0:00 Brad You know, we're giving up way too much agency to these soon to be monopolies or duopolies like these, these players are too big power to the people. Let's open source open weight this, this bad boy. Welcome, everybody, to another episode of the How Many CTOs Does it Take podcast. My name is Brad Hefta-Gaub, and this is my trusty co-host.
0:37 Scott Scott Poirier, how are you doing today, Brad?
0:39 Brad I'm doing pretty well, Scott. How are you doing?
0:42 Scott Good. How many tokens have you maxed out today?
0:45 Brad Okay. So yeah, so you know this will come out like a month from now, but I was on vacation last week for my honeymoon.
0:54 Scott Oh, we forgot to talk about that.
0:55 Brad Well, that's okay. We can say like, hey, I was on vacation for my honeymoon. It was a great time.
1:00 Scott How was your – I mean you went to Hawaii. I think we can disclose that.
1:03 Brad Yeah.
1:04 Scott Where did you go?
1:05 Brad We went to the Grand Wailea on Maui. it's one of our favorite little resorts i haven't been there in a long time that had been like 15 years since we were there last and so it was a great it was kind of a fun felt like a fun little homecoming it was mostly just um hanging out at the pool soaking up the sun drinking my ties
1:24 Scott Well, that's what I was going to ask you. I was going to first ask if you were beach people or activity people. It sounds like you're beach people.
1:30 Brad this time this this this trip we intentionally were like we're gonna like it's gonna be slow motion we're gonna not we're not doing the active i mean normally i'd go do a bunch of scuba diving and we'd go you know into towns and drive the road to hana and do all that stuff but this time we both intentionally were like no you know what we're gonna do we're just gonna be super chill super relaxed and so yeah it was hanging out at the pool yes
1:56 Scott And so it sounds like your pool people versus beach people.
2:00 Brad that is i mean i like the beach uh my wife not so much we
2:04 Scott Oh,
2:04 Brad we did
2:04 Scott really?
2:04 Brad we walked on the beach for the sunset a couple nights so that was nice yeah
2:07 Scott That
2:08 Brad did that
2:08 Scott sounds great.
2:08 Brad yeah it was fun time it was a great time 30
2:11 Scott Well, congratulations on making it this far. That's an accomplishment in and of itself.
2:16 Brad years yeah it's long it's a it's
2:18 Scott Crazy. How's it feel?
2:20 Brad Uh, you know, it's funny, a friend of ours just is having their anniversary in a couple months and he was asking me that and I was like, you know, it feels like both yesterday and also like time flies.
2:30 Scott Mm hmm.
2:31 Brad So absolutely,
2:31 Scott Yeah. Time does fly more than half your life by definition for the rest, assuming you stay married for the rest of your life. That'll be more than half your life.
2:40 Brad that part I am looking forward to the idea of 30 years is half my life that didn't
2:45 Scott No,
2:45 Brad sound right.
2:45 Scott it's not great. No, come on. 60 is the new 40.
2:49 Brad 60s in the new 40 exactly exactly anyway so thanks for thanks for asking about that well you start you asked me how many tokens i maxed out and i'll tell you that the funny thing is i mostly so i will say tuned out from all the work life right i did do a fair amount of playing around and uh while sitting by the the pool and with the wi-fi i worked on some personal projects so that was i mean that was like vacation mode so i was playing but i did get one message from one of my teams that were like oh no our github account says we've blown our budget and we can't use code reviews anymore and i was like what happened you know because this is one of the teams that is like i always tell you like that super high performance team they've got it all dialed in and they actually don't spend that much money on token they you know they they all have like
3:42 Brad basically like they all have like a 200 month a 200 a month stipend for whatever ai tooling they want Mm and they
3:50 Scott -hmm.
3:50 Brad pretty much that's all they use they don't they don't ever they're never there's never massive overages but it turns out that like the budget that had been set on the uh on the organization wide github was fifty dollars And GitHub did like switch something where it's like they went from like all your co-pilots are included with your with your pro account to. Now you're paying per
4:16 Scott Right.
4:17 Brad token. And that happened literally like, I think that happened like the day I was coming home from my anniversary vacation. And so it was like that was a little mini emergency. It wasn't that big a deal. I just said, yes, pay for more budget on that. It's fine.
4:35 Scott Yeah. What happens when people start using. hosted models in their production environments and you get a version of like let's call it DTOS distributed token of service account or whatever so you're somewhere here spiking usage on It's your account
4:53 Brad like a DDoS except for a token.
4:55 Scott you know someone writes a bug and all of a sudden it's looping through tokens and all of a sudden you get a bill for a million dollars because
5:02 Brad Yeah, right.
5:03 Scott you got to put circuit breakers in on that stuff
5:04 Brad You got to put your circuit breakers in. I mean, it's a good thing that Copilot was like, oh, you're over your budget. Like, okay, with a $50 budget, I think we can increase that a little bit.
5:15 Scott I feel like I've been fighting a little bit of uphill battle when it comes to tokens.
5:19 Brad Yeah, so tell me.
5:20 Scott Now, I know there are people at my company, including other members of our senior leadership team who listen to this podcast, so there are 41 listeners. Do you have more in there?
5:30 Brad I think we might be up to 60 now, by the way. There's at least 60 subscribers on the YouTube channel.
5:37 Scott And so I might be getting myself in a little bit of hot water for what I'm about to say. Everybody in our industry is focusing on token optimization right now. So high profile companies have come out and made big splashy changes about that because people were using too many tokens. I have a lot of thoughts about that. So I'm going to come back to the big company thing. Well, let's actually just go there for a second. First is, I think there were places that were measuring AI usage. by how many tokens you were using and
6:09 Brad Yeah,
6:09 Scott so then you had clever engineers who are figuring out how to use a lot of them
6:14 Brad they weren't just measuring usage, but they sort of like gamified it and told their developer like I think there were some organizations that literally did like a token chart is like who's the leader in usage of tokens? Well, well that. I mean, that was asking for right isn't there isn't there like the apocryphal story I don't know if it's apocryphal or if it's actually really true story like in India when the you know British were you know colonized that place they had they like had a they had like a cobra problem and so then they put a bounty on dead cobras and
6:49 Scott Oh, boy.
