Episode 809 | What I Learned Diving into A.I. for 100 Days (with Craig Hewitt)

Rob Walling:
AI is completely changing how people discover brands and content online. And hres has built a full blown SaaS marketing platform to help you stay ahead with over 15 years of real world web data. An AI that actually understands marketing. Hres helps you measure your brand presence, build authority, and monitor reputation across search, social and AI platforms like Chat, GPT, Google AI overviews, perplexity and more. You can also dig into what’s driving your competitors’ visibility and spot market gaps before they do helping you create content that ranks and drive new traffic to your business. There’s no need to juggle a bunch of disconnected tools. Get hres all in one platform to make your brand unmissable in a fast moving world. Try it free@ahhfs.com slash a wt. That’s the letter A-H-R-E-F s.com/a wt. This is startup For the Rest Of Us. I’m your host, Rob Walling. In this episode, I sit down with Craig Hewitt, the founder of Casto, many time guests on this podcast, and we dig in to his 100 days of AI series that he did on YouTube. 

So we talked through the things he learned from that series as he dove into all the tools that you hear people talking about, and frankly, I wanted to find out from him which tools I should consider using. And by the time we get halfway through the episode, he kind of blew my mind and basically says, yeah, no one in our space should really be using chat GPT. There is a much better tool that can do all that and more. And I’m going to make you listen to the episode to find out which tool he recommends. Before we dive into the episode, MicroComp us. Our next flagship event is April 12th through the 14th 2026. It’s in Portland, Oregon. We’re putting together an amazing speaker lineup. This event will sell out. We’ve sold out our last four or five events for the last few years, and this one, especially given the reach and the depth of our audience in the Portland and surrounding area, we will definitely sell this event out. So if you want to attend, don’t make the mistake of missing out on your ticket microcomp.com/us. And with that, let’s dive into my conversation with Craig. 

Craig Hewitt, welcome back to the show.  

Craig Hewitt:
Rob. How’s it going?  

Rob Walling:
So good man. And you are bumping, you’re increasing by one. Your number of appearances on this podcast. You’re catching up to Derek Rimer who is the current leader. I think he has  

Craig Hewitt:
21  

Rob Walling:
Appearances and I tweeted out a while back about the number of appearances you were in that list. Do you remember? You remember your number?  

Craig Hewitt:
I don’t remember my number, but I think it’s a little bit of a cheat if you put the TinySeed tails on there. I think I ran the table for a while, so you did. That’s a little cheaty, but honestly, if I wanted to talk to somebody, super interesting, Derek would be way up there. I don’t know if you’ve had Jesse Hanley on the podcast.  

Rob Walling:
I heard you interview him on your show though.  

Craig Hewitt:
Super interesting. Dude, he’s in Japan, so time zone would be tough. And then Jordan’s been on a lot and Reuben, I mean, yeah, you have the people on who I would want to talk to, which is why I think everybody listens, so hopefully I can fill the bit or whatever today. Yeah,  

Rob Walling:
Indeed. And for folks who aren’t familiar with you, you are the founder of Cast Doses. Your H one is from audience to income Cast doses makes it possible. Transform your podcast from a hobby into an income earning machine, your podcast hosting plus production. You have a productized service under that, and you were season one of TinySeed Tails and you’re now doing low seven figures of a RR. You’ve been on the show many times and I was going to bring you in today honestly to to-do listener questions, but then I realized it would be such a wasted opportunity to not talk about ai. And I think there’s a certain group of folks out there who really want to hear more about it and have it kind of broken down a little bit more and hear thoughts. And there’s a certain group who’s going to delete this episode the moment they see AI in the title, but for them it’s just fine. You have recorded 100 YouTube videos, you did 100 days of AI on your YouTube channel, which is a staggering accomplishment. And I think I do want to talk to you a little bit later on in the show about how you possibly got a hundred YouTube videos done in that amount of time, but realistically high level, you start this a hundred days of ai, what did you learn that you were super, super surprised that you learned in that a hundred days or any high level stuff?  

