about Orgo · at
  • Or we can sell this like one to few which is like selling to the SMBs. Right now Greg if you were to sell this to SMBs like I mentioned we could sell this for like 25K upfront…
    about Greg Isenberg · at
  • So I hope other people go out and build it, too. And, and the beautiful part is, Greg, this entire thing that I built, if you don't want to take the 36 hours to, build your…
    about Bops · at
  • Thank you so much, Greg. Appreciate you always having me on and yeah, love jamming with you in the audience. So guys, go out there, build some cool things, and we'll see you.
    about Nick Mosovich · at

Who and what is mentioned in this episode?

People

  • Greg Isenberg — Person · Host · 6 mentions · first at
    Host of the podcast; asks the questions that drive the practical, step-by-step explanation of building and selling GrokBot-style agents.
  • Nick Mosovich — Person · Main guest · 3 mentions · first at
    Guest who built Bops, explains the stack, build process, pricing, and go-to-market for selling AI agent teams to businesses.

Companies & orgs

  • Orgo — Company · Vendor / product provider (computers for bots) / guest affiliation · 3 mentions · first at
    Company Nick co-founded; provides the 'computers' used to host each bot and is part of the stack for Bops.
  • Twitter — Company · Distribution / social proof channel · 4 mentions · first at
    Nick posted a demo video of Bops on Twitter that received 75K views, helping validate and attract customers.
    Yep. I'm literally showing a video on Twitter of this tool called BOPS that I built. I called it BOPS because I'm it's business ops. And I wanted to, build agents that solve business operation problems. And you.
  • Composio — Company · Connector provider in the stack · 2 mentions · first at
    Composio supplies connector integrations used by Bops to connect bots to external platforms (single API to many connectors).
    Build your own bops. I used agent mail to give it an email. I used agent phone to give it a phone number. I used composio to give it these connectors. So you can see here when…
  • LinkedIn — Company · Platform example for agent automations · 2 mentions · first at
    LinkedIn is used as an example of a platform where a dedicated agent could run outreach and booking flows.
    Are useful. Like I have a bunch of agents here. I actually a quick tip for people like I personally build an agent for every platform. So I have a Twitter agent. I have an…
  • GitHub — Company · Code hosting / repo location · 2 mentions · first at
    GitHub is referenced as a place Nick could have used for version control; the Bops repo is available to fork on GitHub.
    I built this like an idiot. Like I literally just kept going in one single long thread. I didn't even like connect it to GitHub or anything. I didn't push pull like I just did…
  • Hacker News — Company · Monitoring source · 1 mention · first at
    Referenced as a source the bots should watch for relevant posts and signals (example of tasks bots can perform).
    And when I text him, like it shows up here. It says, Shi to the podcast. And he's going to reply and he's going to say hi. And then I That's really cool.
  • Fly — Company · Hosting option · 1 mention · first at
    Fly is mentioned as one of the hosting options for the agent application (alternative to Orgo and Vercel).
    And all of that. so give it the things that'll make it useful. Power it up, boost it up. decide where you're going to host it. You can actually host it on orgo. Orgo is like…
  • Vercel — Company · Hosting option · 1 mention · first at
    Vercel (also spelled 'Versel' once) is listed as a hosting option for shipping the product quickly.
    And all of that. so give it the things that'll make it useful. Power it up, boost it up. decide where you're going to host it. You can actually host it on orgo. Orgo is like…
  • Andreessen Horowitz — Organization · Referenced investor / industry datapoint (previously misclassified as person in input) · 1 mention · first at
    Investor entity relevant as a venture-capital firm mentioned in context of AI tools funding/usage
    You don't know how many people Well, did you see that Andre Horowitz post of like 2% of the US pays for AI tools. So, if you're building with these things, you're taking it for granted how competent of a skill that is and how.
  • Willow Dental — Company · Example potential client (illustrative) · 1 mention · first at
    Mentioned as an example of how a founder might struggle to pick bots for an actual business and how to approach a client like a dental clinic.
    How do I take this and actually bring this to Willow or Willow Dental to actually like how am I supposed to know which bots I need to create? How do I figure out what the pain…

