Gokul Rajaram: How to Analyse for Durability and Defensibility in a World of AI

When you are introducing yourselves, what do you even say?

Because you've done so many different things that I don't even know, like what you take credit for most or what you like hang your hat on.

Well, it all depends on the audience.

If I'm talking to a bunch of car nerds, I talk about winning Le Mans.

If I talk to a bunch of programmers, it's all about Ruby on Rails.

And if it's a bunch of entrepreneurs, it's usually about building Basecamp and 37signals.

So did you, have you ever said Basecamp's revenue?

No, we've never disclosed the full revenues nor full profits.

Um, I think we've said tens of millions or something.

I don't even know.

That's one of the great privileges of being a private company.

It certainly served us well for many, many years when we had the entire project management space in SaaS to ourselves.

And we were laughing all the way to the bank because clearly someone else must have realized this was a highly profitable not even niche, but category of software.

And it just took a decade until anyone else woke up to the idea that this could be a good business.

And then we had competition.

So, um, very much enjoy the privilege of being a US private company and not having to disclose a goddamn thing.

Sam, have you ever been to— if you go to 37signals website, Sam, have you ever clicked at the bottom?

There's this little tab that just says $19.99.

Have you ever seen this?

Uh, no, I'm going to it now.

What's it say?

It's amazing.

So, so in general, I feel like you guys have done an amazing job of like planting your flag in the ground.

Like, yo, this is what we believe.

And Rework was a fantastic book.

Your website right now is basically kind of like Rework.

It's just, uh, 30 chapters of shit we believe, um, rather than trying to sell you on our software.

Like, let me tell you our philosophy as founders and entrepreneurs.

But if you click 1999, Sam, what do you see?

I see a website that— a page that looks sort of like it's from the year 1999, and it's all white, and it says 37signals Manifesto, and it's a 37-chapter book in plain text.

And there's chapters like Chapter 24, My Cousin's Buddy, and it says, my cousin's buddy is a web designer.

And then you guys wrote, then let him do it.

The money you'll save having your cousin's buddy build your website will be nothing compared to the cost of time and money to undo his mistakes.

And so like very much this, like this blast from the past of you guys have been doing this for like, you know, whatever, 27 years or something of saying what you feel unapologetically

and having that be something that, uh, like leading with the philosophy rather than leading with like, hey, here's our features and here's our products, please use us.

Yeah.

So a lot of it is simply based in the fact that neither Jason and I can shut up about what we believe.

So it is a natural byproduct of us existing and working that we keep coming up with takes on different topics in our industry, how to build a company, how to run a company, how to do

technology, how to do design.

And it would be an awful waste in my opinion, if we didn't try to capture some of those observations and lessons and distribute them more broadly.

But there's certainly also a more

economic side to it.

When we got started, I think it's in Rework, there's a chapter about, uh, how to out-teach rather than outspend the competition.

Competition,

and that was really our operating paradigm since the beginning because we never raised VC.

We never had more money than anyone else.

We always had to

earn the money we wanted to spend first, and that meant a whole host of constraints on how we could run our business and certainly how we could run marketing.

Because if you don't have a bunch of money to buy ads and awareness that way, you have to earn it.

And the way to earn it is simply to be interesting.

Now,

I don't set out to be interesting as an objective.

I try to be ruthlessly honest and very forthright in everything that I learn.

And that in turn turns out to be interesting to some people some of the time, because so much of business advice is either being given by people who aren't doing, who are observing

and deducing.

And there's an entire category of that, usually padded to fit exactly 300 pages.

And maybe it has a funny exclamation point in the word fuck instead.

Like, I think that's the most popular genre right now of how to quit my fucking job or something.

That's what I see at the airports all the time right now.

Or it is from entrepreneurs who are active, but very often have all sorts of liabilities in terms of what they can say because they have investors or they're publicly listed or whatever.

And basically Jason or I had none of those constraints.

We were in the thick of it, are in the thick of it, running our own business, having to make more money than we spend.

And it also, we don't have any of the constraints.

We don't have anyone who could tell us no.

Now, some days I wish there were people who could tell me no when I tweet out something stupid or, or write something harebrained.

But, and the net sum, the moving average of that arrangement of doing something that is successful enough that people care about what you have to say.