Craig Hewitt:
Yeah, yeah. I guess just for kind of context on my podcast, which has been on break, I think fairly so because I’ve been publishing YouTube videos for three months straight. I had Brendan Huffer on, you probably know Brendan, he does Growth sprints, super interesting guy. He was telling me on our podcast interview, he kind of came on the scene as he said because he did a hundred days of SEO and this is about 120 days ago. I was like, that is a freaking great idea. I was going to do sales. At first, I was like going to do a hundred days of sales, I have sales background, and I was like, you know what? AI is the topic du jour. I should do this. And so I did and it was super freaking hard. We can talk about the production and the process and all that kind of stuff, but what I learned at a high level, it’s incredible and not there all at the same time. 

What I mean by that is it’s all the stuff you see on Twitter. It’s the, oh, I automated my entire life with AI and this N eight N workflow with 87 nodes and all this kind of stuff. That a hundred percent exists and I’ve built that and it works about 80% of the time. And so it’s that 20% that is still the human in the loop or the human augmented or amplified experience with ai. That’s where we are today. And that applies to everything that applies to voice and video and code and marketing and sales and data. In my experience, it’s not taking any of our jobs entirely today. I think at a macro level what’s happening is exactly what we see. Like Amazon this week, 14,000 people, they’re laying off because you just don’t need that many people to do the same amount of work.  

Rob Walling:
And as founders, that’s a benefit to us because that’s always our time is always the most restrictive thing versus at a large or if you’re an employee of a large company, that’s not a good thing because you’re going to lose your job. So that is the dichotomy of this in the a hundred days of ai, and I heard you talking about this on Jordan GA’s offsite podcast. You talked about a few categories of tools and what you consider the best in class for each, and there’s coding and there was I think just a general purpose kind of chat interface and I especially want to stop and talk about, you have some very opinionated thoughts about chat GBT, which I think are great and then maybe some other, I don’t know what all the categories are, if there’s three or four of ’em, but you want to talk us through that. 

And here what I want to give people an idea is this podcast, the single topic that it focuses on is how to start and grow an incredible company that changes your life. And if you want to go raise funding, you do that and still listen to this podcast because everything I say will apply if you want to use AI or not. Well, everyone is eventually going to have to use AI as founders, but it’s not that this podcast is going to start focusing on that, that I want to educate you as a founder or someone who’s maybe on a founding team, the topics that are super relevant today and that you can’t, there are no books on this stuff yet because it’s changing too fast. There are some YouTube channels that are talking about it like yours, but even those, they’re out of date within a few months. I’m not saying a hundred days will be, but a lot of stuff that I go to seek out education I seek out about AI is out of date because stuff’s changing so fast. So that’s what I want to call out is we’re here to help founders hopefully have some stuff that they can hang onto and kind of frame their thinking around how can I use AI and what if I just want it handed to me the information. That’s what I’m looking for. So let’s go through those categories.  

Craig Hewitt:
So I think probably the easiest one is for code, Claude Code is the answer. Claude code is a bit of a misnomer because you can use it for everything. I built a really powerful SEO and content writing tool with Claude code and if you look at Lenny’s newsletter, he has some really cool examples of this, which is kind of where I got some of the inspiration for that. But imagine Claude code as the best analogy I heard is imagine just taking the rails off of AI and going bare metal with it. It’s in the terminal, you add files and context and all this stuff. So it’s like if you take the chat interface away from something like chat GBT and get to do stuff and plug things into it, that’s what Claude code. So Claude Code is an agent, it has access tools and memory and you can expand it and plug things into it like MCP, but it’s an incredible coding tool. 

So Claude Code, you might say, Craig, that’s dumb, it’s code X or Cursors, new composer model, whatever, some kind of agentic coding tool. Claude code is it for me, I don’t care for chat GPT for almost everything because it’s a consumer product. It’s the thing my wife uses to talk about the stuff she’s interested in. It’s the thing my kids cheat on their homework with. It’s like all this stuff, but the only place I see use in chat GT is in the gpt. So we’ve created some GPT for our teams and use with customers. That’s really cool. So GPT are kind of reusable bits where you do a very specific thing and you can share those and the shareability of ’em and  

Rob Walling:
You train them, you kind of train it and you say, here’s my book, or here’s a bunch of docs, here’s our whole knowledge base, and now you know that that’s  

Craig Hewitt:
How you  

Rob Walling:
Describe it.  