6:49 Brad so that worked for a little bit but then basically because they were getting paid for the cobras a whole bunch of people said oh why don't we breed cobras and so they were basically breeding cobras and then turning them in turning them in for a bounty and then they and then the british you know figured out what was happening and they said okay we're not going to pay the bounty anymore anymore more so the breeders just like released all the cobras because they're just like oh okay and so it was like a policy that was just like a poorly thought through policy that then just literally came back to bite them yeah
7:22 Scott yep no pun intended or pun fully intended uh that's right and then i also wonder like i saw microsoft you know had a big announcement they were cutting back on tokens or whatever and i thought is this just a negotiating ploy with open ai right to put out in the press that we're going to stop using tokens so we can get a better rate on tokens from our token vendor but that's not the uphill battle i've been fighting you know there is in the water you know there's a meme out there not a meme in the humor sense but an actual meme of you know use it you know using tokens efficiently all that kind of stuff and i have been without a doubt in the minority at my company saying you forget about it. Just don't worry about, I don't think we should be worrying about the amount of tokens people are using.
8:07 Scott I think we should be worrying about are we getting value from whatever we're doing. And there will be some point in time when we're spending a lot of money on tokens and we will need to say, hey, we need to dial this back, but we're not there yet. And I do not think we should be doing anything to discourage people from trying to figure out how to use this powerful new tool for good, whatever
8:31 Brad Yep.
8:32 Scott you define good is for good. And I just think the sentiment that's in the air, in the water, not just at the company I'm at, but everywhere is sort of poisoning. poison is such a strong word but it's sort of cutting off our nose despite our face like we need everybody we need all hands on deck to figure out how to harness the power of this thing in
8:54 Brad Yeah.
8:54 Scott good good both good and responsible ways and we're sort of sort of focusing in my mind on this very premature optimization which causes me a little bit of heartbreak and i think it's because it's I mean, it is money, right? And money is just very countable. It's a lot easier to count that than the benefit produced. So people sort of naturally, and people can relate to money like, oh, I'd like to have more of it. I'd like to not spend
9:20 Brad as Right.
9:20 Scott much of it. So that's
9:22 Brad It's harder to relate to story points or even lines of code.
9:26 Scott right. That's exactly right. So I don't remember how we started on this third, but that's kind of where I'm at in the token maxing thing.
9:32 Brad Yeah, yeah. So let me just see if I can play that back. Because I think I think that what you just said is kind of along the lines of my thinking as well. So what I heard you say is that, like, we know that out in the zeitgeist, there's there were there were companies out there that were like, you know, like incentivizing token maxing like it
9:51 Scott Yeah,
9:51 Brad became the
9:51 Scott right.
9:52 Brad thing right token maxing was the phrase. Right. And there's now in the zeitgeist a pretty predictable backlash to that. And
10:01 Scott Yeah.
10:01 Brad I think it was maybe it was Uber that like like the CFO said, oh, we burned our entire annual budget in the first month of setting it or something like that or some ridiculous thing like that. Right. So. And your perspective on that is we're in the venture that you're working at, that's not currently the problem. And so don't worry about that. Don't worry about that problem because you're a long ways from that happening. You didn't make the mistaken incentivization of saying. you're going to get rewarded if you use lots of tokens. And so you're
10:41 Scott No.
10:41 Brad not getting that negative behavior. And instead, you're still like, hey, figure out how to use this thing and we're not going to put roadblocks in front of you while you figure that out.
10:50 Scott Yeah, but that what's out there in the zeitgeist is sort of. you know infecting the culture at our company people already thinking about
10:57 Brad Oh,
10:57 Scott oh yeah
10:58 Brad of course. I've seen it being used. Yeah, 100% people are using this as a, see, it doesn't really work. See, it just blows, it just wastes time and energy and money and water and electricity.
11:09 Scott i don't even get that it's just much of like people will be talking about something it's like well we have to be careful how many tokens it uses and i'm like don't worry about that i keep on saying look at some point the finance team is going to come to us and say say you guys need to dial back the token spend but until that problem occurs like learn learn as much as you possibly can uh like we talked about a minute ago you know put in the circuit breakers so you don't actually like you know start something that you know uses a gajillion tokens um inadvertently but we got to be learning like we do not know we
11:44 Brad do Yeah.
11:44 Scott not understand the power of this tool yet and we we got to be figuring it out
11:49 Brad Yeah. And it's funny, too, with the team that the team that I that I mentioned that like this kind of was a little flare up for them. Right. Like the scale we're talking about is such a small number. right so this is like a team of five that's absolutely cooking with gas and you know they're they're they're they absolutely represent that thought experiment that i often talk about of like every single developer on the team is generating at least 3 000 lines of new business value code every day right and they're they're you know they they're burning through the product backlog you know there was like There was like two plus year product backlog that they basically have completely eliminated at this point, you know, and
12:29 Scott Right.
12:30 Brad they were getting it done with $200 per developer and a $50 budget of shared across the org for code reviews, right? So
12:41 Scott That's
12:41 Brad I'm
12:42 Scott amazing.
12:42 Brad just like, this is not like literally, I was just I just had a team meeting with that team earlier today and they were like. I don't understand how somebody can be using thousands of dollars of tokens like they're like I'm already trying I'm doing we're working really fast like how are you doing how could you possibly be spending more than us
13:04 Scott Well, this leads to another thought, and I mean this a little tongue in cheek, but it might be true. At least Claude is not super clear on exactly. how much tokens you're using or how much they cost and so on and so forth. And I commented to someone earlier this week, I was like, is it possible that tokens are just a hoax, like a totally phony thing? And they literally just made it up so that they can make their pricing opaque.
13:29 Brad Well, I know tokens are a real thing, but but well, OK, so here's one of the other things that I see. happening in. So I would love to talk more about token maxing, but I want to make sure we mentioned like one of the pieces of backlash I'm seeing is, okay, let's go local inference then. So let's basically not use the cloud providers anymore. And can we do the work that we're doing off of open weight, open source models?
14:01 Scott Yeah
14:01 Brad And I'm seeing some really, really interesting. interesting things happening in that space.
14:07 Scott I just had coffee this morning with an engineer who was just kind of telling me about what he's doing and he's running it all locally.
14:15 Brad Well, I think that the newest the newest open weight models out there. I mean, I've seen some statistics that they're performing as well as like, you know, Opus 4.8, you know, or GPT-5.5, you know, so they're actually like the open source models are performing as well as like the frontier models. Now, maybe not as well as Fable supposedly was performing, but now that Fable's not available to anyone, it may be, you know, at least
14:43 Scott The
14:43 Brad at the recording
14:43 Scott table is
14:44 Brad of this.