And then not having that coarse filter of, oh, I wonder what my investors are gonna think about this, has produced over the last 25+ years, a steady stream of observations, mission

statements, crazy attempts to do this, that, or the other thing that very often ended up not being what everyone else was doing, partly because we weren't consuming the same water.

Now, neither Jason nor I have been in Silicon Valley.

We are not products of that.

However much I have tremendous respect for everything that's come out of that place, it does have a tendency to produce uniform thinking around a whole host of big topics.

And 37signals was born in Chicago at a time where that had zero tech community of note.

It wasn't until much later and Groupon and all these other things that some life was bred into the Chicago tech scene.

But when we were coming up, there was no one there.

So we just had the luxury of distance to everyone else's echo chamber, which meant that original thought was possible.

I did— I'm not gonna claim that all the original thought was good, but I will claim that if you look back upon some of the things that we wrote now 20 years ago, Getting Real, the first

book, that manifesto from '99, Rework from 2010, A lot of that stuff still holds up because it's focused on fundamentals and focused on extracting lessons from our work, producing something

that was worthwhile and lasted.

All right, so this episode is all about excellence.

A while back, I shared my personal framework for building excellence in my own life, and the team at HubSpot turned it into a 30-day operating system you can check out right now.

It breaks down the systems it took me 10 years to figure out and shows how I actually use them day to day.

These are systems that genuinely changed my life.

So if you want to build a good life, scan the QR code or click the link in the description.

Now let's get back to the show.

So 20 years ago, you were in your 20s and I don't know how successful you were.

I would imagine mildly successful, but still had a lot to go.

And yet you were giving advice.

I think a lot of people listening, myself included, when I hear you talk about teaching versus, teaching versus spending, I think that's great advice.

But I'm like, I don't want to give advice when there's so many people that are more successful than me.

They should be giving advice.

You know, this sort of imposter syndrome stuff.

How did you justify teaching fundamentals, teaching advice, and writing a book on work when you're still in your 20s and there was probably 1,000 people that were above you that could

have been like, what the hell does this guy know?

Part of it, I think, is genetics.

I've been blessed with a completely irreverent personality that thrives in exactly those kinds of circumstances.

I don't give a damn what people think I can and can't say.

I'm going to call it like I see it when I see it, whether I'm 19 or 15 or 46 or 37.

And The other aspect of that is that both Jason and I did spend quite a long time working for other people.

Now, long time in the grand scheme of things, did we spend 20 years?

No.

But I'd been working since I was 15.

Uh, well, even earlier than that, but I'd been working for other people directly in the sense of I have a boss since I was 15.

So by the time I'm 26, I've been in business to whatever degree that is, either as an employee or running myself for a decade.

Now, I don't even think you need that.

I think what you need is exposure to novel terrain with enough sort of influences and being informed somewhat of what other people think.

And then like your token generation machine will start producing.

If you are of the kind of constitution that I am, which is one of the ways I try to internalize the best lessons from what I'm going through is to share them.

Is the old adage that nothing makes you learn a subject better than trying to teach it.

Right.

Very much true when it comes to this.

Now, the other fact of that is there's this great paradigm of liquid versus

crystallized intelligence.

If you look at Nobel Prize winners who win in physics and chemistry and math and so on, They all do their formative work in their 20s, basically.

Then there's a, a decade of, of application in their 30s, and then they get the Nobel Prize for that work they did in their 20s in their 40s.

Part of that is just biology.

I mean, I remember when Mark Zuckerberg said this and it was hugely controversial.

I think this was in the 2000s.

He said something to the effect of young people are just smarter.

And everyone was like, you can't say that, like, that is just so wrong.

Well, it's not wrong in a certain category of intelligence.

This form of liquid intelligence, smart, uh, fast signups, firing stuff— young people are smarter, just like they run faster, right?

Like, you don't have a lot of, uh, I don't know, 100-meter sprinters who are 42.

That's just not part of the human physique.

Now you get the payoff when you get older that you have more of the crystallized intelligence.

You can make more connections.

If you look at Nobel Prize winners in history, I think the average age is like 80, right?

Like those connections come later.