Products & tech

  • Grokbot — Product · Reference product / market comparand · 22 mentions · first at
    Grockbot (Grok-like personal agent platforms) is the conceptual competitor and inspiration for purpose-built, niche agent products sold to businesses.
  • Bops — Product · Case study / product demo · 2 mentions · first at
    Bops is the GrokBot-like agent platform Nick built in a weekend and used as the concrete example of a sellable, nichable AI agent product.
  • Opus 5.5 — Product · LLM model used to power agents · 6 mentions · first at
    Opus 5.5 (Anthropic model referenced) is described as the capable model used to build and run agent logic in Bops.
  • T3 Code — Product · Development environment / tooling used to build Bops · 3 mentions · first at
    T3 Code was the environment Nick used as the single long-running thread to build Bops quickly (36 hours).
  • Paper — Product · Design tool used in the build process · 1 mention · first at
    Paper was used as the design step (Figma-for-AI) to iterate UI/UX mockups before building Bops.
    Want to build your own Grockbot, you could use our computers. But paper is like Figma for the AI era and I designed everything inside of paper. You simply give your coding…
  • Agent Mail — Product · Email provider for agents (stack component) · 3 mentions · first at
    Agent Mail was used to give each bot its own email address as part of the Bops architecture.
    Build the computers so you don't have to build them yourself You can use them to build your own bops. I used agent mail to give it an email. I used agent phone to give it a…
  • Agent Phone — Product · Phone-number provider for agents (stack component) · 1 mention · first at
    Agent Phone was used to assign phone numbers to bots so owners can call/text the bots directly.
    Build them yourself You can use them to build your own bops. I used agent mail to give it an email. I used agent phone to give it a phone number. I used composio to give it…
  • Hermes — Product · Referenced example of a sold agent (past work example) · 2 mentions · first at
    Hermes is mentioned as an example of an agent Nick or others have sold to businesses for recurring fees in prior work.
    To everything. And I built my first money-making business online, building an app using AI. Then the second biggest way that I've made money from AI was with agents, selling…
  • CleanShot — Product · Tool used for screenshots / swipe file · 2 mentions · first at
    CleanShot is recommended as the screenshot tool Nick uses to capture UI for feedback and to feed to the coding agent.
    So, CleanShot, for those who don't know, it's a screenshot app. It's just better than the native screenshot inside of your apps, inside of your Mac. so I literally will screenshot, the.
  • Cursor — Product · Referenced tool that enabled past app-building success · 1 mention · first at
    Cursor is referenced as the tool Nick used historically to build a money-making app early on (inspiration from a prior podcast episode).
    So I've made money with AI by building apps. So the first when Cursor came out, funny enough, fun story, I watched the Greg Eisenberg podcast. You had Mickey on. He was teaching Cursor. I think this was like summer of 24, I.
  • Claude Code — Product · Alternative development environment mentioned · 1 mention · first at
    Claude Code is referenced as another option for building and prompting agents (alternative to T3).
    Good. It's very simple and you could see it's a very simple design. so I recommend it. It's free and if you already have a Claude subscription, you could just use it inside of T3. It's free. and I think it's open sourced.
  • Meta Muse — Product · Referenced comparable product · 1 mention · first at
    Mentioned as another Grok/Dots-like product in the space used for comparison during research step.
    Very first message I sent inside of building this project of bots, I literally was like I want to build a version of Grockbot or Meta's Muse or OpenAI's dots very similar…
  • Slack — Product · Integration channel example · 3 mentions · first at
    Slack is given as an example of a connector/channel that a bot could have for communication.
    This. I don't want to spend too much time. the bot life. I talk about that, give every agent its own email, phone number, computer. things to make this thing these things…
  • iMessage — Product · UI inspiration example · 1 mention · first at
    iMessage UI is referenced as part of design inspiration for Bops interface (Apple-esque look).
    Applelesesque you know like you're familiar in iMessage and FaceTime and all of this like I landed on that kind of design very you know inspired by Grockbot as well and that's…
  • FaceTime — Product · UI inspiration example · 1 mention · first at
    Referenced alongside iMessage as part of the Apple-like UI inspiration for design.
    On this like Applelesesque you know like you're familiar in iMessage and FaceTime and all of this like I landed on that kind of design very you know inspired by Grockbot as well and that's before any building so that's the second step.

Concepts & topics

  • A.I. — Concept · Core topic · 1 mention · first at
    AI is the foundation of agents, the technology enabling the business opportunity being discussed (selling AI employees).
  • SaaS — Concept · Contextual contrast / market observation · 1 mention · first at
    SaaS is discussed as evolving — the episode contrasts 'dumb SaaS' with custom, niche-built agent solutions.
    I think we're moving more towards like the price for software is trending towards zero and you're going to give away the software and you're going to upsell agents and human intelligence to.
  • S&P 500 — Concept · Data example / request from bot · 1 mention · first at
    Mentioned as an example of a data source Nick asked a bot to pull during a phone demo (shows capability to fetch market data).
    Hey, Bo Poppy. Yeah, I can hear you. Can you pull up can you pull up like a list of companies that I could sell to and do pull up their websites as well? And then also pull up the S&P 500.

Places & events

  • Y Combinator — Event · Referenced accelerator / credibility benchmark · 1 mention · first at
    YC is referenced briefly in relation to launch videos / demos as a benchmark for attention.
    So, like this almost did as good as our YC launch video and it took me two minutes to make. So, you don't need VC funding. You don't need a lot of money.

Media & podcasts

  • Startup Ideas Podcast — Media · Show / distribution · 1 mention · first at
    The podcast (hosted by Greg) is the format for this interview and the place where this idea and resources (repo link) are shared.
    Yeah. And this is the Startup Ideas podcast. if you come up with, a better twist on it. [clears throat] share that, too. And yeah, I think this is just so fun. So,.

Full transcript with timestamps

If you wanted to make money selling AI agents, how could you do it? What is the clearest path to building wealth selling AI agents? And what is the stack that you need? All the tools, all the skills, all the apps to put it all together. Well, today I brought on Nick from Orgo, and he basically breaks down one of the opportunities I think is the most slept on. In fact, I don't I haven't seen anyone talk about how you can take a personal agent platform like Grockbot, niche it down, and sell it to people on a monthly basis. So, today's episode is a full master class. It's under an hour, and the people that stick around are going to understand all the techniques, the strategies, how to price it, how to think about this startup idea. It's literally a sauce factory. So, uh, I'm excited for you to get into it. And by the end of the episode, you're just going to really understand all the different parts, all the different

tools, and I think that even if you don't want to go and build this startup idea, you're going to be way more AI native just by listening to this. Now, if this episode gets more than 5,000 likes, comments, and subscribes, I'm going to include a link in the pinned comment, [snorts] and that's going to have more ideas and more niches for this concept. So, go pause, go and do that and then hit play again and I'll see you at the end of the episode.

Nick, by the end of this episode, what are people going to learn? [clears throat]

People today are going to learn how to build their own custom Grockbot for a business and sell it for $25,000 upfront, retain them on the back end with a downell for like 3,000 a month. All in this video, we'll walk through how to build everything. How do you select the offer? How do you select the stack? How do you actually build the thing? We go through everything in this video and I can't wait for people to build this startup idea out. So stay tuned. So the commitment I need from you is there's a full stack of different tools, agents, um, open- source technology. I need to understand everything about how I can go and start my Gro Grockbot for XYZ because I really do believe that um, you know, selling AI employees Grock for XYZ is one of the biggest opportunities right now. So, I just need your commitment that if people stick around for this, you know, episode, which will probably be under an hour, um they will

have the tactics, the techniques, the strategies to actually implement this uh today.

