- Speaker #0
Because to me, a lot of AI simply doesn't work. It just full stop doesn't work. But then sometimes it does, and it's magical. And if someone gets there and you don't, yeah, that's really, really bad. You know, that's not good.
- Speaker #1
AI is coming for Thass. Will you survive it? Will I? I don't know. Let's find out. Welcome to this new Indie Board Session episode with my guest, Jason Cohen. Jason, welcome to the show. Thank you for being here. So I've been a reader of your blog for almost a decade, and you publish a blog called A Smart Bear, but you have founded multiple very successful companies. So can you tell us a little bit about your background?
- Speaker #0
Sure, the latest company is called WP Engine, and now it's about 16, 17 years old. We started bootstrapping, but then raised money. First a small amount, and then a lot. But then we were also growing a lot. We always raised an amount of money that was proportional to how big we were. So it was a lot, but it was the correct amount. And today we have 200,000 customers and 1,200 employees, and we're profitable and growing, which is kind of amazing still, I guess nowadays. Um, Then before that, SmartBear, which is why I'm known as A Smart Bear online, where the blog comes from, that was fully bootstrapped. I sold that in 2007 and left in 2009.
- Speaker #1
Okay. And SmartBear was a software bootstrap company for like a dev tool product, right?
- Speaker #0
It was for software developers to review each other's work.
- Speaker #1
Yeah.
- Speaker #0
Nowadays, people do that all the time with GitHub. Yeah. It's just a normal thing. Everyone does it. It's free. But when I started it in 2002, very few people did it, and there were no tools, there was nothing free. So it was a big revelation to say, oh, look, we can integrate with version control. All these things you now take for granted didn't exist, so we invented that.
- Speaker #1
Cool. And so the good thing is that you've seen the bootstrap world, you've seen the venture world, and on your blog you mostly talk for bootstrap founders, actually. Recently, I did an episode of a podcast with Robert from a company called Perspective, like an eight-figure SaaS business. And he told something that I was thinking walking to the studio today. And when he told me, today, if you are in a SaaS business and your company is not in chaos, you might not make it, you might not survive. Because you need to kind of... Make sure you can be safer, at least get a chance on the other side. You have the privilege to have been involved with, I mean, obviously very large companies like SmartBear is still a decent sized company today, WP Engines too. What's your thought on that? Do you feel that people need to scramble and get...
- Speaker #0
and get moving or like what what's your thought on the current times well i wouldn't use those words but i understand the idea is that you should have urgency in this new technology which certainly is changing how we do every part of the business so urgency yes experimentation yes chaos sounds like you it sounds like a bad thing to create for a company um also i think While in the long run, it's obvious that AI will transform how we work, just like email transformed how we communicate and the internet transformed that. So in the long run, obviously so. In the short run, it's not so clear to me how urgent and how chaotic it needs to be. What I see is every time I log into Cloud Code, which is every day, it says there's a new update. And there is. And they've done stuff. Sometimes it's minor things to the UX or the CLI. But sometimes it's something enormous and major, something that may not even exist in a month. Or it may be the next new thing. Nobody knows. So in a world where every single day, and even just one piece of software, of course, an important and large piece of software, but still, you could just wait a month and it'll be pretty different. So how chaotic do you need to make it today, given that it's just going to be different in just a month or a day? So I think that's a bad attitude. And I think that people are like, if you're not doing everything now in AI, then you're going to die. Well, actually, the way we're doing AI now is nothing like what we did six months ago, which is not like six months before that, which is not like a year before that. I don't see the pace slowing down in AI. If anything, it's accelerating. So that means we don't know what AI is going to be like in six or 12 months. So what are you being all chaotic about? So I think if you're, I know I'm latching onto those words, but I do think people are causing chaos by overreacting. To me, the correct, to me, of course, I understand. Well, if someone says, I don't need to think about AI at all, that's a person that needs to be shaken up. And I understand if you have to swing the pendulum too far the other way. to just get them to a middle ground. I get that. So if that's the rhetoric that's needed, fine. But if we're going to talk about what's actually a good idea, then let's set the extreme rhetoric aside. Obviously, you can't ignore AI because then everyone who doesn't ignore it will be moving five times faster than you. And also creating features that you don't have that customers actually want. Maybe not today because to me, a lot of AI simply doesn't work. It just full stop doesn't work. Most of the tools I have that say it's AI. And then it doesn't really work. I don't use it. So I don't believe you that like, oh, we'll just put in AI and it'll work because usually it doesn't. But then sometimes it does, it does. And it's magical. And if someone gets there and you don't, yeah, that's really, really bad. You know, that's not good. Look at Anthropic right now. They have now eclipsed OpenAI and they're growing faster because of cloud code. Did anyone predict that Anthropic would be bigger than OpenAI in terms of revenue or users even six months ago? I don't, I mean, okay, a few people, I told you, okay, fine. Most of us didn't think that, right? So, so yes, sometimes, sometimes. And so to me, you should be getting into AI, not because this is how it's going to be from now on, it's not, but because if you don't start now, then when and how far behind will you be not in one tool or one technique, but just in the whole new mindset of what it is to think about. prompting instead of doing and then skills instead of prompting and then agents instead of just skills and then agent collaboration coordination instead of just and whatever like these concepts are a little bit more stable like the notion of a skill is slightly more stable and again the details of a particular agent collaboration are not important they will those those are changing but if you're if like you still think that what ai is is what happens when you go to open ai and just chat with it. Then you are really, really behind. Indeed. I wouldn't say chaos. I would say you're missing the boat. And what you're doing is seeing that chat and going, AI is pretty dumb. And you're right. The LLMs are pretty dumb in the format of the non-contextual one-shot chat. I agree. It's sometimes magical, but largely dumb and not good. And if that's what you think AI is still, you are in fact missing the train. And that is bad. Let's not cause chaos though. Let's decide to pick, you name it. Cloud code, I think right now is a good choice. But you could pick it pretty much anything else as long as it's structured and has skills and you're going to iterate and do that stuff. Even now, you can do at least five or ten times better. And by better, I mean accurate but also productivity than in chat. That's probably where you should be with urgency but not chaos. Because anything that was true 10 or 20 years ago and still is even now, that's a deep truth that's really worth knowing. even if the way we... approach it, attack it, use it, might be very different with AI. Maybe it's not, but let's suppose it is. Okay, but it's still a deep truth we want to attack just using the tool of AI. Well, okay, that's good to know. Other ideas might just be irrelevant. Okay, well, that was a good idea at the time. Maybe, maybe not even that, but now who cares? But things that are still true, but we would do it differently, now that's a deep truth that AI doesn't displace that truth.