14:44 Scott literally a fable.
14:46 Brad It was literally a fable.
14:48 Scott Here's the question, though. It seems like where this stuff is going is the model is certainly valuable, but it's like the harness around the model, the scaffolding around the model is a lot of the value.
14:59 Brad Yeah,
14:59 Scott you
14:59 Brad but
14:59 Scott know in
15:00 Brad those
15:00 Scott terms
15:00 Brad are getting commoditized too.
15:02 Scott no i understand that and so that's pretty interesting yeah
15:05 Brad Here's my quick prediction. By the time that your finance department notices you're using lots of tokens, there will be a perfectly valid, practically free open source version available to you.
15:16 Scott will it have the harness around it though
15:18 Brad Sure. Yeah, absolutely. The harness is the heart. You can already point you can already point quad code at you can already point quad code at an open router, open LLM router that is that's pointing to your local inference. You can already do that.
15:35 Scott Okay, well, I'm just telling you right now, this is the first thought that came to my mind when you said that. Actually, I'm going to tell you the first and the second thoughts because it's the second thought that. Actually has my stomach a little in knots. The first thought was self first just becoming free. And the second thought was, ooh, what does that mean for my 401k?
15:57 Brad Well, first of all, just to be clear, you know me, I'm still bullish on the fact that it's easier and easier to write the code, but that still means you need to know how to be a good engineer. So it's
16:11 Scott Yes, we've talked about
16:13 Brad funny.
16:13 Scott that.
16:15 Brad I've heard some other things about like, you know, is the SaaS apocalypse really happening or what's going to happen there? And, you know, you got people like. There's plenty of people in the SaaS space and other people kind of observing it that are like, no, you still need to build, you still need engineers building stuff. It's not like, it's not like, it's not like any random company is going to spontaneously on their own with no engineering talent decide to create their own implementation of a CRM, right? Like.
16:46 Scott Oh, no. Well, maybe. You got an ambitious guy in finance.
16:51 Brad Right. Okay.
16:52 Scott And we had that guy at TaskRabbit.
16:55 Brad Did he make his own CRM yet with
16:57 Scott It
16:57 Brad Lovable?
16:57 Scott wasn't a CRM. He made his sort of own vendor management system in Lovable.
17:01 Brad And is it working for him?
17:02 Scott Well, actually, he resigned to go work an AI startup and took his Lovable app with him.
17:09 Brad That sounds like a whole different podcast episode.
17:12 Scott totally like um so i want to go back here for a second we were just talking about the sas apocalypse i still think you need someone to talk to customers and find out what they need yeah
17:25 Brad Yeah, of course. There's so much to building software than just writing the code.
17:28 Scott that's something i've been in a little tussle with some people that i'm tasking about is we're trying to set some goals about you know velocity and they want the goal to be you know cycle time we want to improve our cycle time which we do want to improve our cycle time
17:42 Brad That's
17:42 Scott shout
17:43 Brad a good thing.
17:43 Scott out shout out to my team by the way cycle time is like 30 improved year over year is
17:48 Brad That's a pretty good progress.
17:50 Scott something like an average ticket was taken was 10 days now an average ticket has taken us seven days kind
17:56 Brad Nice.
17:56 Scott of thing it's
17:57 Brad Nice.
17:57 Scott pretty sweet yeah it's awesome uh and i've been kind of trying to say like yeah but that doesn't really capture what you're looking for you want to know like delivery value to customers and there's all this other stuff that like slows things down like you know prioritization clarity requirements dependencies overloading the team all this different stuff and actually there's a project that's going through right now that everybody's really It's really frustrating to buy because it's been taking a long time. Look, cycle time on this is fine. When the engineer works on it, it goes real fast. It's
18:28 Brad Yep.
18:28 Scott just there's all this other stuff he's waiting for other people, et cetera,
18:31 Brad Yep.
18:31 Scott et cetera. You actually need to capture this end to end thing.
18:33 Brad By the way, I hear you on that. And one of the things that I think that I've been coaching teams that I work with on in regards to that is like more. get more um essentially accountability to the the builders right like let the builder like let the builders free and like you could do that either by like letting your developers like own more of the product pieces like like a common roadblock i see is like developers ready to code but then they have questions and the product owner has a hard time getting back with them or something like that because maybe the product owner is getting pulled in different direction and so one solution is like hey just like let the let the the developers in this case or whoever you're going to give the building responsibility to get let them set them free like let them let them really run with it right another solution that is like change your org so that you have like these tightly bundled units that i've always have a product decider and a developer and a designer i think you've mentioned this idea and that they're a team that like they're autonomous and they're able to get stuff done right like to me that's
19:31 Scott Yeah.
19:32 Brad that's the answer that and it's not that has nothing to do with ai that just has to do with the fact that as we've talked about in the past it's like it used to be that code was the slow part and so it gave you time to solve problems or sort of answer questions and now it's like no no that's that's the fast part now so now you got to be you got to be ready to you got to be ready to know what you really want and you got to be answering those questions quickly because you're going to be able to crank out the code Very quickly.
20:00 Scott Yeah, one of the most talented engineers that I've worked with ever, I happen to be working with now, said something, he said, the product management part has always been the hardest part of what we do. It's just being exposed because as the cost of building this stuff comes down.
20:18 Brad Yeah.
20:19 Scott Now, we ran a cool experiment at TaskRabbit a couple of weeks ago. Um, I really am super excited about it. I need to get more people to adopt doing this. Um, let's just call it pair programming, but
20:35 Brad Mm-hmm.
20:35 Scott we need another word for it because somebody's already taken that brand name. Uh, there was a moment we're having this meeting. Somebody's like, Hey, is the, is, you know, is the, is the requirements document ready for this feature yet? And, um, someone was like, Oh, you know, it's not, it hasn't been written down yet. Abby has it all. it all in her head she just has to write it down and i i kind of just like i don't know how i guess through just force of will i was like hey i got an idea here could we just get abby And the engineer who's going to work on this, who I think was a guy named Evan, but it might have been a guy named Jared. But I was like, could we just get Abby and Evan to sit down with SpecKit and just skip the whole Abby writes it down and gives it to Evan step? Literally, I just want them to sit with the same computer, ideally with Abby typing,
21:21 Scott and go through the SpecKit process and she can get it right out of her head, right into the spec that AI chews up.
21:29 Brad How'd that work?
21:30 Scott It got positive, worked really well.
21:33 Brad Nice.