Yes. And you will see it. You see it right here. Like I have these agents here. So, you'll have examples and ideas of the types of agents you could build for businesses. You'll have examples and ideas of the types of apps you can build for them. and you have a beautiful app already built for you that you could just fork and use for free and go sell it to a business ready for you all in this video. I walk through every step of the way. Um yeah, there's nothing left out on this on this one.

All right, let's get into it. Let's do it. So, I'm sharing my screen here. I want to show you this is a video of me. I'm going to mute it. I made earlier this month. And you still see this, right?

Yep. I'm literally showing a video on Twitter of this tool called BOPS that I built. I called it BOPS because I'm like, it's business ops. And I wanted to, you know, build agents that solve business operation problems. And you you see here I'm like talking on the phone to the bots. Each bot has its own computer with four screens. And I literally built this entire essentially Grockbot clone in a weekend. Um, I posted on Twitter, you know, got 75k views and it just breaks my brain that I was able to do this in a weekend. And then so we have existing businesses and customers at at, you know, Orga where we build them agents. And I think that this is such a cool startup idea because it's an evolution of like um so many different ideas, which is like with AI, there's a few ways that I personally have made money and and you can see my screen here, right?

Yep. So I I've made money with AI by building apps. So the first when when Cursor came out, funny enough, fun story, I watched the Greg Eisenberg podcast. You had Mickey on. He was teaching Cursor. I think this was like summer of 24, I think, something like that. Um, super early. Of course, you're always early to to everything. And and I I built my first money-making business online, building an app uh using AI. Then the second biggest way that I've made money from AI was with agents, selling agents to businesses, like selling a Hermes agent to a business for 5K a month. Now, the interesting opportunity that we're at today is building agents and apps together and solving like the entire gap in a business with agents and apps. And so, what why am I talking about apps?

So, Greg, have you heard about people who say apps are dead? Software is dead. Most people on X are saying that.

What What are your thoughts? Um my thoughts uh I actually just recorded like a solo episode about this. I think the people are saying that because uh they don't want other people to go and build things. So yes, do I think that software is evolving? 100%. Um, but you know, I think that what's happening is the the era of like dumb tools, dumb SAS, where it's like you have a human going and like pressing buttons and uploading a CRM uh and uploading to a CRM like that is ending. I think we're moving more towards like the price for software is trending towards zero and you're going to give away the software and you're going to upsell agents and human intelligence to go as as like the premium experiences.

That's my take on apps are dead, SAS is dead. It's evolving and people don't want you to know that there's never been more opportunity in this space.

Exact. I fullheartedly agree and like in my perspective it's like the era of generic apps

is what to me feels like is dying and the era of custombuilt like building your own CRM for your business that is like purpose-built for you like your taste that is more alive than ever and it's only the beginning and

you know that concept of like building that purpose-built apps, purpose-built agents for a business and for an industry. Oh my gosh. Like that's what I'm here talking about today when I'm like sell Grockbot to businesses. Yes. Build it. Build your own Grockbot for an industry like insurance or marketing agencies or dental clinics. Like customize it so that the agents come pre-built into that out of the box. each of them having their own phone number, email, um, channel connector, so you could talk to them on Slack, all of that, which I'm going to show you how to build in this episode. Um, and

so what you're saying is basically, hey, you have all these big platforms like Grockbot, Dots, whatever. And they're not in the business of these niche setting up these like niche use cases for businesses. That's one piece. And the other piece is the pe the the the owners of these businesses, they don't know how to set it up and they don't even they're too busy anyways, right? They're just like operating their business. So you're saying, "Hey, there's an opportunity to actually spin up your own version custom. You've gone and validated that. You built it in a weekend. And if you sort of add the right, you know, plugins, templates, it's going to be out of the box like a 10x experience for them because if they go and download Grockbot, if they go and download another one of these platforms, uh, when when they unbox it, so to speak, they're going to be so overwhelmed. They're not going to know

how to use it. And also like they might not want to like the idea of like calling like an AI transformation agency might even be too much for them.

Exactly. Like they just want their problems solved and and AI transformation agency they're going to walk them through all these processes. It's it's going to take them this much time. They're going to have to learn all these things. But if they could just install this thing into the business that is actively already solving all their problems. It has an outreach agent. has a, you know, agent that, you know, manages their community on school, like all these different things. And I can kind of walk through what I use mine for. I have that here if you can see this. Um,

so like if you just had that that you could just install into a business, it's like they're willing to pay so much money for it. So I built bops because actually I wanted to solve I didn't like how all the other uh agent platforms worked and I kind of wanted to marry some things I liked and some things I didn't like. And so I built it for myself and then I post that video and some our c our some some of our customers saw it and they're like, "Oh, like we want that." And so I can actually show you here like one of our insurance customers like right now we're in the process of building them their own insurance agency bops. And you can see here it's like custom branded. So like the computer here has their branding and it's like it's going to be specifically purpose-built for insurance agencies. Um, and so yeah, I can kind of show you like some of the cool things I'm doing with mine. It's it's really

cool cuz what people don't realize is you can like all the same tools that these big companies are using to build Grockbot or dots and whatever, they they make it available so that you can build your own version too. Like they they have their agents API and they have all the same, you know, the voice models and all that available to build on. And so you can actually have a better product in a lot of ways than what they're offering cuz you can glue together just everything that you love across the internet of like, oh, I want an email from agent mail or I want a computer from orgo or like this whatever and you can put it in your own combination of flavors and and really build your own better version of it. So um this is kind of cool. So I have my my Boppy is my chief of staff and I can email him. So, if I click like his profile here, you can see he has an email. He has a phone number. I could text him. So, if I text

him like right now, actually, let me say um I could say, "Say hi to the podcast." And when I text him, like it shows up here. It says, "Shi to the podcast." And he's going to reply and he's going to say hi. And then I

That's really cool. Yeah. And I could call him. So, I'm going to call him on my phone. Like, can Grockbot do this? I mean, I don't think so. [clears throat] So, I call him on my phone.

