Filtering out bad ideas
#27

Filtering out bad ideas

Justin:

Yeah. If you missed last week, I had to leave. And I'm so sorry, Brian, and to everyone else. Like, I here's what happened.

Brian:

I really felt like like I texted you right after. Like, dude, are you okay?

Justin:

Here's what happened. So like many people, I have a do not disturb mode when we're recording, and it's very hard to break through it. Like, can't You can't break through it unless you call me repeatedly and you're in my

Brian:

favorites That's the bat signal.

Justin:

Yeah. And so I got, like, at 01:59, I got a phone call while you were demoing your thing. It was my wife. And then I'm like, look at the time. And I'm like, oh my because it the episode just went longer because and we had a sauna appointment booked.

Justin:

And so I was just like, I knew I could get there on time, but I knew I had no time to diddle. So I just like, I just left it live. Left you hanging.

Brian:

Mean, you made the right choice. I mean, come on.

Justin:

Oh my gosh. That's gonna be a recurring

Brian:

sauna emergency.

Justin:

So the real story on oh, I mean, there's there's so much inside jokes about this now in in chat. Oh my gosh. So good to see everybody. We're here.

Brian:

Yeah. Justin

Justin:

lives. Sauna priorities. That's the that's somebody people are, like, watching it. Here's the thing that scared me is we decided to do that experiment where I left that Riverside link in. So that Riverside link, because I just left our recording going, anybody, any, you know, terrorist could have clicked that link, taken over my livestream, and Yeah.

Justin:

Ruined my life. But thankfully, that didn't happen.

Brian:

It would've been funny if, like, you left that up all week long and then, like, your because I I think we we put it on your YouTube channel. Right? So that Yeah. Then, like, your YouTube channel, like, blows up because of this. Oh my god.

Justin:

That would

Brian:

be amazing.

Justin:

They're just I I mean, there were some funny jokes. People thought, you know, that it was like the, you know, the calming fireplace video or whatever. Someone said they they thought they saw a mouse run by in the background. It was just all sorts of all sorts of stuff. So I I mean, maybe that'll be a recurring bet.

Justin:

I will just leave without notice. And

Brian:

Yeah. You know, you never know when we're just gonna cut it off.

Justin:

Okay. But we were right in the middle of you demoing a secret app. Do you should we can we bring it back to

Brian:

Sure. I mean, I we don't have to go through the whole demo of it. I could talk about it. Yeah.

Justin:

So tell us about it.

Brian:

Okay. So

Justin:

It is not a pack of new filtered cigarettes.

Brian:

No. I decided not to do the the the drop shipping cigarette business. Although that that seems like a maybe a better business than what Filterhawk actually is.

Justin:

That would be that would be hilarious if it was just you just put up a store on Shopify and it's just

Brian:

Well, I will say I have so many different things to to talk about related to FilterHawk.

Justin:

Okay.

Brian:

Overall, I'm really happy that I did the project. I I I see it as a success on a number of different metrics.

Justin:

Okay.

Brian:

But the metric that it is not a success is user growth and customers for to pay. I mean, you just

Justin:

list it. You you sent one email.

Brian:

Yeah. But I I sort of expect it's gonna be I I've talked about this before. Whenever you launch a new product, a new thing, like typically I always get some customers. Sometimes it turns into a real business. More often than not, it sort of gets to that little that like hovering state where I got a few customers, little bit of revenue, chugging along, but it's not a not a real anything.

Brian:

And I and and I often kinda wish that the thing would just fail and I would just get a zero.

Justin:

Yes.

Brian:

I I got a zero on

Justin:

this one. And

Brian:

I'm actually kind of kind of psyched about that in in a way. Like like it's Yeah.

Justin:

Because the the the danger is if you got if you got five paying customers or 10 paying customers

Brian:

Now I gotta support them.

Justin:

Then it was flat forever. Then yeah. But if but

Brian:

I always I love the idea of just building tools and selling my byproducts.

Justin:

Mhmm.

Brian:

Ultimately, I I built the thing because it is a tool that I am using on my Gmail inbox. It's active right now, and it's actually working really well. I'm I'm super happy with it as a little self made product. I'm also really happy that I spent only a week on it. The the big win for me is for Builder Methods.

Brian:

Like, this is helping me refine my product building and shipping process. I can get into that later. But the

Justin:

This is the advantage for you. This is the epitome of selling your byproduct. It's like, well, I'm gonna make it anyway for builder methods. Might as well put a sales page up. And and and by the way, there's some other things about Filterhawk that I think are interesting.

Justin:

Like, the site looks amazing. You did it very, very quickly. Branding looks great. Like, there's lots to glean from this. Like, this this site looks great.

Brian:

I I wanna talk about the process of of building it. But before like, just without, like, bearing the lead, for those of you who've never heard of this FilterHawk thing before, it is my goal with it is it's it's a replacement for your Gmail filters. It's not a Gmail email client. Like, you will continue to use Gmail or whatever email client you use with Gmail. The the goal of FilterHawk is you no longer have to use Gmail's built in filters feature, which I think is terrible and chaotic and doesn't work most of the time, especially if you wanna do anything a little bit more complex with your filters and labeling So and I've been so frustrated with that.

Brian:

You know, we all get so much noise in our inbox. So I started to play around with Claw to see like what's actually possible with Gmail's API. And it turns out you could you could actually do a lot with Gmail's API. And I essentially rebuilt a a so much better version of of Gmail's filters in this little FilterHawk app. And I sort of on on the marketing side, I sort of go through, like, there are four or five, like, really bad pain points with with Gmail filters.

Justin:

Mhmm.

Brian:

Like, you know, because like I like, in my inbox, I do a bunch of things, and it and it's super noisy just like most people's. Like, I get a lot obviously, a lot of cold email. And then I have a lot of, like, customer like, lot of, like, Stripe notifications and money transfer notifications and auto auto paid receipts. I get a lot of that kind of noise. Mhmm.

Brian:

But then I get a lot of replies from my audience Yeah. From my newsletter. I get a lot of survey responses. I mean I mean, like, hundreds, you know, every couple of weeks. It it starts to pile up.

Brian:

I I have to have systems to, like, put all those things into buckets, into, like, separate inboxes so that I can process them separate from my primary feed of like actual business emails that I where where I'm talking to people. Mhmm. So filtering and labeling is like super important to me. And so I built this thing. And it and it works actually so much better than like Gmail's actual filters.

Justin:

Yeah. I was gonna say the the UI you've created here looks so much nicer. Gmail's settings are weird. It's like once you go into advanced settings, it it looks like they haven't updated updated it since they

Brian:

launched Gmail. Least. Yeah. I mean and like, there's so many chaotic things about it. Like, first of all, so I have, like, one filter.

Brian:

A lot of these are sort of copied from or or inspired from Hey, you know, Basecamp's Hey. Yes. Like, I some of their features. I just don't wanna use Hey in general.

Justin:

Mhmm.

Brian:

So, like, one is, like, I have my version of of the paper trail. Right? Which is mostly like receipts or like money transfers, things that are like sort of on auto autopilot, but I still wanna have a record of them in my inbox. I don't want them to be like new emails that I have to click every time.

Justin:

Mhmm.

Brian:

So I have like a complex filter that checks for things like, you know, Stripe payout notifications or Stripe payments or transfers to my bank or, you know, bill was auto paid receipt. But I also have to that filter customized so that I don't miss emails from real customers who might be asking a question about their receipt that they just received from me.

Justin:

Yeah. I've had this exact situation.

Brian:

So so I have to be really carefully dial in these filters. So and like with Gmail filters, you could do these, like, complex queries, but those really never work, honestly.

Justin:

Mhmm.

Brian:

And then and then like the other thing is like you never really know like if I have like 10 or 15 different filters piled up in Gmail, you never know which order they're gonna run-in. So in Filterhawk, you can literally drag and drop and make a priority order. Like when a new when a new email comes in, process it in this order. If it if it passes these filters, do this. If it doesn't, then move on to this filter.

Brian:

You know?

Justin:

Yeah. Yeah. Yeah. I've I've had that ex yeah. This isn't exactly the situation I've run into with Gmail's filters.

Brian:

Yeah. Yeah. So anyway, like, it it works really well, like, on my inbox. I I could and I was actually I almost decided not even to even offer it as a product for people to buy. Mhmm.

Brian:

Because my hesitation was, and this is probably why most people are not are not signing up for it, at least just based on my tweet launch of it. But, like, I'm not surprised. Like, people would a, maybe they just don't feel the same pain with Gmail filters and their inbox as I do.