Craig Hewitt:
Yeah. Yep. Or I want to write this very specific type of marketing email for product announcements and you just plug in, here’s the page, go write the email. That would be a good use for A GBT. There’s a bunch of other ways you can do it. Like a Claude project can do the same thing. Probably the tool, if I decided Craig, you get to use one tool, it would be Manus and producer Ron messaged me last night. He’s like, holy shit, I just used Manus for the first time and it’s incredible. It’s incredible. So Manus is basically Claude code with a web interface. It’s an agentic AI tool that uses Claude under the hood and it’s incredible. It has a browser, so it’s a agentic. If you use chat GT’s agent model like hey, go to booking.com and book me a hotel in Rio for two nights. 

It would do terrible, but Manus would probably nail it, it can write code, it can create reports, it can do all this kind of stuff. And so specifically I’ve been in kind of product creation mode, how I use Manus as like, Hey, I’m thinking about this thing. Help me think through this and help me go do market research, put together a brief for me, help me refine my thinking here. And it’s not just the dumb back and forth with chat g pt, it goes and gets data and provides briefs to you and documents and scripts and all this kind of stuff.  

Rob Walling:
What LLM, is it using Claude? Is it using Claude? Yeah. Okay, so it is Claude  

Craig Hewitt:
Underneath. Yeah,  

Rob Walling:
So as someone, the only AI tool that I use personally, our team uses a bunch of stuff, pod scan and pod squeeze and things to prep show notes, and of course everybody’s using, I think a lot of ’em are using Claude. A better writer is what I keep everyone, everyone who use it says it’s a better writer than chatt, but I really only use chatt and I use a lot. I’m in it multiple times a day and I’m in it. Would Manus be superior for just chat interface stuff?  

Craig Hewitt:
Yeah, it is. So it has either agent mode, adaptive mode, which lets it choose or just chat mode and so you can use it however you  

Rob Walling:
Want. Fascinating.  

Craig Hewitt:
Okay. Yeah.  

Rob Walling:
Chat GPT. Oh no, I feel now I already feel outdated. I remember when I was using chay bet, I started paying for it the first time you could pay for it and I’m like, look how ahead of the ground. 

I’m not an early adopter. In fact, I hate being an early adopter. I don’t like when changes and I don’t like when it messed up my workflow. I was one of the first back in the day, let’s say 20, 25 years ago as a developer, we all built our own machines. You you got all the components, you did that. I was one of the first guys to be like, I’m just buying a Dell now. And people were like, whoa. Oh no. The dark side, you’re no longer technical. It’s like I’m tired of messing around with the printer card. That doesn’t work. I don’t have time for that anymore. I’m billing a hundred dollars an hour writing code. I don’t want to fuck with this. So I tend to want stuff that is pretty streamlined and just works and new stuff often is a little janky. So anyways, I feel like I was ran early on chat GBT and now I’m already falling behind. Man, this stuff changes so quickly.  

Craig Hewitt:
Yeah, so I’ll give you an example of something I did yesterday for one of our clients at CAOs production. So we have a client that’s kind of revamping their podcast and they’re having a hard time nailing the intro section, which is especially on YouTube is so important is you got to nail that first 30 seconds to keep people so that YouTube keeps promoting your stuff. I said to Manus, so by the way, my third one I guess would be super whisper or whisper flow, so I’m kind of voice to text amazing. Or you just talk into your microphone and it types for you. So I said to Manus, I said, go and get the latest 20 videos from Jay Klaus’s YouTube channel. Jay is a big kind of creator economy guy. He does really good podcast on YouTube, which as you know is hard. Get the latest 20 episodes, download the transcript and then analyze the transcript and create a rubric or a framework for how we can input a topic into that and you give me three different intro options and it did, it took about 20 minutes, but it did it and you just can’t imagine chat GBT doing that.  

Rob Walling:
No,  

Craig Hewitt:
You see it, it’s writing code, it’s launching 20 different browsers, it’s downloading this stuff, it’s aggregating it, it’s doing all this Python stuff. That’s the kind of stuff Manis does.  