Hello. Can you hear me? Hey, Bo Poppy. Yeah, I can hear you. Can you pull up um can you pull up like a list of companies uh that I could sell to and do pull up their websites as well? And then also pull up the S&P 500. Let's take a look at that. And I also want you to keep a look at hacker news to see any relevant posts uh for me today.

Mhm. And so I tell I tell him to do that. You hear that? Okay, Poppy. And you can see over here you see on my screen.

So it says what I say. So I'm going to hang up. So I I end the phone call. And now he's going to go ahead and he's going to do all that. And so I literally That's a phone number. I called him on his phone number. And now look, he has his own computer and he has like four screens on this computer that he can control and each bot here has their own computer with four screens and he can multitask and he could do all these different things at once. And so it's like to me this is like a superior product and I was able to build it as one person. Now, obviously, I have my own experience and collection of ideas I put together to get here, but like genuinely, I'll show you how to do this.

Like, Opus 5.5, you can build anything. Like, you're 323 prompts away from building your anything. U in my case, it was it was bops in a week in in a weekend. Um, so yeah, just a a quick demo on that. Yeah, I think like it's so obvious that if you were to install a Grockbot for a business, especially like a SMB, they would want a phone number attached to the bot the bots, not the the bots, the bots, uh, you know, so that they're they're able just to quickly voicemail, call them, text them. It's so obvious, right? Um, and it just goes to show you how early we are in this whole personal agent era that it does feel in some ways like so hobbyist. You do have to like figure you you have to like find out how other people are using it and and then like copy it. You know what I mean? So if you I mean the thing here is you have to understand the ICP really well and and you have to understand their pain

points and you have to understand their bottlenecks and you have to come with them with solutions.

Exactly like you really need to exactly that like and you can see here I even have a managing my Twitter and everything. Um you need to understand the industry specific problems really well for you to even make agents that are useful. Like I have a bunch of agents here. I actually a quick tip for people like I personally build an agent for every platform. So I have a Twitter agent. I have an email agent. I have a school platform agent and each agent owns this platform and is like logged in and is able to automate tasks and routines relevant to that platform for me. Uh so for instance, my LinkedIn agent can outreach to to uh prospects that are relevant for you know um for our industry and people that I might want to sell to and connect with them and then once it connects with them then send a message after the connection and then set the booking uh appointment, get the calendar booked appointment, then I

jump on the call. I have the brief already. Like that's how I use these things. I have a whole bunch of them here. Um, and so yeah, that's just a quick tidbit on like, you know, how to kind of divvy up an agent for what use case. Um, so that is kind of the uh hopefully the value proposition can be, you know, explained by just seeing that. And now I want to kind of walk through how the hell did I build this in a weekend? How can you do it too? But what is the kind of offer that I would take to a business owner to charge $25,000 upfront? You know, maybe you collect half of that up front and then the other half at the end of the implementation of this. Um, but let's go through the entire uh buildout process, shall we?

Let's do it. So, first things first, I want to kind of walk through how I build in general cuz it's a principle thing. You don't need to think too much about prompt engineering these days when you use something like Opus 5.5 on high mode. I mean it just the model is just so smart and intuitive that it kind of just gets it. So um but there are principles that allow me to build well and fast and I would like to walk through those. So the first things first is um well obviously you actually need first things first to decide well how if I'm going to build something like where am I going to build it and so you can build inside of claude code you can build inside of uh T3 this the guy on Twitter he has his T3 code I actually built bops inside of T3 and I'll show you here uh this is the the the exact conversation that I had um you can see it is literally a single long running thread inside of T3.

I I'm not even kidding. Like I did this I built this like an idiot. Like I literally just kept going in one single long thread. I didn't even like connect it to GitHub or anything. I didn't push pull like I just did one long thread for 36 hours. I had loops running overnight and that is how I built this entire thing. So you use something like T3. You open up a project. you select your model like opus 5.5 and the the process is I always start with research. So if you look at the very first message I sent inside of uh building this this project of bots, I literally was like I want to build a version of Grockbot or Meta's Muse or OpenAI's dots very similar research them so you understand first. And so what it does is it goes ahead researches all of them to conceptually understand. And so what I'm trying to do here is I'm trying to frame and prime the model to understand what am I even talking about?

What is it the thing that I'm even trying to build to begin with? And so it researches all of these. It has its sources here. So I I know that the model knows what I'm talking about. And then the next thing I do is I design. So have you you're are you familiar with paper, Greg?

Big fan. Not affiliated but big fan of it. I'm also not affiliated like the only thing I'm affiliated with any all these tools I'll tell you is is Orgo. Obviously I'm a co-founder of Orgo. So we we build the computers so that you could build your own Grockbot. So if you want to build your own Grockbot, you could use our computers. But paper is like Figma for the AI era and I designed everything inside of paper. You simply give uh your coding agent. So, inside of T3, I have Opus 5.5 and I installed the paper MCP. And what that looks like is this. And so, I I did that research step first. Now, my agent knows what what I'm trying to build. And then the second thing I did was I said, "Okay, let's design this." Before building anything, I want to design inside of paper what this would look like. And so, it comes up with these designs.