- Speaker #1
Yeah, I mean, and so I think that it's a... What you're saying intersects well with one question that I wanted to ask you, is that in the last few years, you've denounced a little bit the growth at all costs mantra in the SaaS world. And when you look at it today, it hits for many SaaS founders a wall. And this reminded me of your article about Max MRR. And asking myself, is the problem the fact that now they've reached their max MRR and that's why the growth has stalled? Or is it because something else has happened? So maybe before we dive into that question, can you remind the audience what the max MRR concept is?
- Speaker #0
But first of all, when something happens to everyone across the world simultaneously in every industry, That's not because everyone just happened to hit their own little metric that they computed. That's because there's a secular shift in the world. Obviously, that's what AI is. So I think there shouldn't be any doubt why everyone's growth is slowing right now. So that should be obvious. However, what you're talking about is also very true. This is, again, something that's true anyway, despite AI. And it's not that some founders hit it. 100% hit it because it's just math. And the issue is that in a recurring revenue business, what happens is marketing only grows as fast as you can find marketing channels to spend in. And even then, the channels top out really fast. Like you unlock AdWords. There's only so many searches, only so many words, only so many things with ROIs that work. And yes, of course, you're going to optimize and eke out a little bit more. But there is a pretty clear ceiling on what you can get out of there. And then you're going to get X sales per month or whatever, right? And that's what that channel can do. Okay, plus or minus a little bit, but that's it. And you can say that about every channel. And most channels won't work for you. Companies don't usually say, the way we succeeded is we had 20 channels, each trickling in some stuff. That's not how it works. Only a few will really work for any company. And so what happens is you... get those. And so marketing is chunking in this certain amount every month. And marketing does not care how big your company is. If you have a thousand customers and let's say AdWords is tossing in a hundred customers a month. Okay. Once you get 2000 customers, AdWords is still chunking into a hundred customers a month. And now you have 3000 customers. AdWords is still chunking in a hundred customers a month. AdWords doesn't care how big you are. Right. However, cancellation is not like that. We talk about cancellation in percentages. 5% per month cancel, we say. Why do we say that? Well, obviously, like there's some percentage of our customer base where now's the day. Now, I'm simplifying because really you would want to do cohort analysis. But let's not get into all that. That's what you should do. But let's just keep it simple. So at 1,000 customers, 5% canceling per month is 50 customers a month leave. So AdWords puts in 100, 50 leave. You're still growing by 100 customers. Good for you. But what happens to 2,000 customers if the size of your business is 2,000? AdWords still chunks in 100. Hooray. What's 5% of 2000?
- Speaker #1
100.
- Speaker #0
Oh, so you pull in 100 and then 100 leave. You're not growing. You've stopped growing. So I call it the growth ceiling. It happens to everyone because it's just math. Marketing does the one thing. Cancellation is a percentage. In fact, it's exponential. It's as big as you. That's what exponential means. So how would you get above 2000? Well, here's the answer. If nothing changes, if it's just the status quo, You will never, ever have more than 2,000 customers. I'm obviously rounding off on everything, right?
- Speaker #1
Sure.
- Speaker #0
Ever. Now, wait, that's a big thing to, like, wait, hold on. The status quo, because I thought marketing's chunking stuff in, so if I don't do anything, I'll just grow slowly, right? No. You'll grow slower and slower as cancellations eat into what's going on until cancellations win.
- Speaker #1
Yeah.
- Speaker #0
You stop growing in this little example at 2,000. So you can compute what this number is. What is the number of customers that if nothing changes, if marketing is doing marketing things like it is today, and of course something can change, that's the point. But like, all right, if nothing changes in marketing and nothing changes in my cancellation rate, when will I stop growing? What is the maximum I could ever be? In this case, okay, well, if AdWords is a hundred and cancellation is 5%, somehow the answer is 2000. And the equation is actually really simple. I have an article on this, but it's easy. It's just this new customer's per month. divided by the cancellation rate. So it's 100 divided by 0.05 in this little example. That's 2,000. So of course, everyone can and should do this with their own company and even chart it over time because every month you're new and cancel's rate could be different. So of course, it'll bounce around if you're smaller. If you're larger, maybe it's a little more stable. You should probably do it in cohorts or segments. Okay, fine. But anyway, what I love about this is, first of all, it's very visceral. Oh, crap, this is it. It can't be better than this. That's a very strong feeling. The other thing I like about it is it circumvents the usual debate. I say 5% is not good enough. You're like, yes, it is because my industry and customers and you don't understand. Then I'm like, yeah, but I know lots of companies who also sell to consumers and their cancellation is 2% per month. So you're wrong. Your cancellation rate is just high. Or maybe you've picked a market segment that's crappy. And so you are capped because that's what you did. You're in control of that, aren't you? Arguing, is 5% okay? Is 4.5% okay? What percent is okay? Instead of arguing about that, which is so abstract, it's much better to say, you're never going to have more than 2,000 customers. You good with that? Would you like to move the cancellation rate down a little so that you can dramatically increase that number? Probably. Because also cancellations means they don't want it. Doesn't that bother you? That people looked at all your stuff and decided they didn't want it? It would bother me. It does bother me. So like, there's an emotional component, don't you... Want to find out why they saw all of your stuff and believed in your promise and then left anyway? Don't you want to know why? And then also financially, you need to know why.