Justin:

They're sort

Brian:

of happy with whatever systems they have, which is I get that. But also, would understand, like, people aren't ready to just connect their Gmail to some new bootstrapped startup tool. There's a lot of privacy and trust issues there. Yeah. I did build in some encryption kind of stuff and also the ability to automatically purge your emails from our system even after a couple of days, after the filters have run.

Brian:

But anyway, like that so I'm not really surprised that it's not gonna be like a flyaway success as a product. But Yeah. But I'm still happy that I I'm really to me, it was also like a breakthrough in my process for building in on so many different levels. I mean, we can get into that, but I I feel like I had a a major breakthrough with this.

Justin:

I mean, I do think you should AB test this with this branding instead and just see how it works. Because I think that might you know, you know, the

Brian:

You know, those cigarette companies, they they did figure something out.

Justin:

Yeah. I mean, I think this is there's certain characteristics that are kinda like the death knell for bootstrapped software products. And one of them is, like, if you have to get an engineering team to install a code snippet or configure something in their app to get it to work, that that friction and how it just stops people from signing up and activating, that is hard. And then there's also like, privacy issues. There's also giving it access to my Gmail.

Justin:

I mean, even for me, like, superhuman, I was like, I don't know. I don't know

Brian:

about this.

Justin:

I don't

Brian:

know about that. Yeah. And I did it with this other company, Shortwave, for a while. But but part of it was because it was like, I don't know them and they don't know me. So that makes me feel a little bit better about it somehow.

Brian:

Mhmm. You know? And I to be to be clear, like, I didn't really market this in any real way. I I just did a a tweet and a blue sky about it. So

Justin:

Yeah. In some ways, that's better, though. If if if we're talking about this kind of fail fast approach, you know, I've like over my life, I've tried dozens and dozens dozens of products. And my favorite thing to do is to mock it up as a prototype, create a video of it, start sending to people, and just see what their response is. And as soon as I get back, well, this is pretty good, but I would really need it to do this.

Justin:

Or this is pretty good, but I would need to like this isn't the really the way we do things. Or I'm like, okay. Like, if I get too many of those

Brian:

Yep.

Justin:

Yeah, there's some wisdom in not persisting.

Brian:

Yeah. And but, you know, also now we are in in an era where, like, instead of just having a fake mock up of the thing, I could just build it in a week.

Justin:

Yes. Now that's what I wanna hear about. How because, I mean, people are people are saying a bunch of things here. People are saying they like, you're the design of this marketing site is great. Branding is great.

Justin:

And then yeah. Like, how did you actually build I wanna this.

Brian:

Yeah. Okay. So, like, first of all, I set out like, I had actually a bunch of other projects going on over the past week.

Justin:

Mhmm.

Brian:

So, actually, in reality, if I were to only work on FilterHawk and cut out all my other things going on, I think the whole project start to finish would have been like four days maybe tops. Something like that. You know, there was a weekend in there. Plus there was like some other projects going on. So it it sorta stretched to like seven to ten days.

Brian:

But like, I actually worked on this project like four days essentially. And so okay. Like, when I set out to do it Yeah.

Justin:

So yeah. This is what I I wanna know. Is this is a new project. You're the agent OS guy. Do you just have a, like, default spec plan that you just drop into every new Rails project that just, like, these are the best practices for a Brian Castle app?

Brian:

Well, the the thing I'm really psyched about this project is that I redefined a lot of those best practices for myself.

Justin:

Okay.

Brian:

I took I took a totally new approach on so many different things on this project than I've done in the past. First of all, I actually didn't end up using Agent OS at all on

Justin:

this. Okay.

Brian:

But I did do a version of spec driven development. I still think that's a core thing.

Justin:

Okay.

Brian:

Well, all right. My goal going in was like, let me see if let me see how fast I could possibly build this tool that I want to have for myself. I actually set myself like a deadline of like about a week. Okay. And I knew some of these things that I'm about to talk about.

Brian:

I knew I was going to experiment with these things. And my hypothesis was like, maybe if I do these experiments, I actually can move like 20 times faster than I've moved in the past on everything else that I've built. Even with AI in the past, I still felt like I was very slow. Okay. So like number one was, you know, I've always been like a pure Ruby on Rails stack guy.

Justin:

Mhmm.

Brian:

Meaning, like, Rails, Stimulus JS, Talend, you know, Rails, like import maps. I I always stick like, stay away from React and and Node, know, and, like, didn't wanna touch that stuff. I just pure Rails. Right? Instrumental components that I made was, you know, all pure Rails, Turbo, Rails, Stimulus, all that.

Brian:

This time, I ended up going with Rails, Inertia, and React. Oh, wow. And and SHAD CN UI components.

Justin:

Okay.

Brian:

So totally like my first I've I've dabbled with that inertia React thing just to sort of learn it a little bit. But I honestly, I don't I'm I'm no expert with that stack at all.

Justin:

Yes. Why why the decision to go with React? Just is it just because you assumed AI is so good at React that it would be worth experimenting with? Or

Brian:

Yep. That was part of it. Also, I don't know if this is actually a requirement with Shad CN or not. But I like that stat that combination of things seems to be it's it's turn it's it's a super norm. Like, even like like if you use like Claude just within Claude the web app, not Claude code, like, it's gonna use React and SAT CN components and stuff, you know?

Brian:

Yeah. That was one thing that was totally new. And then another thing was this wasn't specifically a goal of mine, but it turned out to be true for all of Filterhawk, the application and the marketing site. The marketing site is a one page HTML, by the way. That's nothing more than that.

Justin:

Okay.

Brian:

Marketing application and marketing site, I did not hand code a single line of code myself. Not one. Everything was through AI.

Justin:

100%.

Brian:

100%.

Justin:

AI generated.

Brian:

Like generated? That's a weird word. Right? Like, I designed and directed, I would say, and architected everything.

Justin:

For the marketing site, did you give it a mock up or a wireframe?

Brian:

No. Okay. So like like yeah. I guess you could say it's like generated.

Justin:

Okay.

Brian:

But I can get more into like the design phase too. Like, that's a new approach as well. Wow.

Justin:

Did did you record a video on this, by the way? People are already asking about that in the chat.

Brian:

I I didn't. So like one of the things was like, let me see how fast I can do this experiment and as a learning project. And then, like, on the next app that I build, I'm gonna I'm gonna take it slower and record myself every step of the way. You know?

Justin:

By the way, take my money on that. Like, I I don't think I'm the only one here listening to you going, I would pay to see that. Like, I want to see Brian build it. Like, everything you're describing is just like, wow. Wow.

Justin:

Wow. And it has that unexpected feel too. Like that you didn't hand code anything. That's like, okay.

Brian:

I didn't even really set out to do that. Like, as I got into it, I was like, this is how I'm building now. And it's like that's Okay. Maybe later on today, we can talk about like I I I am moving towards that, like doing some sort of course in inside Builder Methods to talk through this process and everything. Well, another thing alright.

Brian:

So when it comes to design, another new approach that I took was I started in claud.ai, like the web app. Talking with Claude about the specific requirements, the problems that this product should solve. We have a conversation about that. And then I asked it to generate a JSON sample dataset.

Justin:

Okay.

Brian:

For for let's say like the workflows page where you're designing your filter workflows for FilterHawk. We sort of hammered out like what what the capabilities and the feature set need to be on this through like a discussion with Claude. And then I said, okay, based on that, now generate for me a JSON sample data set for what the workflow page would require. And just in terms of like the raw data. Okay.

Brian:

And it generates that. So the like, we don't have any sort of, like, real database yet. This is precoding. Like, this is, like, in the planning phase.

Justin:

Okay. Sorry. Is this in Claude code or cursor?

Brian:

This is actually happening in Claude, the web app. Claude.ai. Like,

Justin:

Chat claude. With Claude. With Claude. Yeah. In the web.

Brian:

Yeah. This is like before I even opened up Cursor.

Justin:

Okay. Okay.

Brian:

So I did like a day of this where it was like, okay, we're gonna we're gonna work on like the main workflows page first. And talk through the requirements, get a JSON sample data of that. And then using the sample data, I then ask Claude, okay, now design me a mock up of the user interface for the workflows page using the sample data that you just created.

Justin:

Oh, interesting. The

Brian:

key insight here and I heard this from from a different podcast, but like the the key insight what like what I used to do was ask Claude to do the same thing, like generate a user interface for this sort of app. It needs to do this, this, and this. But without that sample data, it sort of adds in all these other little bells and whistles, or it's missing some bells and whistles that I need. It doesn't quite get the details right. So by going step one, discuss and sort of shape up what's what it's gonna be.