And I'm like, "Okay, okay." And I just kept going. I kept t like back and forth back and forth. I want it to look like this. Plain English, plain, you know, natural language. I'm telling it what I want. And you can see the evolution of how this is all built out. This is before any building. Just all design. Is this is this adding up?

It's crazy, dude. No, it's cra It's just It's insane because the also like the the branding is so good. And I feel like paper paper actually low-key excels at at branding.

I don't know what they did to do that, but it's genuinely true. Like they they've I don't know what they added in the MCP or their app or whatever. Like anything with Opus 5.5 and designing inside of paper comes out like an S tier product design. Like it's insane. Um and so yeah, you can see like this whole like it's a long process. I mean I I spent a lot of time thinking of like the simple design and you can see I landed on this like

Applelesesque you know like you're familiar in iMessage and FaceTime and all of this like I landed on that kind of design very you know inspired by Grockbot as well um and that's before any building so that's the second step so first step we research second step we design

quick question on research and design. Yeah. Um T3 code, why did you use it versus say claude or or or something? Um and any tips there? And then my second question being paper. Uh obviously we like paper, but like what are some best practices in terms of using paper um to get the most out of it?

Yeah. So I chose T3 genuinely out of curiosity. I was like, I had seen this guy on Theo. So, yeah, if you if you guys aren't making content yet for your product, like do it because eventually people will see you enough when you make so much content that they'll try your thing. That's how I tried T3. Um, and I actually So, my take on it is it's very good. It's very simple and you could see it's a very simple design. Um, so I I recommend it. It's free and if you already have a Claude subscription, you could just use it inside of T3. It's free. Uh, and I think it's open sourced.

So, um, yeah, I just I like things that work and are simple and and they don't get in my way, you know. Um, so that's my take on T3. And then as far as paper to get the most out of it, this is a principle uh, it's a design thinking principle, which is um, diverge and then converge. And so what I mean by that is inside of paper you'll see I many times will ask it to generate me like multiple variations. You can see here multiple variations of the same kind of feature idea that I have in mind. And I ask it to give me you know three four variations. And then that's the diverg part. And then I converge and I say okay I like this part from choice one and this part from choice three. And then that's how you're able to really get the most out of something like paper or any just design process in general. Diverge and then converge. There's a a a hack there. Um so going further along into the stack, um after I research, after I

design, you have to choose now what tools you're going to use to build the thing. So for bops, there's a few different components. And it's worth kind of like thinking in this like in this way of uh just like as an engineer like you know engineer thinking of okay I have a phone I have a computer I have an email like where are all these different things coming from and so you have to decide for me I wanted to keep it simple I didn't want to build this infrastructure myself so obviously like I use orgo I'm a co-founder of orgo we build the computers so you don't have to build them yourself You can use them to build your own bops. Uh I used agent mail to give it an email. I used agent phone to give it a phone number. I used composio to give it these connectors. So you can see here when I go to uh login and apps, I'm sorry, apps is the the bots can actually make you know apps. Uh when I go to login here, I can add a

connector and I have all these connectors out of the box from this tool called Composeio. Literally just one one API connection to Composio. I get all of these connectors for my application and then I can connect to them from here. So all my bots can have access to that. Um so deciding the stack that you're going to use after you design it is is the next part. It's like, okay, I always tell it to read the docs and read the API references. Those are the two things I do with every kind of feature. So, I'm like, okay, if I'm going to build the Orgo computers instead of bots, read the API references for Orgo and read the docs. That's what I tell my agent to do.

Next up, after it builds it out, I literally Do you use CleanShot? What do you What What's your What's your screen? You use Clean Shot?

Of course. I've been on it for years. also not affiliated, but love it. You strike me as a cleanshot guy, Greg.

I mean, it it really it's one of those uh upgrades on your Mac that once you use it, you cannot go back.

It's in It's just simple. It just works, right? Yep. So, CleanShot, for those who don't know, it's a screenshot app. It's just better than the native screenshot inside of your apps, inside of your Mac. Um, so I literally will screenshot, you know, the application as the agent is building it. If there's something I don't like or if there's a bug, I will screenshot it and I will send it to Opus 5.5 and T3 and I will say fix this or I don't like this or can you make this more like that? And so screenshots are awesome and and and many times I'll screenshot things from like Grockbot that I like like oh I like this UI element of Grockbot. I'll screenshot that. I'll send it to my agent. It'll implement that inside of my app. Okay, let's wrap.

The screenshot bit is like in general if you're building anything, you want a swipe file. Uh, a swipe file is a place where you have references, screenshots on copy and design, and then you'll feed those, you know, to to the LLMs or to paper to help get the most uh out of it.

Yeah. Yeah. And I I if Yeah. If you don't have a swipe file for like things that you It's just a life hack. You should really have a swipe file for everything. Like I'm talking marketing. If there's like posts that you like that you get inspiration from or apps that have cool UIs that you you really get inspired from, it's good to get in the process of creating like Greg said like these swipe files of just inspiration like your own Pinterest feeds, you know. Um last, let's let's wrap up quickly on this. I don't want to spend too much time. the bot life. Uh I talk about that, you know, give every agent its own email, phone number, computer. Um things to make this thing these things actually useful. The the cool thing when you build it yourself, like inside of Grockbot, you don't actually get a new computer for all of the bots. I don't know if people know this, but in Grockbot, it's actually each bot gets

its own screen, but it doesn't get its own computer. And so when you build stuff yourself, it's like you can give each bot its own computer for privacy reasons and just multitasking reasons and all of that. Um, so give it the things that'll make it useful. Power it up, boost it up. Uh, decide where you're going to host it. You can actually host it on orgo. Orgo is like the computer, but it can act as the server. Uh, you could host it on Fly, you could host on Versel. And, uh, finally, just ship fast. Once again, this this took 36 hours. I made a video on it. I'm literally eating cereal in the video.