- Speaker #1
Of course. One of your articles that you wrote two years ago, you mentioned that the AI wave was different than previous wave, than the mobile wave, internet era wave. Because... In some way, I could summarize that by saying that the big companies are not asleep at the wheel.
- Speaker #0
Right, exactly.
- Speaker #1
I mean, obviously, a lot has changed in two years with AI. I mean, you mentioned even in three months, like, who would have bet that cloud or an open AI, and Anthropic would be much bigger, growing much faster than open AI. What do you think, what do you take of what you said two years ago? And how would you nuance that today?
- Speaker #0
Well, I think most of it is still correct, but not all. And I'll tell you, the part that big companies aren't asleep at the wheel is true. Now, that doesn't mean they're successful at making AI that works. They're not. But who's spending a ton and who's at the forefront of AI? It's people like HubSpot and Salesforce. Google's trying. I mean, these are the incumbents doing AI. So that part is true. Microsoft. So they are not asleep at the wheel. Yeah. Microsoft did miss the internet age. Yeah. Right. And they miss mobile too. They're, they're not missing AI. Now, how much are they succeeding and what about the money? Okay. But just, it's different in the way that I said in that article. Um, does that mean you can't be a startup? Of course you can. Of course you can. But the, but the rules, but the way it works is different than usual in the mobile revolution. If you were early to mobile as a startup, Microsoft wasn't competing with you. And if they tried, it was really bad and they weren't, you know, and so you... there was a certain kind of way that you could try to get enough market share to exist. Nowadays, it's quite different. With AI, that is not the case. So there are breakout startups, obviously, but it's not true that you can just discount the big guys. That is not the case. You have to fight. If you're making a web chat client, you do have to fight against Intercom and HubSpot and Salesforce. Yeah, those are your actual competitors, and some are better than others. You know, yours isn't that good either, actually, because AI doesn't work normally, even for the startups. You know, I'm in some forums, and I would say every week someone complains, I'm using WhisperFlow, but it suddenly didn't work anymore. And other people say, it works fine for me. Well, I use AquaVoice. Well, that worked great for me. Yeah, but for me, it stopped working the other day. In other words, this stuff is amazing, and also none of it's really working well yet. And this conversation happens all the time. These are startups that are doing great work. I use them, too, by the way. I'm a paying customer of these. So, you know, I'm a fan. I'm just saying, also, it doesn't work a lot. That's where we're at. So the big companies are having trouble, but the startups aren't doing great in terms of the products really working well. So here we are. So that part of the article I think is very true. One thing I think isn't true is I emphasize in the article that if you have data, that's an advantage because you can train things on it. And so that's an incumbent advantage because the incumbents are the ones with data. They've got all this customer data and product usage data and I don't know, whatever other data that you might have. And a new startup has no data. So how does a new startup train on stuff? But that turned out to be, I think, fairly wrong. Maybe there are certain areas where it's right and certain. But in the main, that turns out to be wrong. Because I think one person with an amazing set of LLM skills, I mean literal skills, you know, .md, right? They can just do amazing things. They don't need data. So I feel like, sure, there's probably some applications where you do need data. Okay, all right. Then it's maybe quote-unquote correct for that. But that's not what I said. I didn't qualify it like that. So I was wrong about that. And you don't really need data for most things.
- Speaker #1
So you think that even bootstrap founders have a shot today the same way they had when you got started for your previous companies?
- Speaker #0
Well, I don't have to think it. You can just look at bootstrap companies who are being successful.
- Speaker #1
Yeah.
- Speaker #0
Most aren't, but most VC-funded customers also are not being successful, and most of the AI companies will fail, for sure. Of course, there's too much money going into AI. Of course, it can't all work out. We all know all this money is coming in, and the industry can't be that big. It's a net loss. It's net negative EV. And a company is always net negative EV. Every founder that starts a company... is doing something that's a net negative EV because the company probably won't work. All of us doing companies, including me, we're doing something where you cannot take things like probability and expected value and apply it because then the answer is never start a company. And that's not the answer. It's, well, only if dot, You feel compelled to. You maybe have an idea. You just need to. This is who you are. You just have to try. These are good reasons because actually the rational reasons are you shouldn't do it. So, of course, most of the bootstrap companies will fail and the VC companies will fail and most of all of it will fail. Yeah, that's true, but it's kind of not that useful. I guess what I do find it useful, what I do is I say, ah, why is this failure the usual state? And to me, the answer is because so many things have to go right or right enough, not like super right, but right enough for the whole venture to succeed. You have to be able, the product has to work well enough. You have to be able to build it and maintain it. You have to find customers. It has to be cheap enough. Enough of them have to come in and actually convert and keep paying and actually find value in it. with the competition. And like, there's so many things that have to go right enough. And so if you did assign probabilities, which again, I don't believe that I don't believe they're private, but if you did, even if you gave a 70% chance to everyone that you can do it, which is too high, right? But if you did, once you multiply that by that many things, it's still probably not going to happen. Right? So, so the reason why that mental model, while not literally true, although maybe somewhat true, it's like, it's close enough to true that it helps you think through like Well, what aspects of my business are fairly, or would that number be fairly good? Like I'm an engineer. I can probably make it work. Okay, fine. With AI, I'm not so sure, but okay, let's give it to you anyway. What are the areas that for you are weak or unknown just for you? And so those, so you'll probably die there. That's probably why you're going to fail. Could you do something about that? Maybe you have to learn something. Maybe you need a co-founder with a complimentary set of skills that together you're covering this a little bit better. Maybe you could test something earlier. to find out so that you're not wasting so much time before you realize it's actually a bad idea. Spending two months to find out it's a bad idea is better than spending 12 months to find out it's a bad idea. So there's things to do. Once you, once you see that there's things to act on to do something about that. But just in the, in the main, yeah, it probably won't work, but like that's so what? So either go get a job at Dell and don't do this or dig into why is that? And what's, what's, what's peculiar about me and my situation. What could I do about that? That's the only interesting question to me.
- Speaker #1
And so to double-click on this, you mentioned one of the themes or the concepts that you outlined in your blog is the concept of an uncopyable, like you want something that people can't copy.