So I have tremendous respect for that liquid intelligence that's at once both very smart very quick and also totally ignorant, has no sense of the broader world, has no sense of what

you're allowed to do and not allowed to do, have no sense of someone knows more, therefore I should shut up.

Right?

I look back upon that time and go like, I am so glad I embraced all that ignorance with the hubris of early 20-somethings.

That's how we change the world in a very literal sense.

The new ideas very often come from that vector.

That's, uh, no, nobody loves a cautious 20-year-old.

Uh, the founder of, uh, co-founder of HubSpot came on this podcast and he talked about in his first company he was making everything up, and his second company he said, oh, now I, I

know how I'm gonna do everything differently, and it totally failed.

And he goes, just 'cause I was ignorant doesn't mean I was wrong.

And there's, there's something to the, uh, to, to that idea of like you can be both ignorant and correct.

I'd go so far as to say that ignorance is a benefit for a huge class of problems, that you are cursed when you've been through the loop once.

When you know too much, you cannot unseed in the same way.

You will be locked into paradigms and thought patterns.

And if you want to break those paradigms, if you want to break those thought patterns, you kind of got to start from a clean slate.

Now, that's not universally true in all cases.

But I do think it describes the case that Jason or I went through, that we had some input which was working for other people, working at a crazy time in our industry, maybe second craziest

time next to right now, which was the dot-com boom and bust.

Both of us went through it both on the way up and on the way down and saw all the dynamics play out where I think if you compress those years from let's say '97 to 2001, like that's

only 4 years.

That was instruction worth a decade, at least, maybe two, because there was just so much happening.

You got all the explosion of, here's the internet, what are we gonna do with it?

Build into the wild castles in the sky that came tumbling down in 2001 as the dot-com boom burst.

So we had all those inputs, right?

Like all those experiences, and then we kind of started in the desert when Jason and I got going with Basecamp.

This is back 2003.

We start working on it.

It's still a wasteland, right?

Like the dot-com bust have exploded everything.

You got programmers going back to being chefs or gardeners or whatever.

There was simply no work to be had because the whole industry blew itself up.

And in that wasteland, we just go like, uh, do you know what?

We now have to be scrappy.

We now have to rely on ourselves.

We have to be tremendously productive because we can't hire a team of 50.

We can't buy tons of servers.

We have to use open source cuz we can't afford licenses for Oracle databases.

All of those constraints apply themselves to our situation and then just produces an explosion of creativity cuz that's what very often happens.

When you are deprived in all the right ways, you will find out, oh, there's a better way to do this.

You will find out the optimizations.

Like, a parallel to today is the big scary moment for the US AI industry was when Deepseek put out R1.

And some of this is mythology, some of it is bullshit, but the storyline at the time was they could not buy Nvidia CPUs due to export restraints.

They trained the whole thing on 5 million.

I know there's reservations and asterisks behind that, but the fundamental principle was they came up with some novel new techniques of using vastly less compute to still get quite

far.

And I think that applies very much to our situation at the time.

We just, we didn't have millions of dollars.

We had a consulting business that we had to make pay for all these experiments we wanted to do, and therefore we just had to be radically more productive.

I mean, this in my case on the technical side was the birth of Ruby on Rails.

I could not use the same technology as everyone else was using because I had a 10th, a 100th, a 1000th of the capacity.

Therefore, I had to use something else.

I had to use something or come up with something that allowed me as an individual programmer and technical person and the whole operation to build Basecamp by myself on the technical

side.

I am forever grateful that these were the constraints we were under.

If we had raised, I don't know, $20 million and hired 100 people, we would just have used the same shit as everyone else and we would never have come up with Ruby on Rails.

Do you guys wanna hear a crazy stat?

Uh, Nobel winning prize scientists are 22 to 25 times more likely to be in some type of performing class.

So acting, magicians, singing, or some type of like art, like drawing or whatever.

Is that right?

22 times more likely.

I read about that in Bill Gurley's book, and then I just looked it up and it's like, it's like a study.

And the reason I bring this up is I've been, I've been obsessed over that fact.

That just seems astronomical.

And then you brought up the idea of crystallized versus liquid knowledge, and it just solidifies that.

It kind of adds to that.

Isn't that a crazy stat?

That is wild.

And do you know what's funny?