So, like this almost did as good as our YC launch video and it took me two minutes to make. So, you don't need VC funding. You don't need a lot of money. You need a weekend and enough inspiration to go build something really freaking cool and just show the world. And in this case with like if you build your own Grockbot for an industry, you you have two options. Let's let's dive into the how we're going to make money from this and what's the offer and what's the stack. There's one bucket which is like okay we can either sell this one to many or we can sell this um like one to few which is like selling to the SMBs. Right now Greg if you were to sell this to SMBs like I I mentioned we could sell this for like 25K upfront and probably do like a 3K a month retainer on the back end to maintain it. Would that be an offer that makes sense to you? Would you agree with that?

Um, I mean, you know, I'll play devil's advocate as if I'm the customer. I mean, let's say I own a dental clinic. Like, my question to you would be like, how am I going to be able to make, you know, $36,000 plus 25K, $51,000 of value from from this?

Yeah. I I would I would actually just say I would like I would pull up I would pull up the thing that I built you and I would look at one of the bots and I would be like, "Greg, how much how much do you pay somebody like a sales rep to well I guess in your case if you're a dental clinic? Uh how much do you pay your your front desk lady to do all the email and support, all of the billing, all the calendar bookings and everything and appointments and all of that? How much do you pay her a year?" And you're probably going to say, "Well, I pay, you know, 50 or 60 or $70,000 a year for just that one person." And I'm going to show you. I'm going to look. I'm going to say, "Look,

there's that one person right here. And that's just one of the 15 people, 15 bots that I'm giving you here today upfront, done, ready to work."

Um, and you own it and you own this system. And I think like that's the key is when you try um in the past like you can sell agents right for 5k a month as like a Hermes agent to a business but there's a lot of frictions um that a lot of people run into with that model. And what I've seen work really well for a lot of people is like oh they don't business owners don't want to be feel like they're held by the neck on a retainer for these AI employees. So when you sell them this system, you're like, I'm going to just build this system. I'm going to release it into your business.

It's 25K. You own it. And then you downell them after that after a 30-day implementation. And you're like, okay. Cuz they're going to want your help to maintain and things like that. And that's completely optional. That's where the 3K a month comes in. But when you sell them this concept of like they own it and they can have it forever, that seems worth it to a lot of businesses. My take is here's how you can turn that offer to make it completely irresistible. So, a lot of these companies that, you know, we're talking about targeting, their websites aren't super optimized.

And we've shown in this episode how you can create just gorgeous, gorgeous designs uh for apps, websites, and things like paper. Here's what I would sell. I would say it's going to cost you 25K upfront and it's going to be $3,000 a month, but I'm going to include a fully redesigned website with that.

Um I So just completely free. So I think like and then you show and you you have it done for them, you know? It's like this is what it's going to look like. Um and and it looks beautiful. And so they're like, "Okay, you know, we've been thinking about redoing our website." Like, "Yeah, would we pay 25K for our website?" Yeah. So, you're basically getting that for free and then you're just paying $3,000 a month. And the $3,000 a month, I think what's so interesting about this whole like model is um they're considering it as like a higher.

So, like their mental model is $3,000 a month. Well, that's, you know, I don't know, $12 an hour or something like that, you know. So, I think uh that's the other way to think about it in terms of pricing is to actually be 25K up front and to be like $12 an hour.

Yeah. Yeah. That's beautiful. Like essentially going back to that concept of well software is so easy to build and we can build it so fast. We could use it to strengthen our offer and use it as a weapon of like let's just build all these assets and just bundle them in for free or as bonuses, you know.

Yeah, exactly. The other thing you can do to make it irresistible is to say like you you find out what their bottlenecks are and what their pain points are and you could say if you aren't able to get back to people quicker if you aren't able to you know text people you know via the via this like you know you figure out those things and you say if you aren't I will give you back the 3K a month or whatever or like so there's something And it's like, you know, it's no-brainer and you know, you're going to be able to do those things because you're you're setting up the the Grockbot, right? Um, so you just you do those. So it's like, okay, so you're basically getting a free website.

Um, and then you're getting this higher and then you're validating that you can actually do those things or they get their money back.

I love it. I love it. Yeah. There's so many there's so many freaking like value ads that you can give today with AI. Like so many you could build them lead magnets for that you could build them a a dental, you know, build a calculator lead magnet into their website for their dental practice or their orthodontist practice to show so show their clients how much their their braces would cost or whatever. Like these are like lead magnets lead magnet tools. You could build them include those for free. Like there's so many opportunities to just add value today for such little cost and time that it's like

selling shouldn't really be the hard part here. Um so and and so what about this Greg? Like what if I build what if I don't want to sell to SMBs uh a custom Grockbot? What if I actually want to take a custom Grockbot and sell it to an industry? like maybe I do want to build the Grockbot for insurance agencies and have it sold, you know, as a one to many tool. Well, Grockbot, they're charging 200 a month. But if I'm solving this for an industry specifically, how much pricing power do I have? What do you think?

I'm going to say it's 10 to 20x. Yeah. in terms of like you know in general like the niching down like that's another there's a few reasons why you want to niche niche down in general. One is you have more pricing power. The other is you have less competition. Um so I think you you command a higher rate for sure.

Got it. Yeah, I agree. And it's like and you could charge you could do seatbased pricing and have a bunch of expansion revenue and like and and I just keep going back to this point like the crazy thing is this doesn't need like you don't need funding for any of this like you can just like build this stuff like today you know um and it's so cool and so like in here I could even tell you know these bots to like build a website for a dental clinic. Um, let me see what's going on here. Hold on. Let me reset this real quick.