Rob Walling:
I have run into limitations of Jet GPT where I’m asking it to do just a little too much and I can tell it just tries and it can’t. Oftentimes for me it’s like I have this MP three of something, I don’t have a transcript for it. You’re screwed. You just got to go get that transcribed somewhere else, go to Rev and then bring it over. And I’m like, come on chat GPT, you should be able to do this. This is not that hard. I’ll pay extra, but you should be able to do this and it can’t. I’m guessing if Manus can do that, I would completely change my workflow. Huh. All right, well I guess I got to change that shortcut. This man has to remove a chat GPT from I have so many GPTs already trained. This is going to be the hard thing is there is actually a switching cost for me at a certain point.  

Craig Hewitt:
Yeah, I probably can’t leave chat GPT, but I just don’t use it. I mean it’s 20 bucks, so it’s not going to kill me. But yeah. The one other I would really consider folks to look into is talking about agents or automations is some kind of automation or agent builder. There’s a bunch, there’s Innate N, which talked about, there’s Lindy, there’s relevance, there’s Google Opal. I saw Google came out with another thing today. I think however you slice it, whichever tool you use, you probably should be trying to find ways in your business to automate this stuff with either just make.com style automations or these kind of more AI forward AgTech automation platforms. And if you imagine if you just did one of these a month, where would you be next year?  

Rob Walling:
Yeah, tell me, you’ve used this term AgTech a lot on the show already and I have a mental model of what that means. I’m not sure my mental, how accurate my mental model actually is. So just in case folks are listening, maybe they’ve never heard that term before or maybe they’ve heard it as much as I have and they’re kind of like, I’ve never built an agent. I get the idea that it’s a, Hey, it’s a GPT that can then go do some stuff and is monitoring things, and so it’s more like, I think of it as more like a make automation and a Zapier type automation that has AI built into it, but I think that’s actually a limiting description based on what you know, want to talk us through just what agents are and how different they are than just a chat interface type thing.  

Craig Hewitt:
Yeah, so I guess we’re talking about three separate things. One, I would just call a chat bot, right? So chat g bt your claw.ai, you go, you log talk to an LLM. That’s cool. That’s really useful. I do it a lot. I think everybody does that. An AI automation would be kind of stringing together a bunch of if this then thats or loops or whatever with calls to an LLM by itself or an agent. And so that I’d call out an automation and so that’s make.com and ADA and Lendy, all these kind of things. An AI agent has three components to me. It has three components. It has an LLM, so it’s going to call Claude or it’s going to call chat GBT or it’s going to call grok. It has memory. So you think about a chat bot, have I been asked this thing before? Yeah. Okay. I probably don’t need to go call chat GT to find the answer. I can just serve the answer up and it has access to one or multiple tools, so email or calendar or your knowledge bases in a vector store, like a pine cone or rag thing or the web or whatever. So those are the three elements of an agent and an agent can be part of an automation and in the case of Manus, the chat interface is an agent because it has access to tools and has memory.  

Rob Walling:
Yeah, yeah, as you said to all that stuff. Yeah. Here’s the thing I heard, I think it was Andrew Wilkinson on my first million three, four months ago, and he was talking about agents and they’re like, he’s talking about agents. I’m like, great. And he has all these agents that he’s built and I listened to it and I was like, none of these do much. These are toys and I am not going to go through the time to build them because they don’t win me anything. It was like before I get on a call, my agent looks at who’s going to be on that Zoom call and it goes and researches online and then gives me a bulleted list of kind of a bio and a this and that, and it’s like, okay, that’s fine, but I don’t need that. And if I did, I guess it’d be interesting that it would save a little bit of time. There were three or four examples like that, and I’m like, I’m not sold on this idea that it’s going to be worth the building and the maintaining of these things, but I will be sold eventually. I know I will, but I’m just waiting for that. There are a few, an killer example or two of an agent that someone’s using that I’m like, oh, I have to have that. Do you have any examples like that that you’re using or that of?  

Craig Hewitt:
I think a good example of an agent is a customer support agent, and so we absolutely do. We implemented it six months ago. It’s called Doc Spot and it’s kind of from an indie hacker guy. It’s amazing. It’s cut our customer support burden in half at CAOs and we have 4,000 customers. So it’s like a lot of customers every day. It diverts handfuls of conversations from our support team. All it is is it takes our entire knowledge base and most of our website and puts it into vector stores. Vector stores, like easy way for AI to look up the right things and then people ask you questions and it gives them answers. That’s super valuable. We pay 80 bucks a month I think, for it, and I would pay 80 bucks all day to have two salaries essentially and let them go do more interesting things.  