Brian:

Step two, generate just a sample JSON data set. And then step three, based on that, then then you end up with a UI that's like exactly what you need to The exact buttons and functionality are all there.

Justin:

Wow.

Brian:

Right? That was a key insight. Yeah. And then even that is not even what ends up being the app. All that does is it's it's essentially like a mock up.

Brian:

Like, it does code it up like in an artifact in Claude. But I just take a screenshot of that and then I'm bringing that into cursor. So that so then so so that's like phase one. I do that for like maybe three or four of the main sections of of FilterHawk. And then I go into cur For this, I did use cursor two point o.

Brian:

K. Here and there, I I opened up Claude code, but for the most part, I was using cursor to build.

Justin:

Cursor with which model?

Brian:

I flipped back and forth between Sonnet 4.5 and Composer one.

Justin:

Okay.

Brian:

And then also for some of these features, a lot of them, I I I had those two models compete against each other. The same same prompt, build the same thing at the same time, and then I choose the winner.

Justin:

Oh, I didn't know you could, like, have those things Yep. In parallel.

Brian:

That's a new thing in Cursor two point o.

Justin:

You can Interesting.

Brian:

You can sort of spin up multiple competing.

Justin:

And now that everyone's also saying this Gemini three model is very good.

Brian:

Yeah. Yeah. I I haven't had time. I haven't had a chance to check that out yet, but I definitely plan to. And also their new IDE looks really cool too.

Brian:

So that's essentially what I did. So then in Cursor, instead of using Agent OS, I I just did a it's a very similar process. I used spec driven development. And by that, I mean, I used plan mode first Yep. For every big feature.

Brian:

I broke the roadmap down first. Here's the roadmap. It's like five or six phases. We're going to have a dashboard. We're going to have workflows.

Brian:

We're going to have email audits, integrate with the Gmail API. All that is in a road map. And then for each one of those big features, I go into plan mode. And in in plan mode, I give it the screenshot of that mock up that I did earlier. I give it the JSON sample data that I already created earlier.

Brian:

I give it like markdown file that was generated from my ShapeUp discussion with Claude. Like, this is the next thing I wanna build. Now make a plan. So both Cursor and Claude code do this. It's the same feature.

Brian:

And Agent OS does it too. But now now these tools have it built in.

Justin:

Yeah. Right? By the way, are when you say each feature or you starting a new chat for each of these things? Like, it's like new chat?

Brian:

Yeah.

Justin:

Or are you just in one chat doing it? Or like

Brian:

New chat, I would say. Yeah. Okay. New yeah. Yeah.

Brian:

Like one big feature is probably like one one big chat, I guess. Let's see. So so then you then you create a plan. And the other thing that is great, I think, about cursor and Cloud Code, and Agent OS has this too, but but now, honestly, I just interchange like, it's they're doing the same thing. This is spec driven development, which is it's not only just making before it even makes the plan, it asks you clarifying questions.

Brian:

This is so key. Right? Here's what I wanna do. I give it sort of like my big spec, my screenshot, my JSON. Now ask me back clarifying questions.

Brian:

Sometimes I tell it that, but actually these tools will do it automatically. Nice. It's gonna send me four, five, six clarifying questions to really dial in the scope. Do you want it to work this way or that way? Or should we assume this or we're not gonna handle this or that edge case?

Brian:

Right? A lot of those questions, we dial it in and then it generates the plan. It it also generates a task list for itself. And now it has a a that's a spec. Right?

Brian:

Like, that's that's ready to go. And then I have it build. And if it's a if it's a bigger feature, then I will probably use two or three different agents building the same plan at the same time. And then I can compare which one does it best. Now, when you do that, it's gonna cost you like two or three times more.

Justin:

The tokens. Yeah. Like, I'm still on the $20 plan, I think. I haven't done as much, but I'm assuming I

Brian:

actually am too on on Cursor. But this month, like, my overage is like over $150, you

Justin:

know? Okay.

Brian:

Because I I used it a lot. So

Justin:

It is interesting. Like, I can see situations where I'd be like, you know what? For this, I wanna, you know, have it build a few things in parallel and just compare. It's like, okay. Let me see what you did here.

Justin:

Let me see what you did here. I I might like one better than the other. That's really interesting.

Brian:

Yeah. I think it's worth doing that for, like, really big and maybe really critical features.

Justin:

Yeah.

Brian:

But then So then I get a big feature built. And I mean, literally it takes like five minutes, maybe ten minutes tops.

Justin:

Right? And

Brian:

the key insight with all of this is like spend hours, maybe even days in the planning, and then have the agent build it in like ten minutes. Like that's the workflow.

Justin:

Are you how often are you reviewing the code?

Brian:

Okay. So then we get into this next phase, which I think of as like refinement. Okay. I do a little bit of code review, but a lot of it is like in a lot of that is even in the upfront planning. Like, I want it architected this way.

Brian:

The database modeling should work this way. Let's name this that. Let's make sure we have a component for this. I'll include that kind of stuff in the planning. But then when it's done, the design is pretty good.

Brian:

Again, we're using SCADCN components, so there's a lot of quality sort of baked in from there.

Justin:

Was there a reason you chose SCAD over Tailwind components or your own components? Like, why did you make that decision?

Brian:

Well, it is using Tailwind, but not Tailwind components. I think because of, like, the popularity of SADCN and, like, the way that it works so well with, like, React and Inertia. I I don't know. Like, it just seems like that's, like, a popular combo.

Justin:

Yeah. And you wanted the experiment. And part of the experiment is let's choose some frameworks that AI should be really good at because there's just lots of source material. So there's been hundreds of thousands of layouts built in Chad, probably.

Brian:

Totally. And and I know that they've gone through a few versions, and they're really good now. And like, but they're also very, like, plain vanilla, like customizable. Right? Like like, I still have my touch on like, alright, I want the colors to be this.

Brian:

I want the fonts to be this. We we need to smooth out this UI a little bit. Like, you could still customize them, you

Justin:

know? Yeah.

Brian:

So then there's this refinement phrase, right? So it built a big feature. I would say it's like 90% I'm happy with it. And then there's five or 10% that we need to refine. We need to fix.

Brian:

We need to clean it up, polish it up. And so that's where I really love the idea of using Composer in Cursor.

Justin:

Okay.

Brian:

The Composer model is so fast

Justin:

Okay.

Brian:

That it's like it's it's like I just have to have a thought and say it into my microphone. And thirty seconds later, it appears on the screen. It's just so freaking fast.

Justin:

Are you using this microphone mode in cursor too? The voice input?

Brian:

I just use Whisper like, Super Whisper on the you know, that app.

Justin:

Oh, I haven't I haven't tried Super Whisper yet. But there is there is a voice input. I've gone here.

Brian:

But They they added that in, but but I use Super Whisper anyway. So Okay.

Justin:

I gotta check out the

Brian:

Super Whisper. Yeah. The other popular one is WhisperFlow. I've been sort of bouncing between those two.

Justin:

Okay. Man, I'm like a Luddite. I'm like an old guy.

Brian:

I Oh, dude. I I'm like it's so weird. Like, I'm speaking most things into into the computer these days. You know? Wow.

Justin:

That's like, when we talk about living in the future, like, stuff like this and the the fact that this is widespread because I've heard people talking about this. Like, oh, yeah. It's just Whisper. I just use Whisper or whatever. Mhmm.

Justin:

It's like, wow. Like, I haven't even tried this yet.

Brian:

I yeah, dude. I I've been on this for a couple of months, and it has changed the way I interact with the computer. 100%.

Justin:

Interesting. And one of my kids is dyslexic and already uses a lot of voice to text, but this just feels like the next level. Like, you can

Brian:

There's there's little little things that I find kind of annoying. Like, if I need to, like, say, like, the name of a coding variable, it doesn't quite get that right. But, like, yeah. For for like, if I have to write, like, more than two or three sentences, I'd much rather just say it than type it.

Justin:

I wonder if Cursor's voice input is better attuned to, like, variables and

Brian:

things Okay. Like I haven't really tried it because I've I've just had this installed on my Mac. So Yeah. But, yeah, like just the speed of using Composer. I I've heard the new Gemini three is also very fast.

Brian:

I haven't had a chance to to try it yet. But that this is where like speed really matters in a coding model because it's like especially for these quick fixes. Like, oh, let's clean up this UI or change all the instances of this on the page. Yeah. It's just so fast at doing those things, you know?