Let's give this a second. There's a bug. So, in here, I could say, "Build a website for a dental clinic." Um, and load it up, please. And imagine just having this already built into your Grockbot for that business out of the box. It's like it comes with a a website spun up for them. All these bots, you know, a bot that manages the website. I actually have one as well that I have literally a bot that manages my personal website and updates it every week. Um,

like if this is just out of the box ready for them, I mean, yeah, to me it's like it's a no-brainer for a lot of these business owners. Um, and you can see just how fast this thing will build it. Um, while this is cooking, go ahead. Yeah, it's just one of those things that I think that if someone actually wants to go and and run this idea like Gro Grockbot for XYZ, I think uh in-person sales, like you kind of need to see it to believe it. You know what I mean? Like once you see it, you're like, "Oh my god, yes." Like, "Why don't I have a someone managing my personal website?

Why don't I have someone managing customer growth? Why don't I have someone managing LinkedIn?" Um, but you know it you have to see it.

Yeah, you need the inspiration. People don't realize how many things you can actually automate. Like they they don't they need the inspiration in and of itself. And like people like us, you know, we're using these in our daily lives.

Yeah. So much that we have all these ideas for, you know, ways that these agents can be useful. Um, so yeah. um while while this continues to cook. So there's there's those two options there of like if you build your own Grockbot, you sell it to businesses, you can sell it to few and sell for high ticket or you could sell one to many still high ticket uh but just in a more productized approach way. So it's like super scalable. Um, and just like recapping on on on the point here of like when you solve the entire gap for a business, it it commands this pricing power, you know, and you and when you add like these guarantees and you just make it irresistible. Um, that's how you're able to charge high prices. And and if you don't charge this, like a lot of people might have this constraint or like self limiting uh maybe doubt of like, well, if I was able to build this in a weekend, how valuable is that?

You don't know how many people Well, did you see that Andre Horowitz post of like 2% of the US pays for AI tools. So, if you're building with these things, you you're taking it for granted how competent of a skill that is and how valuable that is.

And um yeah. Yeah. Like the vast majority of people haven't heard of Grockbot. They don't know about computer agents. They don't know that you can do this. So, and that's going to that 2% is going to go to 4% and 8% and 12 and 18 and 20 and 25. So, like I agree. I think there is a a limiting belief that some people have around like, well, if someone can create this in a weekend, there's no moat, you know? Um, and that's what they want you to think. They want you to think that, you know, you know what the mode is like? Creating value for customers.

You know what I mean? Like if you create value for for customers, like they're going to stick with you. As long as you create value for customers, they're going to want to stick with you

and they want you. Like I think a lot of people underell themselves. Like there's something specific and unique about everybody. And and when a business decides to work with you, they're it's not it's not always so much that they're making this decision entirely on the product or service that you're giving them. They're also making it a huge portion of that decision is on you because they like you as a person. They want to work with you long term and like so much of this is also like just a relationship thing, you know. Um they want Greg Eisenberg to be the one to build them their their their personal company Grockpot because they like Greg Eisenberg, you know. Um I think people should keep that in mind, too. Like there's this human element of business that's never going to go away. Um

absolutely. So yeah. Um, so while that cooks, is there anything else? So we went through, okay, what is the the startup idea, the the opportunity? It's like sell Grock bots to businesses, build it customized to their industry. How do you build it? We walk through how to actually build like what's the actual strategy when you when you prompt these agents and what is the tool stack that you use? What are all the different components? We talked about that with giving the agent its own email, its own computer, phone number. Um, and then finally, you know, talking about the offer, how do we sell, how do we monetize this? What's what do we charge?

Do we sell one to many, one to few? It depends on the founder type. Um, and what's the value proposition? What are some other constraints, Greg? Am I missing anything that people might hold people back from from going and building this opportunity from them for themselves in a weekend?

There's only one thing which is like okay Nick like I I see I see the stack. I understand all the all these, you know, I see this and but it's like how do I take this and actually bring this to Willow or Willow Dental to actually like how am I supposed to know which bots I need to create? How do I figure out what the pain points are? Um what does that look like? What does that process look like?

Yeah. The so the biggest thing is like deciding your niche is I think a very difficult problem for a lot of people. I and when people get started in business here's here's a common thing that happens. You'll you'll you'll get super amped up. You'll say okay you know what I'm going to start I'm going to make an agency or I'm going to start selling services or agents or crockbots to businesses. Okay. and then you go online and you watch videos on on business education and the first thing you're told is pick a niche and it's like okay well then what's my niche and for a lot of people that might be obvious for some because they're like oh I spent this many years in in dental or I spent this many years in insurance but for a lot of other folks they're like I don't know where to start I don't know what niche to build so you know what the biggest problem that entrepreneurs make is they'll pick a niche they'll start on

that niche they'll build a website they'll build all this messaging and offer and position and they'll make all this content towards a dedicated niche and then after a week they'll realize they don't care about that niche at all and then they have to start all over again with another niche. So my point here being the first step you should do is don't worry too much about picking an industry or niche like right away. What I would do is and this is what I did. I'm speaking from personal experience. I would go through my phone and find the 50 people I know who either are directly like running a business or they're maybe they're friends with somebody who runs a business or they're just like high up at a business. And I would just call those 50 people and I would tell them like, "Hey, here's the problem I'm solving.

I'm building a whole team of agents that can help you run your business and a bunch of apps and like a website and everything all in one solution. I'm doing this for businesses. Is this something that you think you guys could be interested in? And then I would, you know, if out of those 50 people, you're gonna get a few people who are interested. And one of them might be in this industry and one of them might be in that industry. And then I would just do the first one for free just to get a case study of like, okay, asking them a bunch of questions, seeing if we could actually solve these problems for them.