I mean, we all read in whatever we want into factoids like that.

But when I was a kid, what I really enjoyed being was the dungeon master.

So I played Dungeons and Dragons and other role-playing games.

And most people, they just wanted to be a character.

I wanted to run the world.

So I think there were some, um, sort of, uh, early indicators that I like to sort of worldbuild.

Also in game, my favorite game of all time is Civilization I, the original Sid Meier's game, which is obviously all about world building.

So some of those things were present early on and having positive experience building and bootstrapping with others, having to convince others before you even have money to pay them

salaries that they should go in your direction and we should work on this thing together and you should Come into my dungeon is, I think, exceptionally good preparation for running

communities, running open source, running a company.

You know who else's favorite game was?

Civilization.

Mark Zuckerberg.

You know who else?

Elon Musk.

Yeah, both of them have talked a lot about playing Civilization early on, and I feel like they almost had to stop talking about it for a while when the PR got kind of bad, and it's

like they want to play, you know, play God a little bit, but now it's cool again.

Speaking of Facebook and Zuckerberg, you've got a lot of things right.

You know, I think you guys were really early to remote work as a concept.

And, you know, obviously, you know, being profitable as a strategy is that, you know, was, you know, somehow contrarian at the time.

But one thing you got wrong famously was you wrote this post called Facebook is not worth $33 billion.

And you wrote this great post in 2010.

I think Facebook had probably just gone public or was close to it.

And you basically said Facebook's a great success, 500 million people use it, but it's not worth $33 billion.

You laid out this case as to why that was not true.

Looking back on this now, I guess, is it— do you just look at that as I got it wrong?

Is it actually, I think the kernel of what I was saying was right, but, you know, I guess how do you think about this blog post now?

I love that blog post.

And it's funny because people often pull it forward to dunk, right?

Like, hey, look, 17 years ago you got a thing wrong.

And what I love even more is it's often VCs, right?

You're like, hey, wait a minute.

Isn't your business to be wrong 9 out of 10 times?

Wasn't your last fund just wrong on the vast majority of all the investments that you made?

Oh, okay.

Well, let's start from that premise, but I don't even care.

I don't even care.

I love that article because it's both a good example of, uh, being wrong and accepting that in terms of outcomes, yet trying to draw better conclusions than just like, I got it wrong

at the gambling table, right?

Like I should have bet on Facebook.

If I'd bought stock at that time rather than say they weren't worth it, I would've made so much money.

Right?

Yes.

And a great term that I learned from this, uh, uh, poker book is called resulting.

And in poker there's this concept of you analyze, well, not just analyze, you know what the odds are given the cards that you have, the cards on the table, what are the odds that you're

gonna make whatever hand you're trying to do, right?

And if you delude yourself into thinking, I should keep playing 16% hands, I should just go all in on 16% because, do you know what, I won the last 3—

you're an idiot and you're gonna lose all your money.

That's not how to win in poker.

The way to win in poker is to make sure you predominantly play hands that have the best odds just as much.

If you go all in on a hand that has 87% odds of success and you lose, you're an idiot if you draw the conclusion that you shouldn't have gone in on that hand.

Totally should have gone in on that hand.

So resulting is evaluating your decision on the basis of the outcome alone.

Now, obviously you can't divorce those two things, but you've got to do it on a longer trend line.

Everyone is gonna miss a shot.

Not just one shot, that's gonna miss 100 shots, 1,000 shots, a million shots over the course of their careers.

In fact, this is why the VC parallel is so important.

VCs have gotten tremendously rich being right 10% of the time.

I mean, isn't that— I mean, that's an incredible stat for me that you're able to build such wealth on being so wrong so often.

But let's even dig into the merits because I love the case of that too.

When I did that analysis on Facebook, they did not have a monetization strategy.

What they had was a lot of traffic and I pair, or, or I pattern matched that traffic on, I know this is too old, but there was something called Blue Mountain, I think back in 2000.

The e-cards website?

Yes.

Yes.

They were.

Tremendously popular at the time.

They had a ton of traffic, right?

And all the traffic brought them was ruin because they did not have a mechanism for turning that traffic into gold.

It was just trash.

It was trash traffic.

It was just people who wanted an e-card.