- Speaker #0
Or won't.
- Speaker #1
Or won't copy. I was thinking about it, and WP Engine is a hosting business. So it's a very competitive market. And even SmartBear, when you sold it, GitHub had already started.
- Speaker #0
Is there a big market with lots of money that is not competitive? No, no, sure. No, no, but that's a serious question. Because if not, then it's like, right, that's what a big market's like, and there's a lot of money. And so the question, as always, is, so how are you different? Are you different enough that people care and want to pay for that? But that's always the question everywhere.
- Speaker #1
It's true. But if you think about, for example, with AI, where people can build so quickly, where skills and these MD files can change a UI or UX in seconds or minutes versus before, it took much longer. How do you think about building an uncopyable business?
- Speaker #0
Exactly the same. It's always been possible to copy any feature in any UI. Always been.
- Speaker #1
Yeah.
- Speaker #0
It's just easier and cheaper now. And I don't want to say just, that's not fair. It's so much easier and cheaper now that that's really, really true. But it's always been true. What software hasn't been copied by some other, what successful software doesn't have clones before AI? They all did. So does AI accelerate that? Absolutely. But if you're talking strategically, like the success of the whole company on a multi-year timeframe, then the fact that AI can do something faster is kind of not that relevant. You can go faster, so can they. Okay, that's the same, right? You also took six months instead of one month to build this. And so someone else takes one month instead of six months to build it. Right, but you're both on the new timeline. So if you're both accelerated... then the same rules apply. Now, I do agree that there is something about absolute time that's different. If it took me six months and it takes you three months versus it took me three months and took you two weeks, I agree that that's different. I do agree that that's different. But in the broad sense of markets are busy, markets are full, the markets where there's no money being spent are bad markets. where it's hard to do anything. And markets with lots of money being spent is busy. It's busy in terms of many competitors, and also they're all doing marketing. And I would agree that with AI, everything's faster, there's more marketing, it's more noisy. There's also an interesting thing of people will put out so much marketing, even if it's just mediocre. It's just that much harder to get seen. Even if that company goes out of business, there's always companies coming in and going out of business. But meanwhile, they're flooding the market with marketing that you have to compete with anyway. So I do agree that it's harder because there's more noise. It's also easier because you can build faster too. So the fact that you have to be unique in some way, I say unique, I don't even believe that, special in some way, that other people care about enough to pay for is still true. There's just nothing wrong with that statement. How is that going to come about now? Yeah, it maybe used to be that some kind of special code would work. Although, again, it really wasn't. Like thinking about, I've been doing this for 25 years. I don't remember a single moment where anything I did was uncopyable from a code perspective. I didn't do anything ever that you couldn't copy.
- Speaker #1
And so, for example, for WP Engine, the core insight was obviously not the technical. I mean, you had a strong technology foundation. But in a way, I guess the core insight was...
- Speaker #0
the customer service that you provided to these customers or so wp engine is a great example where it's not one unique thing yeah if you're seeking one unique thing that no one did you won't find it sure and and like it's nice when you have one unique thing but um that's more platitudes said by people who aren't operators right that's not how it is occasionally occasionally and that's cool if it is awesome i love it typically it's more we have a certain set of strengths and certain set of weaknesses. that set is unique. That complete set is unique. No one's copied our weaknesses, for example, right? And so if, although we have these weaknesses, if those weaknesses enable our strengths to be even stronger, because they're actually trade-offs is what they are, right? If we've traded off into a stronger strength, then we're special in a certain way, a certain shape. And our competitors are special in other shapes. Or they haven't done this, and so they're just sort of a blah shape, and they don't have a story to tell. Anyway, some customers will like our shape better than other shapes, and some won't. And that's what it is. But as long as we have a special shape of all these things together. So we weren't just fast. We overinvested in things like the very fastest CPUs, the very fastest memory, this incredible amount of attention to all ports of the stack and all this. Is it uncopyable? No, you can do all of those same things. and Almost no one would because it costs money, especially not the shared hosts, big shared hosts, where it costs $2 a month to have a site. They can't afford that. They could physically do that, but they won't because it doesn't match their business model. It doesn't really match their clientele who actually doesn't really need that. So it's not that nobody could copy it. They could, but it doesn't make sense for them and their set of tradeoffs to do so. But for us, it was like, well, but our whole thing is we're the most expensive but also the best. We had better make good on being the best. And that's also service, as you say, because guess what? Good service is also expensive because it's people. Well, now it's AI. Okay, now that might be different. Fair enough. But then, yeah, good service means people who are capable and available. Well, that's expensive. Right. So we charge more, do more. Could other people decide to make a super duper group of tech support people and have an expensive plan that does it? Yes, anyone could decide to do that, but it's very hard. And then at scale, it's extremely hard to do that well at scale. But we invested in that and did do that because it was part of our super special things. But that means we have weaknesses. Like our prices were higher than everybody else.
- Speaker #1
Yeah.
- Speaker #0
Okay. So you can copy our strengths, but are you going to copy our weaknesses? Or does that not work for your customers? Another thing was we said, look, we want to make it, you bring your site to us and it's so fast, so much faster. You're like, what the heck's going on? That's how much faster we want. What number is that? We have numbers, but it's actually not even important. I want you to go. Oh my God, what kind of magic is this? And then I was like, see, that's fast enough. Whatever that is, that's good, right? Well, there are certain plugins, certain code for WordPress that is slow. And there's really nothing to do about it because the algorithms are just bad. So we disallowed those plugins. That's a massive weakness. If you have a website, you depend on this plugin, it won't work here. That's an incredible weakness. You don't want my business? I want it, but no. So when we hold fast to these weaknesses, High price, don't support everything. But because we hold fast to that, therefore, we can, in fact, be the fastest and make good on those promises. And so every company has, of course, weaknesses and strengths. And your job is to well, the important thing is that those weaknesses are feeding, enabling the strengths, not just weak. And then you have a special set. To me, that's a strategy. You have a framework for personal fulfillment about joy, skill, and need. I'm curious, as you've been using Cloud Code every day, as part of the joy you had maybe earlier has become a skill among your Cloud MD files, how have things changed for you in your day-to-day interactions?