Justin:

I can't wait to try that because this is how this has been my normal kind of collaborative workflow is if I'm pairing with somebody, it's often like, okay, let's bring it up. Okay. Well, why don't we try changing this? Let's try modifying this. Let's get in to DevTools and change some things there and see what it looks like.

Justin:

But this feels like an even better version of that. Cause it's probably way faster and just, yeah, perfectly attuned in all the right ways.

Brian:

Yeah. Yeah. I mean Yeah. So like that's essentially how how I did it. And then like, you know, the more manual parts were like deploying it.

Brian:

I I used my my typical workflow, which is I I I like to use a hatchbox.io with DigitalOcean. So I I, you know, I set that part manually and deploy it. And then the marketing site is a one page HTML. So I had I first had Claude help write the copy. And I and I went through multiple iterations on that.

Brian:

And and, you know, I would write my draft like, here here's the problems it solves. And I hear some raw ideas for copy ideas. But now give me some pretty good copy on that. It it gave me back. And then now based on that copy you wrote, let's let's design a marketing site for this.

Brian:

Oh, Claude Code just recently released like, the team at Claude Code just recently released a front end design skill.

Justin:

Okay.

Brian:

So they have, like, a skills feature in Claude Code, and then they, like, prepared this, like, special, like, front end design skill, which essentially gives

Justin:

skills in Claude? What it

Brian:

They're they they sort of have, like, a blog post about this. We can get it linked up. But it's essentially, it gives Claude code, like, sorta some extra muscle when it comes to designing a really nice both UI or a marketing page. And so I I installed this this skill that they were promoting on their blog, and it did actually improve. It like, it it it gave me it gave me a less, like, generic looking thing.

Brian:

Okay. And and I I did go through a few iterations on the marketing site with with Claude. And then and then I did the the video demo things that are on there. I used Screen Studio for that. Yeah.

Brian:

I mean, that's about it.

Justin:

Wow. Is Hatchbox like a

Brian:

Oh, and then and then I hosted it on Cloudflare, and then the whole thing went down. And then

Justin:

But what better time for it to go down than for with a new app? Is Hatchbox, like, similar to Laravel Forge? Is that what this is? It's like a way of Yeah. Think it is.

Justin:

Provisioning.

Brian:

Forge. Yeah. Forge, like, does that too. Right? So yeah.

Brian:

So this is by Chris Oliver. He's the same guy who does Go Rails. Okay. He has this hatch box, which is like a little sass that that yeah. This is essentially the Laravel Forge for for the Rails space.

Justin:

Nice. Yeah. I I'm a big fan of Laravel Forge.

Brian:

I love it. I think at this point, like, Forge I I actually do also use Forge for the for the Clarity Flow marketing site. And Forge is so much nicer in terms of a user interface now.

Justin:

Yeah. They just updated the UI.

Brian:

Hatchbox could could use some love on that front, but it but it still works, you know, really well and really reliably. He's got a he's got a nice little tool there.

Justin:

So Sweet. Wow. Well, that's great, dude. I mean, I think, again, judging from how people are reacting in chat, I think a lot of people would love to see that whole flow as a video. And just even having it broken down, here's how there there are so many times when you're describing it where I was like, oh, I wish I could just see Brian using cursor because actually getting that's where things kinda get unlocked with video.

Justin:

It's like, oh, that's how and you you might not even think to describe it that way. You know?

Brian:

Well, I mean, this is what I've been noodling. I don't wanna like, talk the whole episode here. We can jump over to something you have and and come back to me. But but that is, like, the next thing that I'm noodling on right now for builder methods is sort of, like, putting together this flagship I'm sort of maybe I'll call it, like, the builder method or something like that. Like a Okay.

Brian:

Like, the process. Yeah. Breaking it down into a framework. And and, anyway, like, I I have more to say about that. But why don't we why don't we stop on me for a minute?

Justin:

Sure. I mean, I don't have a lot right now. The thing I was working on today and that we've been working on kind of for a little while was doing a Black Friday sale for Transistor. Mhmm. We haven't done a lot with Black Friday.

Justin:

And

Brian:

Do you usually do a Black Friday sale?

Justin:

We usually do something. But in terms of, like, making it a concerted effort and really, like, sending it out an email and giving it this case, we gave it to all existing customers as well. And the nice thing about

Brian:

Oh, you do?

Justin:

Yeah. Well, the nice thing about being around for a while is you can go back into your history and go, what have we done in the past? And, you know, we've done all sorts of promotions that we track either in Rewardful Mhmm. Just using referral links with a 0% campaign. So even for non affiliates, I have this one affiliate, which is basically me, that's 0%, but I create all these referral links there, and then we can track campaigns.

Brian:

I've seen you talk about that. That that is a really and that's such a great simple tool for it's made for affiliates, but it's so great for that.

Justin:

Yeah. It's just helpful because I mean, sure, things can get blocked and whatever, but it's it's directionally kind of the best. It's like, oh, we sponsored a newsletter. How many clicks did we get from the newsletter? And then how many later actually started a trial?

Justin:

So looking back and then we have coupon codes in Stripe.

Brian:

Wait. So just to be clear, I don't know. Are are you announcing what the Black Friday sale is?

Justin:

Yeah.

Brian:

So like, what's the discount or what

Justin:

It's 20% off all of our yearly plans. So our annual plans are already discounted. You get two months For

Brian:

life or for

Justin:

Just one time. Just for one year.

Brian:

So but does when you say you're you offer to existing customers, does that mean like everyone is just automatically getting 20% off for their next year?

Justin:

So the policy is any existing customer on monthly can switch to annual an annual plan and get 20% off. Yeah. Cool. The the the the thing that made me think this would be worth trying again, we'll see, is, you know, we have tons of coupon codes in Stripe. And when you look at how many times they've been redeemed, like, it's for all sorts of things.

Justin:

We sponsored a conference. We we sponsored a newsletter. We did some ads. We you know, whatever. And you look at the actual redemptions.

Justin:

So there's two ways for us to track did a promotion or a campaign work. One, in rewardful, and two, with coupon codes redeemed. And there's just not very many things that we've done, both paid and non paid, sponsoring creators, sponsoring videos, doing all sorts of stuff. Even some massive audiences. Not much actually results in paid customers.

Justin:

What has always worked for us is the content that I'm producing, the SEO stuff that I'm producing, the brand building that I'm doing. It's like it. Our funnel is affiliates did help early on. We had a few key affiliates that I think helped level us up. But generally, it is not paid acquisition.

Justin:

And generally, it's what we're doing. It's our engine.

Brian:

If I had to guess, like, the hypothesis on that

Justin:

Mhmm.

Brian:

Is it's just like podcast hosting is not an impulse buy. Yeah. You buy it when you're ready.

Justin:

Yeah. You buy it when you're ready. That's right. Yeah.

Brian:

So if you know if if if you've been following Justin or if you've been following the content or if you're googling, like, with SEO Yeah. When when you're ready, when you have the need, you go to Transistor. It's that that's why it's like it's not like Black Friday is gonna prompt all these people to start a podcast this month.

Justin:

No. No. That's right. And here, I wonder if I can show you this without giving out too much. So here here's a little screenshot of my reward full dashboard.

Justin:

I won't show this forever, but I'll just put it up on screen here. So, you know, there's this is for a given time period. But you can just see, like, we've got we've got you know, there's different things that have resulted in a bunch of visits. And then there's this is the actual number of customers we got from each of those things. And so, yeah, you can go down and you're just like, oh, wow.

Justin:

We've sponsored that person and we we didn't get a lot out of it. At least not until I also this wonder if it's

Brian:

like you just don't see it in those numbers. Right? Like you you you sponsor this conference or you sponsor this podcast or whatever.

Justin:

Yes.

Brian:

You don't see those referrals, but people heard about you then. And then when they start a podcast five months later, they come around.

Justin:

I mean, and I see this all the time in the responses to these why'd you sign up emails. Like people that have been I had one yesterday, someone who bought my course, Tiny Marketing Wins, like, whatever, fifteen years ago, sixteen years ago, and has just always been kind of following me. And his boss was like, hey, we need podcasts. And he's like, oh, I know who to recommend. And then we got the sign up from that.

Justin:

Mhmm. So the interesting thing is to look at the coupons. Now, I'm not going to share this. But let me just go to my coupons list here. And redemptions.