Rob Walling:
Yeah, that’s a big deal. And of course the first question any founder thinks is, but is it pissing your customers off? Is it doing the endless when I’m on the fricking Verizon wireless chat thing and I’m like, you so angry right now. You’re such and such. Yeah, you, such and such. So have you. I’m totally when John, John, I’m going to be dead way before John Connor leads the uprising because of how annoyed I’m at chatbots, but how do you keep it from doing that? How have you kept it from being annoying?  

Craig Hewitt:
Yeah, so I mean I guess first of all, Laura Rotor suggested this tool to me. So Laura, I’ve been on the show a few times. She was the one that turned me onto it. It has an escape hatch, so in every response it’s like, I want to talk to you human. There’s a button at the bottom, and if you do that just for us, it dismisses that and then pulls up the Help Scout contact page where you can send a ticket into the team. So we let people get out of the chat bot mode really quickly. Then our team monitors all the logs and all the conversations from the chat bot every week. Basically the cool thing is it’s made our documentation better because the chat bot only references the docs. They’re able to see, oh, it answered this question wrong. The docs are actually kind of vague here or misleading, so it’s getting better all the time. 

Not the bot is better, but because our documentation is better. But to answer your question more kind of holistically, I went through a bit of a identity crisis about 70 episodes, I’ll call them episodes into the a hundred days of ai. I realized I’ve built all this shit that is totally worthless. I built some really cool stuff. I created a AI avatar myself, and I cloned my voice and I did all this stuff like nano banana image generation. It’s like, what the hell? What are we doing all this stuff for? And I pivoted to being like, I’m just going to build products because I find it really interesting because exactly your point is like, yeah, a lot of these AI automations and agents just aren’t helpful right now.  

Rob Walling:
Speaking of 70 days in and having a realization as someone who for a couple years, I forget how long we did it, two and a half years maybe. I did 52 YouTube videos a year and it was brutal. It was so much time and work and the team, it wasn’t just me, it was many people working on it, and I almost tapped out at a certain point and just said, I’m done with YouTube, but we went to every other week. It’s been much simpler since then. I want to talk just a little inside baseball on this, on YouTube, and then I want to move to your, you have a big reveal, an announcement per se that we’re going to get to right at the end and then Cliff hang this, but how did you possibly crank out a hundred videos in a hundred days?  

Craig Hewitt:
It was a lot of work, I think is the easy answer. There’s no magic ai, blah, blah, blah to it. Really, there was almost no AI at all to the whole production process. I had a notion board where I just kept ideas. I borrowed, and I think this is good baseball. I borrowed a lot of ideas off of other channels in my space, so talking about AI automations, the first couple of videos are about AI automation and I kind of borrowed ideas that other people had built and I built my own. There are a couple of what I would call AI thought leader and news channels, and I would kind of follow what they do and make similar videos to, and everybody in YouTube does this. Right later I started trying to do a bit of news jacking, which is like, oh, Google nano Banana. 

This is my first big hit was I dropped the video on 12 Google AI products. You need to know. I dropped it the day that Google’s big image generator dropped just by chance. And then later I started trying to really do last second videos on to do a video on the day something was released and then later most of it was around building product with cloud code. So I think I mentioned all that because ideation is the hardest part. How do you come up with a hundred ideas? The execution of it’s not that hard, it’s just like, God, what do I want to talk about today that I haven’t already talked about? That’s interesting and that I could actually have you watch. It’s a little bit of everything, a little bit of inspiration from the space, a little bit of just what I want to do. 

A little bit of this probably would work, this a listicle or a best X or whatever. I had two times when I wasn’t sure if I would get a video out. The worst was I had done a video that day and the recording didn’t take for some reason the wrong camera was selected or something, and I went to export it at like 9 45 at night and it was wrong, and I was just like, I was sick and I was like, I just don’t want to do this. And so I got something out for the next day. I have an editor who’s part of our team at Casto Productions. He did 85 of the videos probably, and he was incredible. He was incredible. He worked on the weekends. I was never more than a couple of days ahead, and so he would work on the weekends and stuff. 