- Speaker #1
I feel conflicted. Because it is fun to learn new technology. And again, the magic is magic, which is joyful and useful.
- Speaker #0
Yeah.
- Speaker #1
So when that happens, it's fun. And even the part where I have to figure out how to cajole it and figure out how to change those files, iterate. It's an engineering problem, and that is fun for me.
- Speaker #0
Yeah.
- Speaker #1
So that aspect can be fun. But I also find that at the end of the day, I feel drained, not happy, not full of energy. And I feel bad, which is not how I felt when I used to just code all day. I felt energized like I didn't want to stop. I still feel the I don't want to stop, but it's more like a Twitter-style dopamine hit of like, but wait, I have this one more. I have three windows, and that one just finished. It's that kind of bad kind of don't want to stop instead of a good kind. And having talked to people about it, I find lots of people feel the same. It's thrilling to make stuff. either quicker or things that they couldn't have done before is thrilling but um but you feel drained and bad after like you've doom scrolled i think the context switching is part of that you know it it is faster still if you just use one thread it's still faster than you do yourself probably not for all things actually there there are types of things where that is not faster you should just open the file and change the number like get it come on have you forgotten everything you know but there's other things where even a single threaded you're three times faster like prototyping a brand new project or something which is so much crap you just have to scaffold and often blah blah blah and install the stupid you know libraries and just set up the and oh my god it's so much faster than you're doing that and besides that was that was never the fun part so hooray it did for me you know so um clearly it depends on context But it's when you have even just two things at once, maybe three, but certainly two, it really starts to accelerate because you're ping-ponging. And it's not perfect. You're not keeping both at perfect capacity. There's never waiting. But you are getting more done than single-threaded, right? And more than three, I don't know that you can keep it in mind. Even three can be hard for me to remember what all is going on unless it's really distinct threads of thought. But the context switching of that, while effective in terms of productivity, is super draining. It's not fun. In the moment, it kind of feels fun again because of the dopamine thing. Oh, it's done. But, God, I got so much done today, which is totally true. I think it hurts us. Now, is it just the context switching? Is it the fact that we're managing and orchestrating instead of building? But you are kind of building because it's your idea. Does it really matter whether you're nailing the nails in? And the answer is, I think for some people, they never liked nailing the nails in. And so, yes, being able to describe a house and it appears is fun for them. But there's a lot of people, me included, where the craft of building was the fulfilling part. I just love that. I mean, one of the reasons I even did WP Engine is because I love optimization, code optimization problems. I used to just read like... Every implementation of, say, a cosine and sine that any library made. Oh, look, they did this other thing where they rounded the other way or they used this other table. Or, oh, look, with Go, you can compute sine and cosine simultaneously more efficiently than one at a time. And often you need that, like when doing an angle-y thing. And so, oh, look, they actually, for them, cosine is just doing the cosine together and picking off the cosine one. Because if you need both, it's just as fast. That's so cool. I never thought of that. I love that. So I don't do that anymore. So the part of it that I love, not the only part, but part of it that I loved is gone. That sucks. And then context switching and management of things that sometimes are magical, but sometimes they're stupid and you get real mad at it. I told you not to do that. Yeah. I'm so sorry. You're right. I know. All right. But you would say that even if I wasn't right, but okay. Which is patronizing and that's annoying too. Yeah. So actually it's a terrible relationship interaction and you're context switching and you're not crafting. So for some people. That's removing a lot of the joy, even if they're very efficient at being productive. So I don't know. I'm not an expert in this. This is just my experience in talking to a few people.
- Speaker #0
So what's your... How do you deal with it personally? Like, do you sometimes just go back to VS Code or your IDE and code yourself?
- Speaker #1
I'm in the IDE anyway because I'll, like, kind of do a little of each,
- Speaker #0
right?
- Speaker #1
Sometimes I'll even use AI to do stuff like, hey, where is it in the code where this happens? And then it shows me, like, great, I'll take it from here. But that was actually nice because I couldn't remember and find, you know, it takes a little longer. Like, that was actually kind of helpful, you know? And again, like, I find it. fun to make skills and things that are not code or i have just enough code to give it yeah tools so that i can you know that's pretty fun like even just in personal life and you know doing stuff that that can be pretty fun um but there's no stakes and there's no timeline it's just fun so that feels more like playing so it's okay right um uh no i would never want to go back to coding the regular way especially knowing like this could just go really fast it just it just doesn't feel like worth it anymore. And that's really sad. And again, I'm not saying we shouldn't do that. Humanity shouldn't do that. That's bad. I'm not saying any of those things. I'm not getting a value judgment on it at all. I'm saying I wouldn't do it. I'm just saying we can also acknowledge at the same time, it's sad when a craft is lost. There's certain craft in audio design that's lost. There's certain things in photos that's lost. Okay, not completely lost. Some people do. I know, but okay, it's largely lost. And is that better for society that we all have? a phone a camera in our pocket that's that's better than almost any camera that ever existed yes i'm for it you know like i don't think we should stop the technology at all that's not what i'm saying but we can also lament it's sad when humanity loses a craft just like it's sad when we lose a species yeah does that mean we should never do anything no but we can still be sad when we lose a species right i mean we can we can do both we can say both are true i think both are true and so as a as a builder
- Speaker #0
Where do you find your new bliss now in this world? Well,
- Speaker #1
I'm writing a book.
- Speaker #0
Yes.