Justin:

So I said, what is the of all of the things we've ever done, how many coupons do we have? We've got there's no count, but it's probably 50 or 60 things we've tried, maybe more. You know? Here here's Black Friday twenty twenty two. We only got six redemptions.

Justin:

Let's find something else here. I mean, tons and tons of sponsoring things. Like, people say, hey, like, sponsor our thing. And I give them a coupon. Zero redemptions, zero redemptions.

Justin:

Like lots lots of that. But the biggest the most redemptions we ever had was Black Friday twenty twenty one. And that was when we did this this 20% off yearly plans. And we gave it to existing customers, and we gave it to new customers. And here's an interesting thing.

Justin:

I think I can show you this. I did a little do you ever use Sigma that in in Stripe?

Brian:

I I know you've you've you've showed a bit of it before.

Justin:

Okay. I I can't I think I can show you this. But, basically, I I created a little Sigma thing. I'll I'll show it. What the heck?

Justin:

Here, I'm saying I just basically said, compare the lifetime value of a regular customer and the LTV of anyone who used this Black Friday coupon back in 2021. Well, average LTV for Black Friday users is over a thousand dollars. And

Brian:

Oh, interesting.

Justin:

And then the average LTV for all other customers is $543.

Brian:

Alright. Well, just to question that number there, that's Black Friday twenty twenty one. Mhmm. But when you look at the all other customers, is that factoring in, like, recent customers?

Justin:

Yeah. Yeah.

Brian:

So that's gonna drag it down, isn't it?

Justin:

Oh, yeah. That's a good point. I should have I should have I should have

Brian:

Like, you should do, like, all other customers that started Yes. Yes. Yes. Before, like, 2022 or something like that.

Justin:

Yeah. Yeah. I still think because I know our average calculated by Stripe, like, in our dashboard, and Black Friday is still higher. So the I think the LTV we get in the Stripe dashboard is, I'll just say, $900. So this is Black Friday is still above average by 2 or $300.

Brian:

And also because it's like Black Friday, they're getting an annual plan too. So that'll that'll increase.

Justin:

That's the big thing. Yeah. Is Mhmm. Is they're getting an annual plan. It's nice at the end of the year, even from a cash flow perspective for you know, we like to give our employees a profit sharing bonus.

Justin:

So it's nice to have a big influx of cash at the end of the year. We have to pay our corporate taxes, you know, pretty quick after the in into the New Year. So it's nice to have cash for that.

Brian:

That's a good point about the annual. Like, I should have I I do have I actually do a Black Friday for Clarity Flow, and it's always like sort of doesn't doesn't really move the needle all that much, but it's I think it's something like a three month discount. I should just do an annual discount.

Justin:

The advantage again is like in terms of LTV, I think and I can run some more queries on this. But I think generally, our annual customers have been higher LTV because they're paying it. And then it's like even if your payment goes through again, and you're like, do I want to keep this podcast up? Well, I'll just give it another year, and then we'll evaluate in a year. So people give it more time.

Justin:

New podcasters don't just, you know, start and then be like, oh, I'm giving up now. They usually will give it more of a go. So it encourages people to it gives people more time

Brian:

paid for the year, so I might as well

Justin:

Well, it gives people more time to breathe and be creative and, you know, let it happen as opposed to, oh, dang it. It's been two months and I still haven't recorded an episode. In this case, people are like, well, you know, three months in, I'll record my first episode. And, you know, we'll get going from there. So

Brian:

I am gonna probably do a I am just gonna do a a Black Friday on Builder Methods Pro membership. Mhmm. So it'll be interesting. Like, this is the first Black Friday for this new business. So it's like everything right now is like a first time for everything.

Brian:

So I'm getting, like, the very first data in. Yeah. And I've done Black Fridays for most of my products in the past. I don't think I really did it for audience ops, but I did it all the time for, like, for Productize, the course. Like, that being a course sale, like, that was always a big influx of cash was the productized Black Friday sale.

Brian:

Yeah. And then this one so it'll it'll it will be interesting. Like, I don't know what to expect with you know, I'll probably take a $100 off the Builder Methods Pro membership for your first year. Yeah. And I have noticed so we're in November, and sales were pretty good at the November for Builder Methods Pro.

Justin:

Mhmm.

Brian:

I'm getting, like, roughly, like, at least one new sign up a day, sometimes multiple. Mhmm. Which has been great. Like, revenue and and customer growth has been really solid for the first three months straight. And if this continues, I'll be flying high.

Brian:

I'll love it. Yeah. But I but I did notice, I would say in the past one week, it, like, noticeably slowed down. And so my guess there is, like, maybe people are anticipating, like, it's that there's probably going to be a sale on Black Friday. Yeah.

Brian:

That's my one guess. But then if I also predict the flip side of that, maybe the reason why Black Friday won't work so well in this business is because I sense that a lot of the customers buying are just using their company credit card from their employer.

Justin:

Yeah. But still, credit card I have this distinct memory. First startup job. And, you know, Twitter's just starting. There's just starting to be this kind of like, I'm just starting to get into like, these kind of group am I in?

Justin:

You know? So I'm following all these people. I'm following all the Panic people, you know, and they make all all the different Oh, the

Brian:

Panic software?

Justin:

Yeah. And at the time, they had they had that well, they still have their code editor. And I just all the cool people I knew were using Panic's

Brian:

code back in the day.

Justin:

Yeah. And so I was I I just remember, like, me and this other designer at the startup were, like, super into Panic as a company and just like wanted to support them and get in on this the cool kids club that were using Coda. And they had a Black Friday sale, and it just made it so much easier to pitch it to the boss. Like, hey, like, here's this really cool editor. We've wanted to try it.

Justin:

It's on sale. And my boss is like, okay, sure. Whatever. And, you know, we bought it. So

Brian:

Interesting. I mean, I feel like that might have just been your shop. I don't know. I have no idea. Like, I don't I don't know what the real dynamics are in these companies where it's like you get the Like, my I question, like, is Black Friday a real driver if I mean sales are coming from, like, business

Justin:

Let me see. How many corporate

Brian:

I feel like Black Fridays are are more for, like, consumer or prosumer. Like, I'm using my own money, and I'm I'm gonna I need to save a few bucks.

Justin:

I mean, even for us, like, what's let me look at redemption so far. I'm looking just at the top 10, and there's two Gmail and eight corporate domains. So eight out of 10, and at least on this screen, I could load more, but are corporate domains, not, you know, Gmail.

Brian:

So yeah. I mean, it'll be interesting, you

Justin:

know. Mhmm. Mhmm. Yeah. I I think for your product, it makes sense.

Justin:

I think you also have a lot of people that are that are it it also just because, I

Brian:

mean, there there are plenty of, like, independent, like, people using their own. Yeah. And it's it's not all employer

Justin:

based customers coming in. It's like you just need another touch point. Like, one thing I do think that happens a lot inside companies is people need a nudge. So they get nudged once and they're like, oh, I should talk to the boss about that. And then it's like, ah, that's uncomfortable.

Justin:

I'll just leave it. And then there's another nudge. It's like, oh, wow. Now Brian's released all this stuff. I should really talk to the boss about that.

Justin:

Okay. And then, you know, then they're in a meeting and the boss goes, hey, everybody. Like, we wanna spend some money if year end purchases. So if you have anything you wanna put through, just let us know. And then they're like, oh, I'm gonna now's the time to get it.

Justin:

Right? So multiple

Brian:

I know that there's a lot of companies who are like, we we need AI training, so spend whatever you want

Justin:

on that. That's right. What's their number one stress going into the holidays? Mhmm. Anything else you really wanna chat about?

Brian:

Yeah. I mean, actually, like, speaking about that, like, the big thing on my mind right now is it's it's starting to shape up in my mind. Like, this is, like, phase two of builder methods.

Justin:

Okay.

Brian:

And it's it continues what we were talking about earlier, like, the process of like I do think that I I feel like I'm starting to step back and say, like, there's a bunch of things that I that I'm really liking about builder methods. There's a few things that I feel like are incomplete or still have gaps. First of all, like, just the revenue has been solid. Like, first three months, it has already crossed or exceeded, like, my target revenue for for a business. And if this if it just matches or exceeds for every month for the next six months, I feel like, finally, like, this is the business that

Justin:

I'm gonna

Brian:

for for a while. Yeah. Well, that's the deal. That way right now, but, like, it's only been three months. You know?

Justin:

I mean, we should definitely applaud that. That's great, dude.

Brian:

Like like You know, it's like above 5 figures a month in revenue and Nice. Just crossed 20,000 subscribers on YouTube. Like, it's Wow. It's it's a solid thing now.