That was huge. I used a tool called Creator hooks. Creator hooks.com. It’s an AI tool where you put in, what’s the video about, and it’ll give you title and thumbnail suggestions. I had Jake Thomas on my podcast a couple of times. He’s the guy behind Creator Hooks. That was cool. I didn’t have to come up with title and thumbnail every time, and so it’d give you a bunch of different title suggestions and then just the text overlay for the image and then just, I made the image in Canva, but that was kind of the workflow.  

Rob Walling:
Yeah, that’s impressive, man. I can’t imagine trying to do that. But you grew your subscriber base up to what, 11,000 subscribers?  

Craig Hewitt:
Yeah, 11,000. Yeah, we hit 10,000 last weekend, which was incredible. You probably can relate, we had two videos, I think over 50,000, and neither of them are the video that I want to be known for.  

Rob Walling:
Yep, totally been there.  

Craig Hewitt:
They’re just two different listicle videos really. And I think you get a bunch of tam, I guess a bunch of reach on those kind of videos, but the ones I want are the video I did about making the Claude Code project for content marketing. That’s me. That’s super interesting. That’s really super aligned with this MicroConf founder or person, but just the reach of those is not there and the mass appeal.  

Rob Walling:
Yeah, no, we feel the same way. It’s always the push and pull of bigger the audience, but it isn’t actually within, it’s not even ICP, but it’s within the person that I want listening. It’s not the audience that I’m seeking. So as we move on to our final topic, I get the feeling, am I guessing right, that you got into Claude Code and the last 30 videos you started building something and then at certain point did you think, huh, I might want to build this for real. Is that what happened where you’re like, because something you said you’re going to start talking about is what the thing you’re doing at CAOs?  

Craig Hewitt:
Yeah, so just really transparently, I think give folks a little bit of context. We’re eight years in at CAOs and amazing company, amazing, stable, profitable, growing a little bit business, but the podcasting space is challenging. I think competitive. Everyone’s competitive, but I want my own airplane. I want my own jet. And that’s kind of the yardstick for everything is like, is this going to get me My private jet? Casto as itself is not a hundred million dollars company. And so I’m kind of sitting here saying like, Ooh, AI is a giant lever either using AI to do a thing or building in the AI space. Can that get me my private jet? Can it help me return a whole bunch of money to TinySeed and my other investors? And I think the answer at a high strategic level is cast dose by itself as a podcast dosing platform. 

Not a 20 million company I don’t think, but can I layer on a product or to have a suite of things that are complimentary that we can cross-sell because we have an email list of almost 40,000 people and we have 4,000 customers. What can we do to leverage the thing we already have to expand our base if we say that just making the current thing a lot better maybe isn’t the answer, which to be honest, I just don’t think the easiest way to grow the company. And so yeah, I have three different things that I’m in kind of product validation on right now and I’ll talk about, it’s funny being 10 years into this, thinking through how to validate a new product is really different than the first time, which I answered your survey the other day of how did you validate your first product? I was like, we just decided to do it. 

I, I think this will be helpful for folks. I had this conversation with Manus the other day because it was helping me think through some of this. I said, I want to create a product that is in this space, kind of like I can cross sell to my existing customers, but I want it to have these few attributes. The first is it needs to be at least $50 a month at least. We’re going to start at a hundred I think for all of these because selling a $20 a month thing is super or hard. So a hundred bucks a month, absolute entry point for anything that I do these days. And I think for most people, if you can’t come up with an idea to sell a hundred dollars a month thing, I would probably think harder. It needs to have expansion revenue built into the model. So an ESP or a CRM or a usage-based thing, a lot of AI things or usage-based. So the more you like it and you’re successful, our P goes up through expansion revenue and it needs to be aligned with our current customer base and audience at CAOs. There’s a lot of things I want to build that, but that’s hard mode. I’ve already done the hard part of creating a brand that we can just expand on from there.  

Rob Walling:
So that brought you to then a list of three products that you’re evaluating to build as a second product under the CAOs umbrella.  