21:34 Scott You know, if you've been through SpecKit, the first chunk of it. is what are your personas what are your user stories what are your acceptance criteria you
21:42 Brad Yeah.
21:42 Scott know all that kind of stuff which is i mean extremely valuable to building good software and um i think it streamlined the process and at least i asked abby and she she was positive but she kind of gave this me to channel and slack and people were saying how the how it all went and i said i was like i'm not quite tracking here are you like saying this went well and you'd like to do more of it or are you just like reporting how it went and she said no it went well i think we should do more of this so I appreciate you, Abby and Evan and Jared for doing that. If you're listening to this podcast, I don't actually know if they're listeners, but they should be, they will be now. So, so I think that's the future.
22:29 Scott Yeah, I think there's a ton of value in that. I think that like that's kind of like. If if those responsibilities live in multiple people in your organization, then get them together and act and have them act at the moment together because that's going to make the biggest progress. Okay, here to me is the most profound thing of this AI revolution. I lived through the agile revolution. And I got really, really into that stuff for a while. I was going to conferences. I got to meet a bunch of the people and talk to a bunch of people who signed the Agile Manifesto. One of the coolest conference sessions I ever went to was called Agile by the Old Timers. And it was just a Q&A with three people who had signed. The manifesto, I only remember the name of it was, I think it was Ward Cunningham,
23:16 Scott Ken Schwaber, and I forget who the third person was. It was a woman and I forget her name, which is a little embarrassing. But it was just fascinating to hear from them firsthand. How did this come about? this Yeah. thing by that time like it was sweeping through the industry and it was um it was just really fun to just hear firsthand how it came about and anyway the point i'm getting at here is i think ai is bringing us closer to the promise of agile software development than ever and i think that's really really cool like Hey, sit down here with your customer, with your client, ask them what they want. What do you want it to do? Who are you? I am a user. I'm a, I'm an accounting manager. Okay. As an accounting manager, what do you want to do here?
24:02 Scott I want to reconcile these bills. Why do you want to do it so that my accounts zero out at the end of the month. Right. So like I just. As someone who for, say, five or seven years totally nerded out on that agile development process stuff, I just think it's pretty cool that it can kind of maybe bring some of that stuff really to life. I mean, I've been on lots of teams that have done weekly sprints or bi-weekly sprints and had weekly demos and shipped software where we really did it by the book and really did a great job with it. But for a lot of teams, they don't, right? It's really just kind of a two-week planning cadence and a very waterfall situation. Right. And I think this will make it easier to do it in a more iterative fashion.
24:49 Scott Well, it's funny that you say that. First of all, agreed. And then second, what I'd say is that I do think that I've seen a little bit of a I've seen a little bit of a backlash against SpecKit or not SpecKit specifically, but SDD kind of. kind of more broadly speaking by people who are advocates of agile where they say like hey the sort of like over emphasis of knowing exactly what you want ahead of time feels more water folly than agile and so it's interesting that you actually are calling it out as an example of a tool to help us do agile right and I've actually seen other folks that have have had a little pushback back i think both can be true you know because i think that the pushback i've seen is a little bit of the thing that you and i've talked about in the past which is one of the parts of agile is you want to be able to get enough that you can criticize you want to put something together enough that you can like you know interrogate it and criticize it and so if you spend a lot of time and energy up front trying to like be polish your specification
25:55 Scott then you're slowing down your ability to get something up and running enough that you can interrogate it and criticize it, right? And so it's this balance, right? Yeah, I mean, look, you can do, use spec kit iteratively, right? Each step along the way can be a little spec. I think if I understood the purpose, this is my interpretation of Agile, right? It was a response to the fact. act that the software we're building is not what people want and it doesn't work so Yeah, it doesn't solve the actual problems that the people have. well that's why that's the first thing but the second thing when it doesn't work like it actually doesn't function like it doesn't compile it doesn't work right and so okay let's do this thing that's very iterative where we show it to people frequently to see if it's actually what they want
26:45 Scott and we deploy it frequently at the end of every sprint so we can verify that it works right Right. That's fair. and Yeah. so to me that's like that was a response to whatever was happening before where we were building stuff that didn't work and that wasn't what people wanted and Well, I remember those days, right? Like early on before, I mean, in my career, it was like, you know, like building Windows software. or something like that right where you have like you know nightly builds you know and then the build comes out in the night you know the day you know the morning you get your build and and like literally you could have developers who broke the build like they somehow checked in code that like broke the build Yeah. and you you get the little dunce camp cap that you walk around and give to the developer that broke the build you know the build buster you
27:33 Scott Build know and Buster. Yeah. I experienced both dunce caps and build buster stickers okay Yeah. so um i hear what the adjunct with a critic you know someone saying well it doesn't you have to do this upfront stuff but like i guess what i'm trying to say is You can get that iteration loop really Yeah. fast. Yeah, really tight. You know, not a week or two weeks, but like, okay, let's talk for 45 minutes. Let this thing do its work and come back in 45 minutes and see what it's like. Or Yeah, you can get totally. real, and if you combine that with some of the stuff you talk about, of course, which is ephemeral, like review type environments and continuous deployment and all that stuff, like you can, I would imagine the right team,
28:20 Scott the right environment, the right product. I mean, you can. like a blazing fast and Yeah, so basically we have another violent agreement here. We're not going to argue on this one. and blazing fast is actually like what you've talked about with one of your clients who has gotten to the end of their product roadmap and it's Right. like what do you what else do you want us to make for you like Right. and actually it gets back to the thing we're talking about it's just shifting the bottleneck upstream which is like well we need more product managers and researchers and designers to go out there and figure figure out what it is that people want. And I would argue that that's the piece of the case that you're making for like the pair, the pair programming, if it's not pair programming, it's like the pair agile thing, you know, where it's like, make your product owner and your developer like right there, make them a unit that they they they're collaborating together and and as soon as the the implementation is done,
29:18 Scott you're ready to move on to the next thing. And this is where I do think that what you're going to see in like smaller and smaller, like I just saw, okay, just today I saw something on LinkedIn, it was like, you know, a recruiter saying like, oh, this founder is trying to hire a CTO and they're demanding that the CTO needs to still be coding. Mm-hmm. And there were, you know, this big, huge debate about that. And the thing that was interesting to me about that is that I think that. ai is doing is going to do that same type of thing where it's basically you know you've seen these other things we're not going to have managers anymore everybody needs to be a doer right and so this idea that like you should be able to be builders like you should everybody you know like everybody on your team you make make put your systems in place so that everybody on the team really can be builders it's going to cause you to basically