Justin:

The funnel's working. People are seeing videos.

Brian:

Yeah. Like, it like, I have a distribution channel. The email list is growing every single day. Customers are buying almost every day. The workshops have been selling pretty well.

Brian:

The the the areas where I feel like I think it's pretty clear that I'm missing the flagship content piece. I I think I have the product in Builder Methods Pro. Like, that's gonna remain the thing that I'm selling. But what you get inside feels incomplete still.

Justin:

Okay. Okay.

Brian:

And I think the thing that's missing and I I have had, I would say, two I think, like, two people have have asked for cancellations and refunds. And and they're basically citing like, I was I was sort of looking for more structured training.

Justin:

Okay.

Brian:

You know? And I don't really have that. Like, when you get inside right now, you get the Discord. And I do have a bunch of, like, private videos in there, but it's just me. Like, here's here's the thing that I built today.

Brian:

Let me show you how I built it. It I don't have, like, a structured curriculum or course or framework. There's Agent OS, but that's sort of like a separate thing. And and honestly, like, if I if I'm really honest about it, I feel like Agent OS still attracts a lot of users and interest in my stuff. But I'm personally starting to distance a little bit from Mhmm.

Brian:

Using Agent OS because the underlying tools are getting so much better now that, like, it it it starts to become a little bit too heavy handed for Yeah. Like

Justin:

It is actually better for you to just take that philosophy and help people. Like, instead of you having to maintain this package that people are installing and, you know, have questions about, like, just let Cursor deal with that. And then you can still you can still, you know, get all the benefits of having done Agent OS because now you're the spec development guy. But now it's like, hey, you know what? Cursor does this great, but here's the philosophy.

Justin:

Here's how to use it.

Brian:

That's exactly you you nailed it. That is exactly my my I've started to use exactly that kind of language in a lot of my videos lately. And and so my thinking now like, now that I can reflect back on the process that I experimented with FilterHawk, I'm gonna run that process again on whatever my next little tool's gonna be. Yeah. And then again after that.

Brian:

And so now I'm starting to develop an actual development process for because I really think that this is what we, as professional builders, actually need.

Justin:

Mhmm.

Brian:

It is I think part of it is like we we all still a lot of us still think in terms of being hand coding, like, full stack developers and and using AI to help us hand code faster.

Justin:

Mhmm.

Brian:

But we're now entering the phase where it's like, no. Like, we have to rethink from fundamentals.

Justin:

Mhmm.

Brian:

We are context engineers now. We are agentic. Like, we our role is to be product managers, product architects.

Justin:

Mhmm.

Brian:

And the actual coding generally does should be be done by AI, and we interface with the AI agents. And we have complex, highly capable tools for doing that now.

Justin:

A strong stance.

Brian:

It is. But honestly, that is where we are. I know there's a lot of people who are not on that page yet. I get that. But look, these tools and the models underneath them have advanced so much just in the last, like, four or five months that it's I mean, if you just look at this new Google antigravity thing, I haven't even used it yet, but, like

Justin:

Google anti

Brian:

It's the it's like they're they're they're competitor to Cursor.

Justin:

Okay. Okay. Man, this place the space moves so freaking fast.

Brian:

So fast. It's so so fast, dude. It's insane. And so, anyway, my my thinking here is, like, I I do wanna develop this process. I'm thinking of it calling it, like, the builder method.

Brian:

And it's and and it's what I described earlier, like, how I went about building, like, feed FilterHawk and and my general process of, like, you know, idea to the design phase with the JSON to the you know, getting mock ups and then developing specs with plan mode, building them out with with agents, doing some refinements and getting it shipped. And I think that if I can build it into this like structure, then all of the like, the comp the the tools and and the and the tactics are always gonna keep changing super fast.

Justin:

Mhmm.

Brian:

So we can always maybe you like cursor. Maybe you like Cloud Code. Maybe you like anti gravity. Maybe you like Versus Code. Like, it doesn't matter.

Brian:

Maybe you like Agent OS. Maybe you like Plan mode with Cloud Code. You can still run a spec driven development process using any of these tools. Yeah. You know?

Brian:

And, like, that's sort of, like, the mindset. And and we in the community, we can we can be talking about, hey, have you checked out this this this workflow or this new tool that just dropped or this new model? Like, does that slot in to the builder method way of building? You know?

Justin:

I do like that the builder method. I I think there there does seem to be a lot in here that's interesting to me. One is that there is some friction. Like, as you're saying, like, this is the way people are building things. Like, this is the way to do it.

Justin:

I can hear all of the people on the opposite side of that that aren't convinced yet. And I think there's a lot of interesting content. I I even noticed the the video you teased together inside of builder methods is Yeah.

Brian:

I dropped one today. I was actually speaking to

Justin:

that. To unfair advantage. So

Brian:

it's perfect content. The topic for for today's video. I was on the Ruby oh, I was on one of these, like, Ruby on Rails podcast last week. And and, you know, a bunch of a lot of Ruby on Rails developers, just like a lot of Laravel developers and everyone, they're still like, man, I feel like I'm losing my craft

Justin:

Mhmm.

Brian:

To AI. And I'm and I'm trying to push back on that. Like, look, we still have our craft, and we are actually at an advantage having having been full stack developers and engineers and architects. But the tools, like like the actual writing of code, that is a solved problem now. It like, I'm I'm sorry.

Brian:

You just cannot make the argument to me that, like, it alright. It makes, one mistake or or code something in a slightly different style than you would, and and it's ruining your whole code base. Like, I'm sorry. No. That's not true anymore.

Brian:

Like, might have been true a year ago. It's not today.

Justin:

What do say to the people, though, that, you know, there's gonna be a lot of experienced developers that are like, I like that part. Like, in some ways, that was, you know, that one discussion we had with Adam, that was the the the clip that went viral was Adam saying, you know, AI just automates all the boring boilerplate stuff and lets me hand code, which is the the thing I like. But now it feels like you're going another step and going, so do you have like, I think people can still hand code, I'm guessing, and get benefits. But

Brian:

Look. Peep you can just do things. Right? You can always just do things. And if you if you love the art form and the poetry of hand coding your own programming on the page, I totally get that.

Brian:

I I love that too. But at the end of the day, in a professional context, at a certain point, it just becomes ridiculous. Like, it's one thing if you're doing it for fun and for craft or for a hobby or for your own little MVP on the side nights and weekends. You wanna you wanna craft it the way you wanna do it? Fine.

Brian:

But you can't make the argument that, like, it's not like the that these tools are going to slow companies down or produce worse products than like like, not in 2025 and not in 2026. Like, this is a new time. Yeah. I'm just like I I know I sound like I'm a whatever, like a AI boy on here, but, like, that that is where we are. Like, it's so it's so much better.

Justin:

You need to have a strong opinion about it. I think that I think that part's great.

Brian:

I I think I but also it's like, why did I get into it in the first place? Right? Yeah. I like getting technical with code and everything, but really I'm a product guy. I learned Ruby on Rails because I want to design and ship products.

Brian:

And so we cannot, like AI is incapable of knowing what product to build, knowing how to solve the problem for the customer, understanding the customer's need, job to Having be intuition. Having Even even from architect from the technical architecture, like deciding on the right tech stack or the right technical approach for solving this problem. Like, that's still us. And that that is where we come to the table with an advantage over the nontechnical vibe coders.

Justin:

Right? Yes. Yeah. Yeah. I mean, you should see the chat.

Justin:

It's gone crazy because I think we were hitting on a nerve here. This is why this is such a great content topic for you because I think there's all sorts of other interesting things like for the person that loves to hand code. There's lots of opportunities for you to teach these models how you do things. Like, you could walk through how you're coding things, record the transcript, or use Whisper or whatever, feed that in and say, hey, I want you to hear my process of thinking as I'm writing this code. And I would like you to replicate that style, that those kind of decision making matrixes into your model.

Justin:

And you can still hand code, but now you're just also using it to train the model and make it better. And one day you might, you know, be like, man, I'm behind, and I need I need something that writes code pretty close to the way I do.

Brian:

Yeah. And, you know, it's it's not just about going faster. Like, we are like, I am building insanely fast. Like like so much faster than before. But it's not even about that.

Brian:

Like, I I actually think that I'm building much better final products when I can when I when when my mode of building is interacting with an agent. You can go deep on this. You can get technical on this. It's not just a box on a on a lovable interface in your vibe coding. Like, there there are more layers to this if you start to lean into it.