And what are you gonna do with that?

Maybe today you could make it work.

They certainly couldn't at the time.

My analysis was Facebook is trash traffic.

It's just a bunch of people talking about all sorts of shit that isn't intentional in the way that a Google search, for example, is.

At the time, the surveillance capitalism paradigm was probably already under, well under construction, but not widely distributed like that.

This was the alchemy that was gonna turn absolute trash traffic into gold because Surveillance capital flipped it on its head that it did not matter whether intent was there.

If we know everything about everyone at all times, all that matters is eyeball time, because now we can pinpoint whether an ad should be shown or not shown on the basis of who they

are, not what they're talking about.

So they could be talking about the most deranged, stupid stuff, and we can make better ads happen because we can target those individuals than a golf magazine online could do on golf

ads, right?

Like, I don't know if that person is currently having buyer's intent for buying a new set of clubs, but you know what, Facebook probably knows whether you're within 10 miles of the

golf store right now, you're driving towards, whether you could be swayed one way or the other.

So I didn't see that.

I did not realize the power of that alchemy.

And by the way, nobody else did either, because if they did, they would've gone like, holy shit, Facebook has invented alchemy.

Why are they only worth $33 billion?

They should be worth half a trillion.

Let's just go all in, bet the house.

No one did that, right?

So anyway, I love it for all these things.

I love it for resulting.

I love it for this example of alchemy.

When someone really does figure out like a complete unlock on the entire internet business in that right moment when they also have all this trash traffic that would otherwise have

been worthless.

And then I just love, you know what, I was wrong.

And I remind myself about that on like at least a yearly basis, sometimes more often than that when some VC needs to dunk and pulls out a 16-year-old blog post.

And I, I just sit back and smile.

Yes, you should have some humility about your predictions.

Can I tell you guys about a funny interaction I had with your business partner Jason?

So hey.com.

So if you guys are listening, go to hey.com.

That's one of my favorite landing pages.

You guys are wonderful at writing copy.

You're obviously great at design, but like copy, which I think is like the forgotten skill set in internet— internetting, copywriting is sort of like a lost art.

You guys have not lost it.

You and Jason both have it.

And I texted him and I go, hey, I'm working on a new landing page for my thing and I'm using hey.com as inspiration.

What's the conversion rate of your landing page?

Because it looks great, it sounds great.

Is it effective?

And he goes, you know what, I haven't really looked at it.

And I was like, what do you mean you don't look at the conversion rate?

You don't know if it's working?

He's like, no, I've never really thought about it.

I just know that like it felt good and I think it's good.

Therefore we just ran with it.

And I was I was in awe of that reply.

That was the opposite of what I would have expected.

Can you talk a little bit about the tension between good taste, which I think you have, and also following data because you have bills to pay and you want the— you want a good outcome

and you want to be great.

You know, you race race cars, data and like the weight of your car.

And I don't know anything about racing.

Precision matters, data matters.

And yet you seem to be a follower of taste first.

This is the luxury of margins.

We have had the luxury of margins since day one because we built something people liked and we did not balloon the company to follow that interest.

We grew the company as we needed, exactly to the point where we were not deadly frustrated by our lack of progress on the ideas that we had, which meant that right from the early days,

and through some period of times we had ridiculous margins.

I mean, truly, when I was telling in confidence others about our margins, they were like, no, no, you gotta mean gross, right?

And I'm like, no, that's net.

And they were like, what do you mean?

I've never heard of a business like that.

Are you able to give any, uh, frame or any ballpark for previous numbers if you're comfortable?

So we understand.

Not really.

I mean, let's just say ridiculous.

I think that's a, that's a good ballpark, right?

And certainly if you compare to any of the public SaaS companies, like, as we were in a different industry, right?

Like, often industries have a tendency to coalesce around a certain profit margin.

And it seems like public market SaaS has coalesced around like, I don't know, -10% or something, -20%.

Ridiculously long stretches of un— uh, unprofitable growth, partly because there's all these other hacks of how you get money outta the business while leaving nothing for the shareholders,

like stock-based compensation and so forth.

But that's a different discussion.

Our focus from the beginning was the more margin we have, the more freedom we have, the more freedom we have, the more we could just focus our time on what we like to do.