30:09 Scott be able to build really a lot of things really quickly and so now Okay. it's like the the question then turns into like do we know what we're building and why and what's our plan for that I think there's this thing that's definitely out there. People have this idea. I've heard the expression, the alpha is going to flow to the builders. Do you know what I mean by that? The alpha? Yeah, yeah, You well, we talked talked, about yeah, like, where's the value going to go? Right. And so I think I've heard people say this and they think the value is going to go to the people who can build. But I have a different point of view, which is the value is going to go to the people who can. who can figure out the right thing to build. Yeah, yeah, I agree with you on that. I think it's really exciting for a lot of people who never knew how to, who never were capable of building before,
30:56 Scott because they either by, as a result of talent or training, they didn't have software programming skills. Now it's really exciting and they say, oh, I'm going to be able to build and do my own stuff. I can create all sorts of value, but I think they're really missing it. That's definitely out there, and I think they're really missing it. Well, again, like we're going to be in violent agreement on this. It's going to probably be hard for us to get a debate going, We but I have to find something we disagree on. think that like I think that the. challenge here is as you've said do you do you know how to figure out what is worthwhile to build like that's Yeah. that's kind of some people talk about that as the taste you know do you have good taste um and then you and i've also said and i i know that
31:43 Scott I feel very strongly about this, that the tooling that's out there today is not really good enough yet to make sure all the engineering eyes are dotted and T's are crossed for Mm-hmm. security performance, you know, that stuff. And so you still need a fair amount of expertise and wisdom applied or at the very minimum, you know, as you know, I'm a strong advocate of. of build the scaffolding around the vibe coders, build all the engineering best practices scaffolding around your builders, and then you can start to trust that what they build will not bring the whole site down. That's right. And yeah,
32:29 Scott you know what? Okay. Can I tell you what I think the final frontier is going to be? Software. getting a lot cheaper let's imagine we can figure out the product management problem and we can have a crystal ball and figure out what needs to be built i think that's a problem that's not solved i personally experienced that problem i'd love to do you know a little side project to have something going on i don't have i've done that before in my career i don't have anything going on right now and um that's gonna be fun i don't have an idea right now i don't even know how to get an idea anymore I know it's not a shoot down my own ideas, Sure. but let's say I had one. I think the final frontier is what I will call distribution. Let's say I had a crystal ball and I found the thing people wanted.
33:16 Scott How do I get that thing in front of those people efficiently, cost efficiently? Because the tension is so divided right now, whether it's. advertising social media whatever like how do you break through and get in front of the people that you need to get to I just I guess I guess I'm just thinking out loud here I guess the answer is you just do those things right you buy advertising you do social media you sponsor social media you do you do PR you go you do community you just do all of it right Yeah, I think one of the thesis I have one of the theories I've heard out there is that like creators who already trusted brands like this is like an argument that the brand is what matters the brands are.
34:01 Scott are you know brands can bring out any product and as long mm as they -hmm yeah are consistent with their brand ethos and mission then they're the ones that are going to be trusted by the consumers to to buy that's right and that's just you know essentially influencer marketing is what you're doing creator influencer marketing and so i don't know it seems to me like that's the final frontier of all this like old industry which is like okay everyone's going to build it for free everyone's going to have ai that could go out and talk to customers and figure out what to build but like how how do you i guess we should just give up call Well, I don't... it a day just you do you remember professor falcon Uh, wait. The movie War Games with Oh, Matthew Broderick. yeah. Right.
34:46 Scott He The best just gave way to, the only way to win is to not play. exactly, exactly. And he just gave it up and he went out to his island retreat and was like, this is ridiculous. Well, Yeah. I'm not there. I'm still definitely very bullish about the idea that like, you know, I'll just repeat. I think that like the people that have ideas. I agree with you that distribution will be a piece of it, but the people that have ideas will be able to build those ideas. If they have good scaffolding around them, those ideas will be built in a way that is appropriate and, you know, secure, et cetera. Mm-hmm. And then, yeah, there's distribution. And I think that, you know, one of the theories that's out there is that.
35:33 Scott You know, there are a whole bunch of smaller niche needs that were not practical to build before, but now they're practical. And so if you can reach the people that have that need, then those are the people that you'll be the one solving it for them. And there's there's a viable viable way of doing that. That's what I mean by the final frontier. Okay, so like, let's say you need some app for some very specific purpose. I'm going to give you an example of that specific purpose. Yes. I snore. Oh yeah. It's a fact. I have to admit it. I've never heard of it. I never heard it. People tell me that People tell me it's funny. You never really hear your own snoring, do you? there happens to actually be one person in particular that tells me about it quite frequently.
36:18 Scott Let me guess. And it has become an issue actually between us. Not a major. Well, I was going to say not a major issue. you but depends on who you are in this equation so my wife treasures her sleep she's a person who sleep and i've turned over the last five years into a real snorer i don't know how it happened and um sometimes i wake her up and sometimes i keep her up and um that really bothers her and then it just you know leads to all sorts of dysfunction i'm like oh i had a great night's sleep and she's like i Okay. want to i want to kill you right now right okay now I got an app. It's a phone app. Obviously, Mm it's -hmm. a phone app. And you just put it next to your bed and it Scores your snoring. that's exactly right.
37:04 Scott It listens overnight and it scores you. Okay, now my wife is a little bit of a snorer. she's apparently i'm just like really saw and logs in there and she's more just like a little little stars and so i was like well i wonder if it picked up her snoring too and so you want to if you had a particularly bad snoring night you want to try to blame her well that was your snoring not mine not not not especially but you do need to disambiguate that so i thought well there could be an app where we each put our phones on the bedside and if they're both running they can use some sort of i don't know sound you could probably figure this out you're an online you're a digital audio they could use some sort of sound differential to like figure out who's snoring with who's the Yeah, for sure. only point i'm trying to so i that was literally i got the app i got the snoring app last night i used it for the first time last night i scored 55 out of 100 and i'm not sure if i'm trying to get more or less points i have to go back and look up but
38:02 Scott and I think the guy thought you of this want problem less this one and I thought and I thought of this app idea that I'm just described to you which is we could have two of these but the point I'm getting at was what you just said if you can find a problem and reach those people then you know you can have a suffer product and reaching those people you have to reach the double snore what do those ads look like do you and your partner snore might it have also been the dog do i need a third app in our room for the dog who also sometimes does little cute little dog snores Yeah. This is getting over the top now. man So now my dog is on Apple Watch? it's funny you say this i'm pretty sure that like uh you know in the there's also a fair amount of of like um i watch like streaming news you know so i have like sling and so i'm watching the news is like on my tv via sling and i've noticed that i get targeted ads that are very clearly targeted like they are 100 ip targeted to me based
39:03 Scott Based on Yeah. my, you know, internet searches, you know, and it just cracks me up how far that's come. I feel, of course, like partly to blame for that as helping invent behavioral ad targeting back at the turn of the millennium. But you I'll tell you what was very, very visceral about that for me. The day I turned 50, my get ads changed. your AARP letter in the mail. Well, there was that but like is instantly online the ads I would get online were different You know your 40 to 49 set when the person's going in there and clicking 40 advertised to 50 plus Suddenly I was getting different ads. There you go. I mean those are overnight This is a lot.