Brian:

And also like in a professional context, like for companies building products, like there is a reason why CEOs and CTOs are telling their teams, like, we need to get on this. Like, they're not just following trends. Like, there there is a sense of, like, sort of missing the boat, falling behind. And so and some of that is, like, justified.

Justin:

I will say that I think for those people with experience, your experience is still going to be incredibly helpful in how you guide and review code produced by by, you know, AI. You know,

Brian:

I That that's also the big misconception too. Is like, I think I think a lot of people are like, oh, AI is just gonna replace us. Like, I don't believe that at all. I I think software profession professional software builders, we very, very much have a role here, and and that is architecting. That is prop owning of the products.

Brian:

Like every every individual on a like like any professional SaaS is still gonna need to have software professionals using agents. Yeah. You know?

Justin:

I think what's challenging about AI in particular is it it is it is way bigger and different than other abstractions we've had before. So nobody writes machine code anymore. And when we went from machine code to the next abstraction layer up, there were a lot of machine coders who were like, no, we shouldn't be doing this. We're gonna have you're gonna have memory leaks. You're gonna have all sorts of problems that we can deal with at this stage.

Justin:

You know, you're gonna build bloated software. And then it was like, well, no, let's get to I don't know what was first c. Was it c or c plus plus? And then it's like, well, here's, you know, how many people write in those languages or COBOL or any of those. No.

Justin:

We have and then you even get up to like Ruby, which is like a very abstracted language. Yeah. That's running in the browser. And then it's like

Brian:

And then you got Ruby on Rails. Right?

Justin:

And then you got Ruby on Rails. And it's like the all these arguments was like, oh, Ruby on Rails adds this whole scaffolding and everything that who would ever use this? It's such a then and it's like, well, it's actually better. You know? I I don't wanna have to write all that boilerplate.

Justin:

And the so this is there's one argument, which is just we're just following in that same abstraction line. But AI is different. It's not like we're just like this isn't not just like another nice layer on top of abstraction. This is even bigger than that. It's more multidimensional than that.

Justin:

So I can understand Yeah. Strong feelings around it for sure.

Brian:

I yeah. I get it. But I also like I I think because it it is such a huge step change, like, step change is is not even a strong enough word. It's like what you're saying. It's it's a much bigger thing.

Brian:

It's that's why I'm thinking in terms of, like, I think I need, like, a, like, a process, like, a course or framework that says, like, let's rethink our whole process from the ground up

Justin:

Mhmm.

Brian:

With an AI first approach. Right? Like, the design phase, we do that, you know, in collaboration with with Claude and a JSON sample data, and then we generate a a design off that, and then we iterate off that. Like, instead of like hand coding in Figma or hand designing in Figma, you know. Or or yeah.

Brian:

Like like, it's it's just like, I think I think the the and this is where, like, Agent OS is starting to feel a little bit outdated. It was like that was like training the the agents to hand code like we hand code. But really, we need to we need to go deeper than that and say, like, if we're starting from scratch in in a world where we have these AI tools available to us, what's actually the optimal process from the ground up? And, like, that's how we need to start to think about it. You know?

Brian:

Yeah. And and like and and I and that's exactly what I did last week, And it resulted in building what I think is like a pretty nice little finished product in four days. You know?

Justin:

Yeah. And there will be lots of examples of shitty products made with AI. Like, that will be true. Yeah. Just like there's lots of examples of shitty products made by indie hackers before AI.

Justin:

That's always been the case.

Brian:

Yeah. There's shitty products products from huge companies too.

Justin:

Yes. Yes. And I I think a course like that would be great. I think especially if you do it in cooperation with some teams, like start to figure out how teams because even I can see like a pull request review process totally changing. So instead of saying, oh, hey, like this, this, and this, go, hey, we don't really, you know, structure things this way in our code base.

Justin:

Why don't you take this prompt, bring it back in, and say, we want to get it kind of in this area. And you're going to be there's going to be more back and forth in that that kind of way, you know?

Brian:

That actually brings up one more thing I wanted to get your feedback on. Like, I'm thinking about like the areas where I have gaps right now in in the business, in my visibility, in my interaction with customers. And I get a lot of survey data. I get a lot of feedback. I get a lot of just straight purchases of Builder Methods Pro.

Brian:

But I still feel like I don't fully understand my customers, or I don't really know all the different customer avatars and the and the and the true pains and problems and things that they're dealing with. And and especially in the Teams context. Like, you know, I just have my small shop here. It's just me and a couple of assistants here. And a lot of it is just me solo.

Brian:

So I know that a lot of the buyers are coming from these larger companies with larger teams, which I don't have as much personal experience with, at least not lately. So you know, and I do these monthly workshops where people are coming in and they ask a lot of questions and stuff. But I'm starting to I'm starting to kick around the idea of like maybe maybe one of the new benefits in Builder Methods Pro will be some sort of office hours. Maybe once or twice a month, I'll throw up a Zoom. And it's not a workshop where I'm talking.

Brian:

It's more like, hey, members, whoever wants to join in, it's sort of like open q and a. Like, you come and bring your questions and

Justin:

Yeah.

Brian:

And I'll hang out. And and and I'm I'm just here to connect you all. Like, I'll I'll give you my opinions, but but really, I wanna hear from team leaders. I wanna hear from independents. I wanna hear from nontechnical people.

Justin:

Especially if you targeted the office hours. This is office hours for teams or team leaders. This is office hours for individual contributors. I think segmenting those, you'll get because the worst thing is to have a big office hours. One person saying, how do I start my computer?

Justin:

And the other people are saying, like, there's just a mismatch. So I think segmenting it would be good. The the other thing I think I would offer if I was you is you're already getting inbound from teams. I would pitch a team on saying, hey, my normal rate is whatever. $500,000.

Justin:

Instead of you paying me 500, pay me 200,000 and agree that we are going to document the whole thing. So we're going to document we're going to make a little documentary on this, which is just we'll do calls. But I'm going to hear from your team members that are pushing back. I'm going to We're going to work through some of these common issues that people are Real people are having. When I talk to the people here in my co working place, it's all over.

Justin:

People who have they're like, I I know I should be checking this out. I haven't even looked. Like, I've used ChatGPT for my own productivity, but I haven't even pasted any code in a web chat box yet. But I know I should. And to other people who are like, I'm trying to convince my company, you know, my company just got acquired, and I'm trying to convince the boss we need to be doing this.

Justin:

We're moving so slow. And it's I I need some help in convincing the boss that we got to do this. And here's the benefits.

Brian:

You know, it's interesting. I had a few I had a bunch of leads like that in in, like, two months ago. Some of them were from huge companies, like 400 plus employees. Some of them were from 50 employee companies. A couple of them were, like, smaller shops.

Justin:

Mhmm.

Brian:

I had a bunch of sales calls with those people. I pitched different quotes on like me giving a private workshop to their team, to their large teams, having like a one week open office hours with their teams, and sort of had a bunch of calls around like, these are the pain points. And exactly what you were saying, like maybe everyone is sort of like slow to adopt, but there's like 10% who are super hyped about being champions of AI within the company. And it was weird. I had a lot of, like, leads and prospects and really, I thought, good sales calls, but, like, no no bites.

Brian:

Or I had a few small small engagements that went through, but, like, nothing like, it it wasn't like, oh, like, shut up and take my money. Like, they sort of just went cold. And then and, you know, I I did some follow ups and stuff, but not not too hard. And then and then at that point, I just started to really focus on the monthly workshops and the Builder Methods Pro. And and in Builder Methods Pro, did add a a way to buy multiple seats, and a few people do that.

Brian:

But for the most part, it's like individual sales. Although, I I think that they are using company credit cards Yeah. For for a lot of them. So I don't know. Like, part of it was also like, I don't wanna spread myself too thin, but I I do still agree with you that there is I hear it all the time.

Brian:

I see it. There's a need out there with larger teams. And if like, maybe there maybe it's it's just like a special sales page for Builder Methods Pro that's aimed specifically at teams. But but really, you're just buying, like, 50 seats for Builder Methods Pro.

Justin:

Oh, I I think you should offer that for sure. I I I think this is a hard ask, especially for a lot of these companies that are having this challenge. Because a lot of the companies that are having this challenge are not the kind of companies that wanna be a case study. But it might still be worth trying it a few more times, like a call out for people sorry, companies that might be interested in being a case study and being specific. I'm looking for a company, a team that is considering adopting AI, but hasn't really and there's lots of pushback internally and confusion.