Craig Hewitt:
And I can share that, man, I don’t know if I’m ready to share.  

Rob Walling:
You don’t have to share. You don’t have to share the ideas if you don’t want to. The thing I’m trying to get at is, it’s interesting to me, there would even be three ideas I would think, but is that the Manus speaking? Because I feel like I’d have a really tough time coming up with a list of viable ideas and that one of them for me would just leap out and grab me and I would just get the madness for it and be like, I want to build that right now versus having three that I was actually validating.  

Craig Hewitt:
So that’s a really good point. So I think there objectively several different good options. One of them talks to me personally more than the others, and I think that is pretty important because one thing I’ve seen with CAOs in my whole kind of cycle with it is I got to be excited about showing up every day and doing this, especially starting from scratch. So I’ll talk through the three ideas. One is I started on a channel, it’s called Link Berry, link berry.ai. It’s a AI writing tool for LinkedIn. So the idea is a lot of people that I talk to cast us are like, I know I need to be on LinkedIn where all the thought leaders are, but it sucks and I hate it and I hate writing for it, and I just don’t want to do thought leadership content. And so we’re like, cool, how can we go get the ilk of who you are from other content you have online and just create content for you? 

That’s pretty compelling. So basically saying instead of you hiring a ghost writing agency that’s like two grand a month to pay a hundred bucks for us and we create 20 days or 30 days of content for you at a time, that’s pretty compelling. I think the other is a tool that we built internally, which is this AI content writing tool. Dude, it creates such amazing content, but it’s in code and I have to freaking open up cursor and go into the terminal and do this stuff to have it right, because just like that, taking the rails off of AI that Claude code allows for, it’s like, cool, can we sify that, give a ton of context to it and have it create really good content. So like an AI SEO tool, there’s a lot of value in that. I think the market for that is much bigger. I think the challenge of that is it’s probably harder for people to see the value of that.  

Rob Walling:
There’s just too many competitors. The way I see it too many is too strong, but there are a hundred tools that are basically saying they can bring up blog content. And here’s the thing, yours is really good and it’s better than all of those. No one else will believe you when you say that. You know what I mean? It’s like that’ll be the big challenge there is how do you prove to other folks that when they’re evaluating, should I buy this one or this other one? But anyways, but you’re right, the market for that one is bigger.  

Craig Hewitt:
And the third one is kind of an offs spin of that, which is talking about agents and you’re like, I think I noted agent is, but is we are creating, and we’re using it in Casto, a few really specialized AI agents like, Hey, go get me the SEO data from our website, our Google Analytics data, go get me a Google search console information, go write this article for me in Slack. And so commanding army of marketing agents from Slack. And so the writing agent that we have is kind of a part of that, but so are like, Hey, go write me a LinkedIn article, go me LinkedIn post, go write me ad for Facebook ad. I think that’s probably like, I’m just not there yet because that’s the first one and the second one put together along with a Slack bot and a bunch of really complex stuff. So yeah, I might do that in the future, but it’s probably, I think we should go raise a bunch of money if we do that. It would be a really good business, but that’s a different kind of train.  

Rob Walling:
Yeah, that makes sense. It’s a bigger, more ambitious, more competitive, there’s more to build, right?  

Craig Hewitt:
Yep.  

Rob Walling:
Huh, interesting. How are you feeling about building a second product talked about on this show with Ruben? I get this question every once in a while of should I build a second product? And it’s like by default, no, but the not build a second product thing, it’s usually somebody who’s got 20 grand a month in revenue and it’s like, that’s not your, don’t build a second product. This is not happening. You’re obviously in a very different situation than that being eight years in. And as you said, you kind of hit the S-curve a little bit. You’ve heard about the S-curve that dmh, I mean, it’s before DMEs, but basically every product plateaus at a certain point, and sometimes that’s half a million and sometimes that’s 30 million. But it happens eventually. And then you do, you look at HubSpot and Salesforce and all these big companies, they’ve had to layer on multiple products in order to keep growing. 

You just can’t. The total reachable market for these things isn’t that big. But beyond that, not about do you think you should be building this product because you think you should, right? You’re like, no, this is the way to grow it. You’ve already talked about that, but as a founder, there’s got to be some excitement in it, but there’s got to be, is there a little bit of kind of dread or anxiety around it’s the pro and the plus the thing that makes you happy and the thing that makes you squirrelly? What are those emotions?  