39:49 Scott This is a pretty long and windy road. How did we get from token maxing to AARP and cruise lines and erectile dysfunction advertising?
39:59 Brad We're
40:00 Scott Interesting.
40:00 Brad talking about the end times.
40:03 Scott All right. Well, here's my question to you. Here's
40:05 Brad Okay.
40:05 Scott my question to you. Okay. My question is, I want to talk a little bit about are we going to get like are people going to be running away from these, you know, AI frontier model labs because of either pricing or. The other thing is wanting local inference so they can have privacy and they can be not controlled by. not be controlled by the man in other words and i don't mean like controlled by the man like you know guardrails around bioweapons or something like that like of course you know i think i'm obviously a supporter of those kinds of guardrails but um i mean like you know i was really set back by the fable basically saying like oh we're gonna like change we went from zero data retention to now like we're gonna keep 30 days and uh they also were like in the back end they were running a classifier that basically just
40:58 Scott He decided if your query was allowed to use Fable or not. And if it wasn't allowed to use Fable, at first when they wrote it out, they were silently dropping you down to Opus 4.8 without telling you, but still charging you the Fable rates for it. Like there's some super shenanigans going on there. And I'm like, this is like. But to me, that was the that was the the Klaxon blaring of like, hold on a second, like we're way we're way allowing, you know, we're giving up way too much agency to these. you know soon to be monopolies or duopolies like these these players are too big like this is power to the people let's open source open weight this this bad boy yeah
41:45 Brad So the answer, your question is, do I think we're going to, what is your question? Do we just think we're going to see more of the open source usage or are we going to?
41:52 Scott are we going to see like you know People fleeing the big providers to go to run it themselves, to essentially democratize this.
42:04 Brad I don't know if I would use fleeing. Here's my take, having thought about this for about 90 seconds. I think we'll see three different types of pressure as CTOs to move away from the. main big models. One will be financial. That'll be CFOs saying, hey. got del those token back then i heard from my friend that you guys could run your token your little models you have over there whatever cute engineering stuff you guys do you guys could do this on your laptop and save us a lot of money so like i think that'll be one thing we hear and i think the other side of that coin is to me that's the same as the colo versus cloud hosted argument which is like yes it would be cheaper to do it on our own
42:51 Brad But we have to maintain all these servers. And so the total cost of ownership is more expensive. So I think we'll get pressure to use locally hosted models. Um, but people won't be accounting for the cost to maintain those things. That's one vector of pressure. The next pressure I think we'll see, we'll get some of the free from the man, right? I think that'll come more from the individual engineers. To me, that's more of a political acts, like people saying, I don't want to, I don't want to have my stuff owned by the man, but there will be, I think in some areas. privacy concerns but i don't think like just
43:30 Scott Well,
43:30 Brad for coding let
43:31 Scott let me challenge you a little bit on that one, then maybe we might actually be able to get into an argument here.
43:38 Brad me can i just throw out the third one we
43:39 Scott can Yeah,
43:40 Brad come back
43:40 Scott sure,
43:40 Brad to
43:40 Scott third
43:40 Brad genres
43:40 Scott one.
43:40 Brad for context the third one is what i would call you know the small language models
43:45 Scott Yeah,
43:45 Brad I think there will be vendors who come along and say, look, we've got a much more cost efficient model for coding because we've just tuned this thing just, oh, you're a Python shop. We've got a Python coding model here.
44:00 Scott yeah,
44:00 Brad Like you don't need to. These generalized models are so expensive because you can ask the same model, how do I tune my guitar and how do I write this software program?
44:08 Scott yeah.
44:09 Brad And, you know, who won the Olympics in 1948? right like you can ask this thing anything and so that you know someone's going to come along and say here is the python programmers model and
44:18 Scott Exactly. And it's
44:19 Brad it'll faster faster
44:20 Scott and cheaper and yeah.
44:22 Brad cheaper and
44:23 Scott And maybe and maybe even better performance, like not just speed performance, but may actually be better at if like if you really tuned it on Python, it may actually be better at Python, less likely to hallucinate Python.
44:36 Brad I think there'll be some pressure there, but that'll be challenging because then we'll have to go sell this to our CFOs and say we want to buy this special model and they'll say, well, what's wrong with Gemini?
44:44 Scott Right.
44:45 Brad And so, and then, so that's the third one, but then the one where we think is interesting has to do with the local for like privacy security reasons,
44:54 Scott So
44:54 Brad which is
44:54 Scott let me
44:54 Brad interesting.
44:55 Scott let me
44:55 Brad I want to hear your take on that.
44:56 Scott let me give you the scenario where your your CFO might come to you and ask you. this question hey are all of our intellectual is all our intellectual property going up to um anthropic i'm not or open ai i'm not sure i i'm not 100 sure i trust them to not use our intellectual property against us and
45:17 Brad Right. Right. Well, supposedly our enterprise agreement with them says it's not.
45:23 Scott And then with Fable, they changed that and said, we're going to keep it for 30 days and there's no opting out of this, even if you have a DPA with us. Now, their reasoning was around, you know, this thing is so good, it's cybersecurity, it's, you know, we want to we want to make sure nobody's using it for illicit purposes. OK, but that's a hold on a second now. you're who's watching the watchers man like
45:50 Brad Right.
45:51 Scott right like you're the one that gets to decide whether my query was a invalid use of cyber security you know like maybe i was just trying to vet that my software didn't have a security hole in it or maybe i'm on the red team and i'm doing i'm doing ethical hacking as a red team member trying to make sure that you My internal software is secure.