And then once you do that and you have that case study, now you might be at a point where, okay, I like this industry. It might be an industry you've never even heard of. Like for me, one of my first one was like a promotional product industry. I didn't even know this industry existed where they make like swag merch for for companies. I was like, "Wow, really interesting industry. Big, untapped, it seems." And like I would have never found that if I didn't just be open-minded and just like reach out to the people who were receptive and and just work with that. And so I guess my point being same way earlier how I said diverge before you converge. Same thing is true with like industries and niches. Just try test out a bunch of them. See what the market appetite is with people that you know and build for them. And then once you're building for them like just genuinely sit down. Don't try and scale right away. Sit down with

that one first client. Try really to solve their business. Watch them, shadow them, ask ask them to like share their screen and go through a process of like something that they do manually like creating an invoice or you know creating products into a presentation and sending that to a customer. And you will see how many specific automations and problems can be solved. and it will be very intuitive hopefully for you to build those agents and applications out for them. So

yeah, couldn't agree more. The one thing I'd add is if you are going to go and do a free case study for someone, say a dental clinic, one thing you can do is you can say, "Hey, I normally charge $50,000 a year for this service. I'm going to do this for you for free. All that I ask is if this works, you can introduce me to three of, you know, colleagues in in in in the dental clinic space in the local area or the next town over and just give me an intro via text saying like, "Hey, you know, Nick did this thing for me. It, you know, it was successful. You should consider it." And

I love that. Yeah. And and to make it even easier for them, you can say, uh, and I'll give you, you know, a $5,000 credit, so it makes them look good or something like that.

Yeah. Yeah. I'll credit this towards your next month if they're on like some retainer with you or something like that. And and you know the beautiful part about this whole process that you just described too, Greg, is like at no point ever are you selling cold because you're always if the first person you know is someone you know

and you do it for free and then they refer you to someone they know, you're always warm. And that is the key I think with sales is like you never want to be on a call with somebody who doesn't know you and they don't know who what you sell and it's just cold. It's like it's like dating, you know? You want to I want to meet you through a friend. All right, I want to meet you through some kind of warm mutual interest or connection. That's absolutely true for sales. And when you do it like you just described, you're always going to be selling warm to somebody uh who has some knowledge about you and what you do. So,

I also think that for for many people, including myself, like selling is icky. like you get kind of this weird feeling, but it doesn't need to be that if you're selling something that is so valuable um at a price point that is essentially unheard of that you're teaching them um something that they're really interested in learning about. So there's so many reasons why selling could be this helpful thing that is like they want in so many ways. Um, so I wouldn't again like limiting belief like selling isn't icky at you know if if you're building something like this and that's why I wanted to bring you on to talk about this because to me this opportunity is a really big one. It's one that you know could could create value, you know, right away. Um, and I think it's one that when you show people the power of this, it it really just like you can see their the dots connect in their brain.

Yeah. Yeah. I think this is like the most exciting time to be building honestly like services or just like any kind of business, whether you're building a productled business or a serviceled business, an agency, you're selling agents, like whatever kind of business you're building. I'm like, Greg is rooting for you. I'm rooting for you. Like, it's the best time ever to be building. Uh, and yeah, super was super excited to share this startup idea with everyone. It keeps me up at night. Um, so I hope other people go out and build it, too. And, and the beautiful part is, Greg, this entire thing that I built, if you don't want to take the 36 hours to, you know, build your own, you could fork it. It's on GitHub. It's you literally go to like uh the you can go to bops.

We'll include it in the show notes in the description so people can can click it.

Yeah. Go to open source tell point your coding agent to this repo. Fork it. Customize it for yourself. Go sell it. Go make a lot of money. Good luck. Have fun. Um

and let us know. Yeah. And let us know. Yeah. Yeah. And this is the Startup Ideas podcast. if you come up with, you know, a better twist on it. [clears throat]

Um, share that, too. And yeah, I think this is just so fun. So, thanks for having me on, Greg.

Nick, thank you for being generous with your time, your sauce, your ideas, your tactics, your stack. Uh, and I'm happy you open sourced this. Um, I could speak to you for hours. I would love to have you back on. So, you know, comment below what you what you want Nick to come back on to to talk about or just anything in general. Let us know. Um and uh I'll include links for where to follow Nick on the internet. Um and uh yeah, hope to see you soon, my friend.

Thank you so much, Greg. Appreciate appreciate you always having me on and um yeah, love jamming with you in the audience. So guys, go out there, build some cool things, and um we'll see you next time. See you next time.

Questions about this episode

Who speaks in "The $1M+ Solo Grok Bot Business (Full Course)"?

Greg Isenberg (Host), Nick Mosovich (Guest).

What is "The $1M+ Solo Grok Bot Business (Full Course)" about?

This episode unfolds a practical masterclass on building and selling AI agents as a scalable business, using a Grockbot-style personal agent platform as the launching pad. Nick from Orgo breaking down how to niche, price, and package a full-stack solution—covering offers, tools, and workflows to land $25k upfront plus ongoing $3k/month retainers. Listeners will come away with a clear blueprint…

What is mentioned in this Greg Isenberg episode?

32 people, works and ideas are mentioned, including Grokbot, Greg Isenberg, Nick Mosovich, A.I., Bops, Opus 5.5. Each one on this page links to its own page with every episode that mentions it.

How long is "The $1M+ Solo Grok Bot Business (Full Course)"?

The episode runs 50 min; the transcript is about 9,044 words in 94 timestamped sections.

More from Greg Isenberg

About this transcript: the episode's spoken words as captioned on YouTube, split into timestamped sections. Reclip.Me locates every person, work and idea in the text, keeps the time it is first named, and quotes the words around it verbatim. Last updated November 5, 2026.