Craig Hewitt:
Yeah, I think that you shouldn’t do this. We have tried to grow organically and I mean organically within the kind confines of what Castus is to achieve significant growth for a long time and have brought in some extremely talented, smart people to help us and it’s not happened. And so the one thing I would say to everybody else is don’t think go building another tool just because tried to grow for a month is the answer. We’ve been trying for a year and a half and have brought in the most talented marketing people that I know and paid a lot of money and it’s just not happening. I think we’re at that natural point in our business and yeah, I’ve talked to Reuben about this and he’s like, Hey man, it’s just not going to happen. 

And so we have a responsibility to ourselves and our investors to make a ton of money. And so if that’s not the way, then this is the way I am very excited about building a new product. I think Link Barry is a really compelling value prop when you say, we’ll create a month of LinkedIn content for you so you don’t have to, and it’s a hundred bucks, people are like, oh yeah, amazing. Take my money. I think in the AI era, you kind of almost need to work backwards from the aha moment to the product, and that’s, it is like, what’s the problem we’re solving? Which is like people need to post on LinkedIn, but it kind of sucks and they don’t want to, we’ll just kind of do it for ’em. We’ll just pull a bunch of information from them from the internet, aggregate it and kind of spin it into content and it does a really good job. So I’m excited about it from a product perspective. I’m starting to bring the team into it a little bit, which that’s a whole dynamic. I’m nervous that we’re starting from scratch and it’s been eight years since we’ve had no customers, and I’m scared that it’s not going to work. And I get this in talking to founders who have exited a lot, they’re like, was it luck the first time? Can I do this again? And so I’m having that from within the company still.  

Rob Walling:
It’s got to be exciting. I love the fact that you, because and I have been, obviously we talk pretty frequently and I have watched you try all the avenues to get CAOs growing faster, and it is frustrating when none of those work, and I feel like in a business that’s not growing for me, I’m going to speak for myself. I’ve got bored gotten demotivated. It’s like, I’m not going to keep spending my time here doing this if it isn’t working. And so it’s like what’s the excitement? What do founders actually get their juice from? I know that we talk about, well, it’s to change your life. Money is a goal I think of almost all of us, and that’s great. I have no problem with that. But also it’s the excitement, it’s the learning. It’s the seeing something go up into the right, it’s seeing more customers or more MRR. There’s always got to be something, right? And I think the thing that maybe has been lacking for you for a bit of just to keep you motivation, right?  

Craig Hewitt:
Yeah. I mean motivation, a little bit of fear and excitement, I guess. Casto is a stable business, for better or worse. It’s really stable. Yeah, I need to sizzle it up a little.  

Rob Walling:
Well, I’m excited for you, man, and folks who have known you for a long time is running cast hosts, the podcast host. We all got to kind of expand our thinking and that’s it. Yeah. And going to get excited to see as you launch this, folks want to keep up with you on X Twitter, you are v Craig Hewitt and on YouTube, youtube.com/the Craig Hewitt. And if folks were interested in, they’re like, I want to see what you build email. Do you have an email newsletter where you’re going to be like, Hey, this is what I’ve settled on? Yeah. Is that@craighewitt.com?  

Craig Hewitt:
Yeah, Craig hewitt.com. There’s a little button in the top right to join the newsletter. Yeah, I send the newsletter out as often as I can, which is about every other week. And then if folks want to check out Link, Barry link, barry.ai, we have wait lists there now of a few hundred folks, and we’ll be opening doors for early access in a few weeks. So a week or two after this comes out,  

Rob Walling:
Going to link barry.ai right now to enter my email address. And yeah, it’s been great having you on the show, man. Thanks for coming back.  

Craig Hewitt:
Awesome. Thanks Rob.  

Rob Walling:
Thanks so much to Craig for coming back on the show, and make sure you follow him either on Twitter or YouTube or sign up for his email list to see what he’s up to, because he’s putting out a lot of really interesting content. And I, for one, am looking forward to where he’s headed as he starts building this second product. Thanks for listening this week and every week. This is Rob Walling signing off from episode 809.