- Speaker #1
It's called Hidden Multipliers. And you can get it at hiddenmultipliers.com. Look at the sales pitch. Amazing. It's about how you can grow because, of course, everyone's maybe not. And so it's a bunch of ways to jumpstart growth with your current budget and team. So there, I did the big pitch. I had my indulgence. Now we have to get back. Anyway, so I'm writing the book. And that's the answer, in fact, because… If it's my ideas and my writing and my everything, that's why people would buy the book. If AI wrote the book, you wouldn't want to buy it. The only reason you'd ever buy this book is because you're listening to this now and you're like, this seems pretty good. So if you have 300 pages worth of specific things that help me grow, I want to at least, yeah, maybe I can find something for me. That's the only reason. And then also I have my own writing style because I've been writing for 20 years. Maybe you like it, maybe you don't. I do not write the way marketers write or the way they tell you to write like a fifth grader, really short sentences. That's all you do. Like really short paragraphs. I don't do any of that. My sentences go on and on and then there's a semicolon and we're still going. I mean, I don't want to write poorly, but I just, I'm just not going to write like for a fifth grader and they're not going to insult my reader's intelligence. And so for some people, they, I guess they want, they want it to be more simple and just, just give me a worksheet to fill out. And I'm not, again, I'm not going to do that. I don't think that's reality. It's just not my style. But for people that want long form that's been really thoughtfully written and not written by AI and ideas that form a cohesive structure and narrative and hopefully with little bon mots and fun turns that make it fun to read and not just the information. If that sounds fun, well, that's what I did. And so that takes up a lot of time. It's very slow writing. All my articles take dozens of hours because, I mean, it's just it's hard.
- Speaker #0
So this book is actually only for is not only for software businesses. It can apply across a spectrum.
- Speaker #1
It's not only for software. It's not only for SaaS. However, I take my own advice. One of my own pieces of advice for software or for books is that you have to have your absolutely perfect customer in mind. Of course, in software, we call this the ICP, the ideal customer. So you have that in mind and everything you do has to be for them. The truth is, because you were so specific and clear there, a wide swath of other people, and you might say a bullseye, are also interested. Not all of them are interested because they're not the bullseye, but a percentage are, and there's way more of them. And so in total, most people who read the book will not be my ISP. Most people that buy your software are not your perfect customer. Nevertheless, that's your North Star. That's your vision. That's your strategy. That's what you're working for. It's focus. It helps you make decisions. It makes everything clear on the homepage. It's very important to do that. Okay, so I'm just saying all that just because that's useful for everyone in all cases. And I have two chapters in the book, by the way. briefly making the case I just made with an example that's very fun, and then spending the whole time saying, well, how do I find my ICP? How do I know which customer profile is ideal for me? How do I go about really articulating that so that I go do everything you just said? So that's in the book. Anyway, so I have to do the same. So my ICP is a bootstrap founder with $2 million in ARR in a SaaS business where growth has slowed, maybe stopped, and you They don't know what to do. And by not knowing what to do, I don't mean they have no ideas. I just mean they're like, well, I can't keep doing what I've been doing, obviously. And I have a list of things I could do, but I'm not really sure how to choose. And maybe the best thing I could do might not be on that list either. And even if it is, I don't know which of the 10. So I just don't, I'm just, I'm stuck in that sense. I'll do work tomorrow. I'm not, you know, but like in a strategic or a higher level sense, I'm stuck on how to undo this or get growing. That's who I've written it for. go. Do you have to have a software or a SaaS business? You do not. I can even point to certain chapters, like I just said, where that applies to every business and even books. Hooray. But I can point to other chapters, like there's some stuff about retention. Or if it's not a SaaS business, that doesn't mean anything. So you could skip that whole chapter. So that one doesn't apply. So I would say the further away you're on the bullseye, some applies, some doesn't, of course, because I wrote it for one very specific person.
- Speaker #0
And I assume that it's a mix of... it's a set of principles that are almost independent of the times, meaning that the fact that now it's not as much SEO, it's GEO or whatever. Oh yeah, no,
- Speaker #1
there's no tactics like that.
- Speaker #0
No tactics. It's more about, I guess, product marketing.
- Speaker #1
Well, what it's about is understanding and strategizing about your business.
- Speaker #0
Yeah.
- Speaker #1
These are mechanisms that are common to most businesses because they're just built into how it works.
- Speaker #0
Yeah.
- Speaker #1
And so when you have this... Now it's more, it's more like knowledge and learning and understanding.
- Speaker #0
Yeah.
- Speaker #1
And then having, of course, having examples of what people actually do, but those are all just illustrative, right? You copy exactly what someone did. That's that, that probably won't work. And as you say, with AI, like maybe it really will be totally different. So those are only to illustrate the lesson or the deeper concept. Those are the things that you take forward. For example, this thing with the ideal customer, I've already had the notion of, Hmm, so I have this whole process for finding it. Well, what would you do with AI? Well, you can't ask AI, who's my perfect customer? It doesn't know either. It's just going to generate something that sounds good. You already did that. You could do that yourself. That's not right. But what I find AI is good at is facilitation. It can't give you the answers, but it can prod you pretty well. So I built a set of skills where each one would do one of these steps, right? And the idea is that you're iterating through this thing. But AI is maybe helping to brainstorm. It's certainly prompting you. And it's also being a foil. It's kind of being a devil's advocate. Oh, you said that was good, but is it? Because what about this? And like intentionally being, you might say mean, attacking you for your benefits, either you're still right, but now you really understand why, or maybe you're not right. And okay. So it's very hard to do that alone. You can't really be your own devil's advocate. You need a facilitator. Well, who has a facilitator? That's hard. And so all the consultants and raise their hands, me, I'm the coach. You're right. That's part of what a coach is great at, not telling you the answers, but Facilitation. I agree. Anyway, most of us don't have that and so on. So if I could turn the structure of the book into, let's say, a facilitator who's helping you do that, I think that'd be a fantastic use of AI.
- Speaker #0
And are you doing that?