46:14 Brad I mean, the same could be said of the Facebook algorithm, right? Like, yes, yes, real, yes, yes. The answer is yes to your thing. I can see that I would see it more come from the legal department than the finance team. But like, yeah, like, that's a real concern. And it's already a concern I have because. The one thing I've tried to impress upon our team about, you know, we want a lot of experimentation with AI tools, but I've really tried to impress upon them like, hey, we buy this tool or whatever, it's too expensive, that's a two-way door. But if people start putting our data in here and we don't have the right agreements and this company isn't trustworthy or they leak it, that is a one-way door. We cannot get that back. And.
47:04 Brad I mean, it's interesting what you're telling me about Fable. I wasn't aware of that. But I can see for overall for that concern, like we just do not trust that we can keep our data safe. And then I can see it even getting pushed further. Like, so, you know, TaskRabbit has partnership agreements with vendors. So not vendors, I mean partners. So like, for instance, we do deliveries for Best Buy, right? You go into Best Buy and buy a big flat screen TV. and say i want to deliver it well really it's someone from task grabbing it coming to pick it up and bring it to your house and a lot of those partnership not a lot of them pretty much they all ask us to attest that we have some sort of cyber security and
47:41 Scott Yeah, they have DP you have DPAs with them just like you would for yeah, of course
47:45 Brad we you know they want us to see a SOC 2 or an ISO or whatever right and i could start to see a world where they're saying hey our data cannot go and anybody else's models now
47:58 Scott A hundred percent.
47:59 Brad that stuff's always at least in the partnership space it's always fuzzy because it's like whose data is this really yeah
48:05 Scott Well, and I think that like, I guess that's the other piece of this that I'm saying that I am predicting that there's going to be more backlash on this in this regard. I think that I
48:15 Brad i
48:15 Scott think that
48:16 Brad could see that
48:17 Scott There used to be a, hey, it's worth the risk because the frontier model, like I trust Anthropik, they're the good guys, or I trust OpenAI, right? You know, there's this notion of like, oh, the agreement says they're not training on our data as long as we have a subscription with them, right?
48:36 Brad Mm-hmm.
48:36 Scott That's what the agreement says to
48:38 Brad Mm
48:38 Scott people
48:38 Brad -hmm.
48:39 Scott starting to challenge that and say, you know what? Now that I can literally run an open weight model locally that I fully air gap, that I know is not ever sending anything outside of the machine it's running on, right?
48:52 Brad Mm-hmm.
48:53 Scott Like it absolutely is not sending it anywhere. It is only on premise, on that device. I think there's going to be more and more pushback of being like, well, why would you ever give that data away?
49:06 Brad Yeah, I think that's right. I think I hear what you're saying. I'm picking up what you're laying down, which is I always think about how what changes the dynamic of an industry, right? Like there are often certain players who they make that decision, like we don't want our data to be safe and then they're big enough gorillas in the industry, they can start forcing other. players to the ecosystem
49:30 Scott Like state
49:31 Brad around them to
49:31 Scott of
49:31 Brad change.
49:32 Scott Texas buying textbooks.
49:34 Brad I wasn't thinking of that. I'll tell you the example I was thinking of. And it's interesting, it didn't come true. And I was kind of surprised that it didn't come true. I'm not trying to get political here. I'm just using this as an example. Obamacare, the Affordable Care Act, the marketplaces for insurance. I was always puzzled why, so the way those things work is if you're a company of over 50 employees, you have to provide insurance. Otherwise, if you're under 50 employees, you have to go into the market, you'd send your people to the marketplace. And I never understood why it was that way. I didn't understand why the biggest companies in the country, you know, the JP Morgans and the Microsofts weren't just like, this healthcare stuff is a headache for us. We would
50:20 Brad How about we just give the money to the government or the employee or whatever, and they can just go in the marketplace and deal with it and we can just wash our hands? How many people do you think Microsoft employs to manage their health care benefit program,
50:31 Scott Right.
50:31 Brad right?
50:32 Scott Thousand at least probably.
50:33 Brad Exactly a thousand. I never quite understood why big businesses didn't actually push the Affordable Care Act to include them as opposed to exclude them, right? And then I read this article in the paper one day. back when we read papers, by the way, that Walgreens. had built a healthcare exchange. They hired the same company that built the, were one of the same companies that built the exchange for the government to build them their own private exchange internally. And they were shifting gears and going from providing health insurance to just giving their employees money and saying, go into the exchange and buy your own. And I thought to myself, oh, okay, that's going to be the tipping point. What's going to happen is that all these big companies are going to start doing that. And then after a little while, these big companies are going to say, wait a second, why are we running an exchange when the government runs an exchange too?
51:25 Brad Why don't we just persuade the government to let us just... put the people in that exchange. And I'm only telling this whole story, which I'm not being very long-winded to say, I'm just trying to think, I always try to think about the dynamics that will make something shift. And I thought something would shift with the Affordable Care Act, it never did. And so my point is, what's going to make it shift with these major frontier models? And I can see it. being customers who demand privacy are going to say you cannot share our data with these frontier models and that's going to necessitate me as a vendor to have to go out that that might be the tipping point for a lot of companies who are willing to take risk but ultimately they're going to get their hand forced to bring these things in house
52:07 Scott So there you go. I think it's going to be, like you said, a combination of eventually the CFO will say, hey. hey tokens bend too much and it will be some stakeholders whether it's your own legal department your own internal team or your extra your external customers are going to say hey i want guaranteed privacy of my data i want guaranteed protections of my data and then what was your third one oh
52:32 Brad My third one was the small language models.
52:34 Scott yeah that there'll just be that those there there are versions where The models perform better when they're actually tuned to exactly what you want as opposed
52:44 Brad Right.
52:45 Scott to just sort of tuned on or trained on the whole world.
52:47 Brad Here's an interesting thing about the middle one, the customers who want private data. How does that play with... the fact that everyone's on the cloud not everyone of course but like you know like so you come to me and you say okay i don't want you to you know you're you're my customer and you come to me and say i don't want you to use anthropic or gemini or or anybody you got to like run your own stuff i'm like okay but i'm going to run that inside of aws so is that going to be okay with you
53:15 Scott I would think that's more likely to be okay with you than the. you know having the model i i
53:22 Brad I
53:22 Scott exactly
53:22 Brad guess AWS isn't training itself on the
53:26 Scott you're
53:26 Brad content of the content
53:27 Scott just running
53:27 Brad of
53:27 Scott you're just running an ec2 instance you're just running a you know you're running a pod you don't aws doesn't know what's running on that pod you know
53:37 Brad Yeah,
53:37 Scott And
53:37 Brad that's true.