And you know what?

It just so happens to be the night that Jason or I love running minutiae A/B tests to squeeze out 0.01.

And that's very often what's needed if you wanna tweak something.

You run 1,000 of these experiments, right?

Like not 2, you just, you stick to it and you keep grinding away.

And I think grind is actually a good word here because this is something I often hear entrepreneurs talk with such pride in their voice.

Yeah, I, I grind.

Like, this is why I'm working 100 hours a week or 120 hours a week.

I'm just grinding, grinding, grinding.

I'm like, okay, I don't want to do that.

My life is far too interesting to waste it grinding.

Grinding is the stupid shit you do in World of Warcraft when you're peon.

No, thank you.

I don't want to do that.

So can we just get to a place where our intuition affords us the success of our intuition affords us to just do whatever the hell we want whenever the hell we want it, that'd be a great

place to be.

And we arrived at that place very early on in the life cycle of the company.

But I will also say what's interesting about this is that I like numbers.

I actually like statistical analysis.

I like rigor.

I like the confidence intervals, not to the point that I wanna run these A/B tests myself, but to the point that like I have great respect for the statistical tradition.

And we for many years had essentially a data scientist sort of on staff, someone who would crunch those numbers, do those reports, try to eke it out.

And after trying that for over a decade, Jason and I finally came to the conclusion in honesty that we never did what the numbers told us to do.

What we would do was we would do whatever the hell we wanted to do, and then if the numbers supported that, we'd go like, Those are good numbers.

And if the numbers didn't support that, we'd go like, yeah, I don't know, there's probably some factor you haven't calculated in.

And then we just thought at the end, why are we doing this?

Well, partly why we're doing this, because we listened a little too much to other people telling us this is how you're supposed to do things.

Now, I am not saying that this doesn't work for some definition of work.

I'm saying we didn't need it.

And that is a much nicer place to be.

And if you are able to get there, you should embrace it with Full gusto.

If you have the margin to not waste the precious hours you have allotted on this earth doing shit that would bore you mindless, don't.

Like,

we never had the obligation to squeeze the lemon for its last drop.

We never had the investors going like, hey, why is this quarter 14.79%?

You promised me 16.25%.

Go get the last 2 percentage points for me.

We never had that.

We never had that pressure.

And therefore we got to act reckless in some definition of

prudent business operations.

And I'm immensely proud of that recklessness.

First of all, because we have outlasted about 7 generations of competitors and others in the SaaS space.

Who did all of that, right?

Who did all the statistical analysis, who, who squeezed the lemon in all the ways, who

debased their copy because for a hot second it converted slightly better and we didn't and we're still around.

So part of what I take away from that is A, what a nice run.

B, maybe all the statistical analysis.

Is a bit of hocus pocus.

And the reason I say it is not because the numbers aren't right or the math doesn't math, but because very often the missing ingredients to getting to the capital T truth is unknown.

When you keep drilling down on a set of data that is highly limited and does not describe the world as it exists, you're gonna steer yourself blind on that.

You're gonna think you found the truth.

And you're not hearing from all the customers who's not signing up.

You're not hearing from the word of mouth that takes 18 months to materialize because those things can't get computed.

But isn't that an absolute— I mean, I think your way works wonderfully and I identify with it and I wanna copy you, but you're on the board of Shopify.

Uh, they probably break so many, uh, dogmas or rules that you believe in and it's worked out really well.

Totally.

This is the great thing.

Um, ��ižek, is this Polish or Czech?

I think philosopher has this great monologue about how he hates wisdom.

I'm like, how can anyone hate wisdom?

And the reason he says that is for every wise axiom, you can find another that says the opposite that also sounds wise.

And both of them were extracted from moments of success for that individual in those circumstances.

And that's a bit of a deflating thing because it could very easily sort of descend into nothing matters, nothing is true, and everything is just postmodern.

I don't believe those things at all, but I do believe that wisdom is very contextual.

This is all the way back to Newton and, uh, uh, Einstein and the theory of relativity.

When you go fast enough, like all the neat physic rules that are really helpful for calculating the apple dropping on your head.

They just start breaking and suddenly time and space and everything is relative.

So you gotta know your universe.