Justin:

And I want to just do three or four calls with people on your team and see if we can work through this. And, I mean, I did actually Stripe invited me. Stripe does these like team office hours that are so interesting. So they they assembled the whole team that was around this one feature. And then they had me call in and just say, hey, like, just want you to talk about your experience with this particular thing.

Justin:

And then I talked, and then they all just got to ask questions like And I could see something similar with you just like, hey, like, let me call in, have your whole team assembled, and then let's go back and forth. Like, what about this? What about this? And I mean I

Brian:

sort of pitched versions of that. I I I probably could do a better a better follow through on it. But I yeah. I think a lot of I I still think like the whole it's like these teams and the and the leaders, the CTOs, the CEOs, like, they feel this need and this and this pain, but they're also like, it's so new that they don't know what to buy or or how to solve that problem. They they know they feel it.

Brian:

They don't quite know the shape of of, you know, what they need to do about it.

Justin:

And a lot of it is going to be giving I mean, I think a lot of people are already taking your stuff, especially this video that you just is this public yet, this video, or is it it's not released yet?

Brian:

Which one?

Justin:

The one you just talked about here, Master These this morning. Oh, okay. Okay.

Brian:

On my on my channel. Yeah. I actually made it, like, the cover video on my channel today. Yeah. Yeah.

Brian:

Master These Skills.

Justin:

I mean, that's the kind of video that people are gonna share internally. Go, hey. Like, I know we're having these conversations. Like, here's something else that might here's listen to my friend, Brian. He's he's got some ideas, and maybe we should just explore this as a team.

Brian:

Also, part of it is like, don't I I still need to learn a lot, especially from what's happening in teams. Like, I I have this concept for this process, the builder method, but I need to see how that lands, how that gets adopted in Teams. And that's that's where I think, like, a a series of office hours and hearing questions and hearing like, yeah, I tried to implement this with my team, and here's here's what went down. And like, I I need to get a feel for that because I don't know that I feel qualified to talk to to a team of know?

Justin:

As as long as you can as long as you are just hosting in a sense to say you can say, listen, I have a guy that's done a lot of this. I've explored this. I haven't been a part of a big team in a long time. But I I do have a perspective that might be helpful for you. Because some of the questions might just be like the CEO going, well, you know, I haven't heard much about this.

Justin:

But like, what do you think? If we were gonna invest in one thing, what should we do? And you could say, well, here's a couple ideas, you know, cursor or cloud, you know, code or whatever. Like, I would just get started on that.

Brian:

Yeah. I also just think like having a framework. Like having I don't think that it's AgentOS anymore. I think it needs to just be like a more general process that any tools can slot into. Because that's the other thing about teams.

Brian:

It's like, usually most teams are like, well, we're we're really all about Jira. Or or I we have the whole team using Cursor. We have the whole team using Versus Code. So, you know, or or Cloud Code. So, like, it's like you should and I and I think that that's health.

Brian:

That's the right way to think about this era that we're in. It's like you could use any tools. They're all still following the same process. You know?

Justin:

The other thought that just came to my mind was you could do a video. You can have videos and content that are like the one demo that will convince your team to adopt AI. Mhmm. Like, I think there's lots of companies. Like, there's this, like, tiny iceberg of companies that have even done any sort of adoption of AI.

Justin:

And then there's this giant massive iceberg even within technology. And we're talking about like enterprise software and consulting. Like most of the GDP is probably actually here. And they are dragging their feet. They are not in this.

Justin:

And I think people are looking for tools on how what can I show my boss? How can I convince my team? What can I what assets are you gonna give me? Like, you know, six Slacks six Slack messages to send your CTO about I mean, that might be a bad one. But, you know, like, those kinds of assets, I think, are are I think will be interesting.

Brian:

And yeah. But that's also like I don't mean to push back on this because I think I do agree. I I I felt like there there has been a missed opportunity here with teams and with larger budgets than independents.

Justin:

Mhmm.

Brian:

But at the same time, I gotta be honest, there's part of me that's like, well, I'm an independent builder. So, like, I my comfort zone is talking to the solopreneurs, the the small SaaS companies, the the transistors of the world. Right? Like, the the the teams of five ten. Like, those are the people that I actually relate to.

Brian:

And and and I do and and I also do know for sure that there is a pretty big segment of customers in Builder Methods Pro who are that. Like, it's it's not all teams. There definitely and and there are also some people who are, like, truly nontechnical. Like, they're I've had the question come in, like, hey, I know you're sort of marketing to professional developers. I sort of don't know how to code, but I've but I've been following your stuff and I'm into it.

Brian:

So I I still get a segment so it's like I'm I'm still trying to figure out who my ideal customer is, both from where is the most demand, but also, like, what's the best fit for me, you know?

Justin:

Yeah. Here's here's a here's one more perspective. Independents, fast moving independents have always been the ones that are kind of out blazing the trail and then writing blog posts and creating videos that then get shared around big companies. It's the it's the independent, the small, agile, crazy one that can figure a lot of this stuff out. But that stuff absolutely gets shared in big companies.

Justin:

I have this hilarious example, which is this post on Hacker News. How remote work changed my life ended up being how actually, we started this coworking place because there was a guy working in Vernon who worked for SAP, one of the biggest enterprise software companies ever, that saw this blog saw this post on Hacker News, read it, and shared it in their local chat, whatever they were using. And, you know, ended up generating a great discussion. And he used it as kind of an asset with his boss to say remote work can work, you know, for a company like ours. And I'm I was just a little independent working for a small little software company.

Justin:

So I I think we have we've always had more influence as, you know, the the crazy ones, the ones that are out doing stuff. And in some ways, enterprise companies want think I this is why thirty seven signals still has so much voice. Is I think there's all these people in crusty old companies that are like, hey, finally, I've got something I can show the team and say, I know we can't go all the way, but here let's use this for some of it. You know? And

Brian:

And you know what? There are indies who are really using AI. And it so I started recording this new show. It's not published yet, but I've I've recorded three episodes. I it'll probably just end up being called called, like, the Builder Methods show.

Brian:

It'll be on the YouTube channel, probably a podcast version too, but it's very visual. I'm inviting fellow independent builders, people building. Show me your screen. Show me your workflow. Show me the tool.

Brian:

What do you like Cloud Code? Do you like using GitHub issues? Do you like using cursor? Like, what's your what's your flow? How do you lay out your screen?

Justin:

Like Yeah.

Brian:

It's super fascinating. So I had Arvid Kal on the first one. Oh, cool. Spoke to Mike Buckby. Spoke to Christian Jenko.

Brian:

Talking I'm to Brendan Dunn next week. So these are guys who and what's so fast I'm so psyched about this new show. I can't wait to get it out. It's so fascinating how every single person is doing it so differently. And and these are guys who are fully embracing AI in in their workflow.

Brian:

Like, Arvid is like his product, like Podcast Scanner, I think it's

Justin:

called. PodScan, I think?

Brian:

PodScan. Yep. Like, he's all about Cloud Code. Like, it's it's it's generating all of his work, you know? Mhmm.

Brian:

And he showed me his whole screen and, like, everything. And it's fascinating, man. And it's just so interesting to see how we've all got these different little tools and best practices. And some do use sort of like a spectrum and some don't at all. And it's a totally different approach, but they're still using it.

Brian:

You

Justin:

know? Yeah. Yeah, dude.

Brian:

I just gotta figure out how to get this, like, edited and post productive.

Justin:

I maybe just gotta do it live. You've got all this practice just doing things live now. Look at us. We're we're we're live we're live streaming this.

Brian:

I know, man.

Justin:

Alright. Well, let's leave it there. Yeah. This was a good one. I did oh, I gotta go.

Justin:

Alright.

Brian:

Alright. I'm gonna go then. This a

Justin:

good one. So much great interaction in the chat. Folks, if you're listening at home or watching at home and you missed the live chat, we record most Thursdays at starting at 12:15PM Pacific.

Brian:

Yes, sir. I guess next week, it'll be Thanksgiving over here. So

Justin:

Yes. We'll probably skip next week, I'm guessing. Come back next week because Dave Giunta will have written his opinionated blog post about AI and teams. I want to hear that. Yeah.

Justin:

And oh, yeah. Keith, you're right. Cold plunge part two. That's Canadian Thanksgiving already happened. There's no reason.

Justin:

Could join me in the sauna next week as we go live.

Brian:

Yeah. There you go.

Justin:

Thanks, everyone. Talk to you soon.

Brian:

Alright. Later, folks.