53:38 Scott I just think I really so I guess the other piece that I'm piling in on this is a combination of like AI backlash and like, you know, the big frontier models have not done a good job of like. marketing themselves as being friendly actors you know they like when the CEO is up there saying oh it's going to cause job loss you know and oh it's this massive security problem it's just like i
54:04 Brad Anthropic's done the best, but it's tough.
54:07 Scott i think I feel like it's funny. I thought Anthropic had done the best and they seem to be sort of like the good guys, but I definitely feel like they've fumbled this whole fable thing in a way that has led me to not to start to question them. The other thing I'm a little worried about or I'm feeling, I guess, not really worried about is I feel like they're kind of in Boy Who Cried Wolf mode where they've been some of the loudest. voices for like um this stuff is really dangerous right you know i don't know what's the line me thinks the lady doth protest too much give
54:45 Brad That's one line from the bard, but there's the other line from Jurassic Park.
54:50 Scott me that line
54:52 Brad We were so busy wondering if we could, we didn't think about if we should. And so that's what I worry about the most. is that people are so fired up about doing this stuff that we're letting genies out of a bottle that will never get back in.
55:11 Scott Yeah. Well, and I guess that's the thing that I would say, like, I don't buy that I'm not feeling that any of the frontier models are any better actors in that regard than than any others. Right. They're all they're basically all. acting in the same way.
55:29 Brad They're just capitalists.
55:31 Scott There you go. So
55:32 Brad Or
55:32 Scott that's my so.
55:33 Brad Ouroboros.
55:35 Scott Yeah. Anyway, so but that's still my summary. My summary is, look, it's making my team so effective that it's not going anywhere. And so therefore, don't allow yourself to be just completely tied into a single vendor. And
55:49 Brad That's
55:49 Scott and the good news is the good news is that it looks like the open source open weight models are getting good enough. Like a year ago, you and I were talking. And we're talking about like, hey, is this stuff good enough to like write production code?
56:01 Brad That's Yes.
56:02 Scott like a that's like a done question. Like, of course, it's good enough to write production code. Right. And so now I'm saying, like, guess what? The open source models, the open weight models are going to be good enough to write production code. And
56:16 Brad Yeah.
56:16 Scott so the game over.
56:18 Brad Yeah, I gave a talk to my team today. I mean, I shouldn't say I gave a talk, but we had a meeting and I was talking about AI. And I said, I feel like there are three truths we're adapting to right now. The first is that it's. changing our jobs and changing our industry. The second is it's not going away. And the third is it does, in fact, write production-grade software. And those are things that a year ago or 18 months ago were not true. And I think it is, at least they were not true to me.
56:50 Scott Or they
56:50 Brad But
56:50 Scott weren't universally accepted as true.
56:53 Brad I think it is true. And I left my team with this. Maybe this is a fun place to stop. I have this picture of myself with a floppy disk. A floppy disk, for those of you who don't know what that is, three
57:06 Scott Wait, five and a quarter inch or seven and a half inch?
57:10 Brad and a half. I never possessed a seven and a half. I did possess many, many five and quarters from my Oregon trails and my Olympic games and all that kind of stuff. Never possessed a seven and a half. I knew a guy, though, they did. But I have this picture of myself with a 3.5 inch floppy disk. Somebody on my team said, why is Scott holding a picture of the save icon?
57:38 Scott Ow!
57:39 Brad uh they were just giving me a hard time of netscape navigator 1.0 oh
57:45 Scott Nice. Fit on a single floppy disk?
57:47 Brad it probably was like five of them to be honest now let's take a look i actually actually happen to have that up here let's take a look it's right on my computer somewhere um i think it's just one it says netscape navigator 1.0 for windows installation and for installation instructions read the readme file file included on this disk so um the point that i was making to my team today is like i get it all this is kind of threatening and scary i know people at both ends of the spectrum up here who after 25 or 30 years in the industry gave it up just said i don't think i'm ready to go learn all this new stuff and a young man who's a research college graduate who wanted to be a computer programmer and said i don't think i want to do it i don't know if this is a i don't think this can be a career for me i don't know if it'll really be here for a good solid 30 40 years and so i get the concern
58:47 Brad But the reason I bring up the Netscape Navigator disc is I said to them, I feel like a huge asset in my career. I feel incredibly fortunate in my own career that just by the fact I was here first and when I say here first, I worked on it started working with the Internet professionally in 1995.
59:04 Scott Yeah.
59:04 Brad I mean, Netscape Navigator came out in 1994. Like so. And just because I was the first one on the bus, I got to pick what seat I wanted to be in, you
59:12 Scott Yeah,
59:13 Brad know, and
59:13 Scott exactly,
59:14 Brad we
59:14 Scott exactly.
59:14 Brad are all at the beginning of this AI thing. And I know that to the people on my team, I know you're all nervous about this. I'm nervous about it. Don't get me wrong, but we're here at the beginning. We get to shape this and lead it and figure out where it's going to go. And that is a massive gift for all of us in the next. chapter of the software development story.
59:38 Scott Love it. That's the way to land it. That's the way to land the plane, Scott. You just did it.
59:43 Brad All right, Brad, what do we want these people to do?
59:46 Scott Smash that subscribe button, rate and review.
59:49 Brad And tell all your friends about it, including your mom.
59:52 Scott Including your mom. Exactly.
59:53 Brad All right.
59:54 Scott All
59:54 Brad See
59:54 Scott right. Have
59:55 Brad you next
59:55 Scott a good week, week.
59:55 Brad Brad.
59:55 Scott See you next week.
59:56 Brad Bye.
59:56 Scott This episode of the How Many CTOs Does It Take podcast was...
59:59 Brad Produced and edited by Saffron HeptaCow. Our theme song was AI generated by Brad. Please follow us on all the social media platforms at HowManyCTOsPod. You can find us on YouTube, LinkedIn, Instagram, Facebook, Threads, X, TikTok and Patreon or at our website at HowManyCTOsPod.com and remember to rate, review and subscribe on the podcast platform of your choice. I slide, thanks
1:00:26 Scott And I'm for the
1:00:26 Brad listening.
1:00:26 Scott sea of the day.