- Speaker #1
Yeah, I think that'd be great. Of course, anybody could do that. I'm happy about that. I don't want anyone to wholesale copy the book word for word and publish it as their own. That's too far. But when people... I'm actually a huge fan. I've written about this specifically. So I'm on the record saying like, I love it when people take my ideas and go. So my entire blog, so 18 years of articles, right the second you could go to my name and there's a section called, in the title bar, it's like about me or whatever. And there's a section called special links. And in there is an RSS feed with not only every article in the metadata, but all the content of every article in a simplified markdown blob, including linked... images and the images are like lower res. So if you process those, it wouldn't be as many tokens because I like to have high res on the website, but you don't want to use that for tokens. It's all right there. It's all, it's of course, it's automated. I make any change that's updated. So if someone wants to, and of course people have done this, if you want to ingest that in any way, you can just do it. Like just download the thing. That's it. You did it. So that's just what I'm saying is I love it when people remix or, you know, people have taken some things, put their own thing on it and then published it as like this is my consulting thing you know great like i love the remixing and the stuff because i don't i've never made money writing or on my on those ideas and i still don't want to make money okay i want to sell a book but like that's that's just because it's dead it costs money to make the book so you know you know like i'm still not trying to like make a lot of money writing a book i know that that won't happen um so i love it when people take the ideas remix it perhaps and do shit with it. So if other people make those skills, awesome. So yeah, I might try to make it, or maybe someone wants to work with me to do it. That'd be great. If I could just give all that away, that'd be awesome.
- Speaker #0
Maybe the interesting thing is that you're... pushing your bliss to some other avenue, meaning that you're not having as much fun today writing code because of AI, but you're having fun playing with the materials because if people can leverage your book, your content to help them in ways that was not possible before because they couldn't access you.
- Speaker #1
Yeah, I mean, I've been writing for so long that... you're right, but I don't feel it as a shift just because to me writing, I've been doing it for so long that it doesn't feel like a shift. It just feels like something I've always done.
- Speaker #0
Yeah.
- Speaker #1
Right. But, but you were right, but you're never, you're right. It is a shift in that I'm kind of putting it on there. I wouldn't quite say I don't like writing code anymore because the part where you can be more effective and do stuff is super fun. So I would just say it's, it's, it's both, once again, it's both at the same time. It, in some ways it's more fun than it's ever been because the stuff you hate also just kind of goes away.
- Speaker #0
Yeah.
- Speaker #1
I love that. So kind of some parts of it I love more than ever. But then other parts, no. And it's just both. I think anything that's an expression of humanity is not something AI is taking away. Now, I'm not going to prognosticate about the future. I just said I don't think anyone can. I include myself. However, I can point to lots of examples where AI is way, way, way better than humans. And also, we don't care. Like in chess. Chess has been better than the very best human on earth for 25 years. Yeah, and has only gotten better. So okay, does that mean we don't play chess anymore? On the contrary, chess has never been more popular in the history of the world as today. There's people with YouTube channels with millions of subscribers, and also Twitch with things live, and also smaller YouTubers, and also there's never been more money being made in courses and tutorials and teaching ever. But computers have been better than us, not just us, like us little folk, better than the best humans for decades. We just don't care. I mean, we use a computer, right? But in terms of our interest in chess and even our interest in consuming products, such as courses, tutorials, videos, buying chess boards, going to tournaments, you know, just the products in the more broad sense. Our interest in buying products has never been more. And our excitement in the game has never been more. We use computers in new ways, and that's interesting. The computer did not supplant chess at all. Why not? Because it's a human thing. It's a human struggle. There's sport. Okay, is it physical enough? It doesn't matter. It's a human effort. So whether you're trying to do better yourself or you like seeing other people doing that, it's in the suspense of sport. That transcends, like, of course the computer's better, but all the sport's gone, and we're here for the sport. We're here for the humanity, and also I can do it too. There are some Olympic sports you look at and you're like, okay, I'd never do that. But chess is something where, of course, at your own level, you can do. Fortunately, there's 8 million people also on your level, and they're all online. So you can all do that. How cool is that? That's a very human thing. So I think those things are fairly insulated. Not that AI won't change it. AI has changed chess substantially. People's styles have changed. We've learned things. Openings are different. Opening prep is different. We learned that you can advance pawns in a certain way or even get rid of pawns, and that's okay as long as certain other things are true. Computers taught us that. No human ever thought that. In fact, it was contrary to what you said humans need to do. So we have learned from computers. It's changed the game. And studying chess, because you study with a computer, it's changed studying the game. And yet here we are doing it more than ever. So it has not stolen the game from us. So me writing the book is like that. Who reads books? I don't know, but people weren't reading books before AI either. So that was already, that trade's already left. I know that. Again, I'm not trying to sell a million copies. I am under no illusions. I'm doing it for me and my craft of it. And for the people who like that too, which is small, it's like, you know, I get that, but for them, they'll enjoy it too. And that's, that's enough. And, um, you know, even now, like when I see a YouTube video, it's clearly like a computer. I just click off of it. Now, maybe some people do and some don't, I don't know. But to me, I'd rather have a human doing a bad job at a video than an AI doing a great job.
- Speaker #0
What I was trying to say is that the book or the content that you've written in the last 18 years is incredibly useful. But sometimes we are bad as humans as remembering all the articles or all the concepts that somebody has shared. Yeah,
- Speaker #1
me too. That's why I tweet stuff. And they're like, oh, man, I need to hear this today. And I very often respond, me too. That's why I wrote it. I was like. Oh my God, I forgot this. I probably said it a hundred times. I still forgot it. Yeah, of course we do. We're just humans. It's okay. We're all trying to get along.
- Speaker #0
But that's a beauty in a way that at least for me, that's a beauty of AI in the sense that some of the concepts that you've developed over time, like for example, in your book. If they become tomorrow skills or what have you, I can read the book and then I can get the book almost as a companion in my life.
- Speaker #1
I would love that.
- Speaker #0
Where I'm saying, oh, I read this book. I like this concept. Let me try to see. Let me pick Jason's brain on that problem. I'd love that. And that's the beauty of it, I think. I mean, that to me is a. It's a completely novel...
- Speaker #1
I think you could just pick up that skill and not read the book, and it's probably helpful, and I'm fine with that. I think it's better if you know what's going on instead of... Yeah,
- Speaker #0
because it will click more.
- Speaker #1
Yeah, you won't really know what's happening. Yes. And plus, just as a human being, are you an automaton too? Do you just feed something on a computer and do what it says? Because now you're the automaton. It might be okay, and actually, I would say for many tasks, sure, I'm fine with that because those tasks don't matter to me.