You gotta focus on like your set of circumstances and what works in your context and what works for a startup and a small company selling to other small companies is not gonna be the

same as what works for a Shopify.

What, what, what's the most interesting thing you've learned from being in the Shopify business, or Tobi, you know, if I just said, what are the biggest kind of like lessons, takeaways

that anytime you let a smart person go experience kind of a new novel environment, I'm always curious, like, what's the one or two things that stand out to you?

Good question.

Some of it is just awe.

Like Shopify last year grew almost 30% year over year.

I'm like, I would have been happy to go 30% year over year.

Like, I don't know, very early on in our lifecycle.

They're doing it from a base that's just unfathomable.

Like the law of big numbers gets kind of wacky, right?

When you're already doing billions and billions and you're growing at that rate, it just, it feels like you found a glitch in the matrix somehow.

And what I love even more about it is you can come up with all sorts of reasons for why that is, but you know what, even at the board, even Toby.

We don't really know.

Like, why did it take Shopify— what is it gonna be, 22 years— to get to this moment where it just had its best year since the craziest of the peak COVID years?

Isn't that fascinating?

Like, in many ways, the company is doing what it's done since Tobi created Snowdevil.

I just want a toolkit for me to create a great-looking store rather than be in the, uh, bazaar of Amazon, right?

Like, pretty simple mission.

And they've just been chipping away at that for like 20-plus years.

Suddenly the charts just go like, dong dong dong, wow!

And all these theories are just that.

Like, we don't actually know.

The amount of stuff that people don't know is amazing.

I mean, I wrote a blog post a while back basically putting that to the point of nobody knows anything.

And I think that's a good general take on the big questions.

Nobody knows anything, whether we're talking about climate change or what AI's gonna look like in 2 years, or why Shopify's able to grow 30% every year in

year 22.

Uh, no one knows anything.

We can have all these theories.

Very, very little of it gets to be reduced to irrefutable proof in the sense of a math equation.

We could talk about this board of directors of Shopify forever.

It's so fascinating.

On, on Tobi's bio for board of directors, the last sentences, he says that he was on the core team for Ruby on Rails.

So I assume that you've known him for close to 25 years.

Is there any cool stories that you've had about him or leadership lessons that we can learn?

Because it's very rare.

I mean, building— he's built like the 17th largest company in the world.

That's rare.

It's also very rare where he's been at the helm the whole time.

Uh, is there anything, anything cool about like the evolution of like a computer nerd to being like a proper CEO of a 10,000-person company?

Is there anything cool that you can, I can learn about that?

What's funny to me is that you posit that as he went from one form to another form.

Not true at all.

He's still a computer nerd.

Absolutely.

As much as I'm a computer nerd, actually, he's more of a computer nerd than I am most days.

The, direction of travel of him sending me cool tech stuff and me sending him cool tech stuff, I, I think it's in Toby's favor.

So he's still that person.

Now, obviously what he is also is now the head of this giant corporation that is responsible for, I think I saw something like 15% of, I mean, world e-commerce, a notable share of world

GDP.

I mean, that's mind-blowing in a way that I'm really happy that Toby at least doesn't give any appearance of internalizing into his character composition.

He's very much still the computer nerd.

And I think this current phase that we're in with the AI and the agent explosion just demonstrates that to a T.

He was incredibly early on this stuff.

Some of the public memos that are out from Shopify, you can check their dates and then you can check Topi's predictions.

And I just got to give immense respect to that because I was not that early.

In fact, Topi helped drag me into the light on some of this.

And some of it was also just like, I had to see with my own eyes the agents do what they're now capable of doing before I fully flipped on it.

But he was just really early on it.

And so much of that came from that nerd perspective.

Most— we've had a lot of really amazing people on here, and like the like ballers, like Dharmesh from HubSpot, they will reference Toby as like he's the guy of doing that, of like being

able to see a little bit into the future and behaving in such a way that is— he's just right a lot.

Totally.

And I mean, some of it is because he's really smart.

Some of it is because he's got that nerdy

conviction.

I think similar to what I have, I'd like to believe that we're similar in many ways, and that's why we get on so well.

We've known each other for 20-plus years.

Is that a little bit of isolation, maybe because of that nerd upbringing that you have the whole world telling you X and you actually go like, eh, let me just check under the COVID

Oh, do you know what it says?