- Speaker #0
Yeah.
- Speaker #1
Right? And I... do want to spend my time on things that are important to me, and this isn't one of them. So, okay, good. Yes, you're saying, yes, I am okay with that. Okay, great. But then there's another context where maybe you're not okay. There's plenty of people who, for example, they want to draw. So you don't want the AI drawing for you because the whole point is drawing, but you're bad at drawing. I know, but the whole point is I want to draw, you know? So I play the piano. Not better than all the people I listen to. I know, but it's my own expression of it, and it's my own doing, and it's my own self-challenge of doing. And of course that's what it is. And so it doesn't matter. So you should, it's, it's wise to pick which things just need to get done. And so, yeah, if AI just does it, I don't even need to research, screw it. Great. And other things where like, it's the strategy for your whole company. Maybe you want to be a little bit more engaged in that and yet have AI help assure it, you know? Yeah. I think, I think in fact, probably the best way to learn would be that you make the skills, you take the book. And then maybe you take a markdown file that also I might provide, right? So that you don't have to type it in or whatever, right? So you have the book. You have the thing you can feed in. Then you make the skills. Because that would force you to see what good looks like in terms of facilitating this. Now, it's that whole thing, you know, if you tell me I forget, if I do it, I might remember. And if I teach it, I understand. Yeah. Right? So the equivalent of that is you're teaching the alum how to do it. So by the time you read it and you make the skill, sure, from now on, you're going to use that skill and go fast. But you've done the teaching part, which means it's super deep in you. You are really good at this now. And yet you have the automation. To me, like when I personally use LLMs, that's what I do. Even when I see there's a skill somewhere, I'll read it just to see like, yeah, I got to learn from other people. But a lot of times I'm like, but I'll learn if I make this skill. And then I'll have a skill I can run a thousand times. So I get both. I gain the understanding and I gain the efficiency. For me personally, that's fun for a lot of things.
- Speaker #0
You get it to do something in the direction that you want.
- Speaker #1
Yeah, you're right.
- Speaker #0
Because otherwise, you're stuck in the direction of somebody else. And in a way, it's like when you're using somebody else's software and you don't like it and you don't understand why the UX is shit.
- Speaker #1
Well, and it might be wrong for you, right? Just because it was right for them. So, okay, there's parts of it that are right for you and parts that aren't. So either you start with theirs and then massage, or you kind of read it through, get some ideas, and then start your, you know. All of that, I think, is really good. But I think, yeah, I think when you make the skill and you continue to, like you said, you didn't like the output. Well, that's a very personal thing to say, I didn't like the output. You're making a judgment there and making it whatever you think is good or whatever matches your context and so on and your personality. That's freaking cool. And there I think you're really engaging, learning, teaching. That's being alive still and not being an automaton. So I think that's pretty neat. So, yeah, I'd love it if people do that with the book. or maybe I'll have things that... I think work pretty well that maybe you could start with for people that want to start with. Or, I mean, again, like I'm not judging if people want to just take it and say, I'm just going to run this, get an ICP and move on. I remember what that means. I don't care. I can just do it for the craft of it, for trying to get my best ideas out there, my best way I can. And also there's stuff if it can be free, because again, I'm not trying to make money or anything. What if someone goes off and they, good, hooray. I mean. To me, that's also a lot of the real original open source spirit, right? It's share alike, you know, take, remix. That's awesome. How many things have I taken and remixed even by accident? I can't even count. Okay, good. Well, I would like other people to have the same, enjoy my stuff the same way. We all do it. It's great.
- Speaker #0
So you're a trained technologist that became a very strong, I would say.
- Speaker #1
marketer in a because you're a lot of what you write is more marketing but in a strategic way not necessarily in a tactical way yeah i never write about like here's how to write an ad here's how to do it never ever it's yeah it's more of the structure the structure is really what it is is engineers want to just write code yeah and they don't want to do everything else yeah and so I was always writing for that person. So the last thing I need to do is talk about writing code. That's what they already do and they already read about. And I have nothing new to add about how to write code or how to make a unit test. That's already there. So for me, it's like, no, what I have to add is I'm an engineer too. And I'm telling you, you've got to do this other stuff. And I can explain it, I think, in a way that, you know, will resonate more with you or maybe cajole you into doing these things. And so that's what I'm going to do. So I'm going to write all about those other things. Not because you shouldn't write code or that I don't write code. It's just... I have nothing to add there. I have something to add over here. Yeah.
- Speaker #0
Well, thank you, Jason. I wanted to ask my last question, which is a question I always ask. What's the most important thing you've learned from your mother?
- Speaker #1
You always have to be your own advocate. No one else is your advocate. Even people who are really good and want to advocate for you still. are not as good as you are as your own advocate. I watched her take care of my dad as he had multiple myeloma for nine years, 10 years. And we had good and bad doctors, good and bad nurses, good and bad, everything, advice and everything. She's the one who was the best doctor. And even though there were some of this is the best doctor for this and that, and I know they're very good and I know they're trying very hard. They can't be as good because they're seeing a lot of people. They can't be as good. So That doesn't mean you don't listen to them. That's not what I said. Of course you do. But you advocate for yourself. Wait, but this doesn't seem right. But hold on, this happened. And oh, yeah, you're right. And the people miss stuff all the time. And so whether it's a self-promotion as people do on social media because no one else is promoting you or at work or something like that, which is more heavy but also true. This notion of like it's almost like that phrase of you go around worrying what. you look like what people think of you but no one's thinking about you they're thinking about themselves what people think they look like right no one's thinking about you right so in that same sense no one's advocating for you even when they kind of would but they're just not that's just not what they're thinking about um so um so you need someone advocating for you really either yourself or in the case of of my parents say then someone who truly is advocating for you you need that person. which could be yourself or maybe it could be someone else, but you need to be your own advocate in the world.
- Speaker #0
Thank you very much, Jason. Thank you for joining in the board session. Please make sure to like subscribe, add a five-star review or anything that you can do to make more people discover our show.