Y.

I'm gonna bet on Y even if the whole world is in on X.

And to see that continuity, I should say, is incredibly rare.

And I think part of that is the answer to why now with Shopify?

Yeah.

Do you know what?

Compounding determinism over 20-some-plus years

has a tendency to add up.

And it adds up in irregular ways where it sometimes takes quite a long time before boop, the inflection point comes.

Now, I mean, we say this, but Shopify went public, I mean, over a decade ago at a $1 billion valuation.

I mean, that's a Hall of Fame.

Early investment you could have made because Shopify didn't do as so many other companies do now, which is essentially reserve all the upside to insiders and their VCs and then go public

when it's already kind of plateaued, right?

Like from where they were to where they are now, an amazing ride for the public investor if they were along for it.

But yeah, I mean, tremendous respect.

It's funny because I mean, I just joined the board, uh, a year and a half ago.

And it came out of a conversation actually where I was making the argument to Toby that boards are bullshit and that all the boards I've ever been on were boring snooze-fests dealing

with minutia to rubber stamp things for auditors and whatnot.

And he was like, ah, I don't know, I think we have a pretty good board.

And I was like, uh, okay.

And we started talking and then we had talked for a few hours or 5 and then suddenly I did come to realization, do you know what?

Yeah, I've probably just been bored because I've been dealing with things at a scale too similar to my own and I'm not learning anything.

Shopify is dealing at a universe apart in terms of scale while still using some of the same things that I'm passionate about.

Shopify is the biggest Ruby on Rails shop in the world.

Obviously I'm very passionate about that.

Um, many of the other aspects of it has great overlap, but I don't know that scale.

I'm like, that sounds like a fun challenge.

Let's just do it.

And of course Toby was right.

Like, is there bullshit on a board of that scale?

Of course there is.

But it's actually a small minority, and most of it is just really interesting stuff dealing with problems I would never see day to day in my own business.

You mentioned with AI, you're like, Toby kind of dragged me into it a little bit.

What's been your evolution with AI?

So, you know, a couple of years ago, what was your opinion?

What were you playing with?

And then now what do you see?

I thought it was amazing since day one and been a ChatGPT user since the first days of that interface and just loved that we were able to make computers do this right from the get-go.

I really like computers.

I just like them for their own sake, not even always for productive purposes.

I just like computers.

Some of my favorite time is just sitting in front of the computer doing computer stuff, learning computer things, even if it's not directional towards a specific goal.

And this, even from the get-go, even when ChatGPT was dumb as rocks and couldn't count the Rs in strawberries, was just fascinating.

I'm like, I still don't get it.

I mean, now I look into a lot of it and how Transformers works and whatever, back and forth and temperature and all of it.

I'm like, I still don't fully get it.

Like, how is it?

Yes, I know next token, but how does it turn into this?

How does it cosplay so well as an intelligent being?

Maybe the reason I'm surprised is because I have the wrong model of human consciousness.

Maybe human consciousness is far closer to token prediction than any of us would've cared to admit or even investigate seriously before.

And

the models as we have them, as they came to be, were quite close to that.

And we have discovered intelligence in some sense.

But the timeline is still that from the get-go, awesome, love it.

It's not changing my day job.

And the reason I say that is I'm very particular about how code looks and what it feels like.

And, um, I was about to say correctness.

That's not even like at the top part of it.

I'm like, I'll get to correctness when I love the aesthetics.

And on aesthetics alone, the models were just not that good for a really long time.

That's one part of it.

The other part of it was the early 8, not even agents because they weren't agents, they weren't doing tools.

The early LLMs, the early AI was autocomplete, right?

You're in your editor and then it tries to complete the rest of it.

And I hated that.

It was like having someone interrupt my train of thought every 5 seconds to usually be wrong.

And I say wrong, not in a factually correct way of does this program execute, but wrong in the sense of like, this is not the code I wanna write and I don't like what you're producing.

So I didn't like that form of using AI and therefore didn't really use it that much for quite a long time.

I used it as a pair programmer, as a better Google, as a better Stack Overflow, as a here I have a problem, can you have a look at it?

Oh, you taught me what the solution was.

Let me go off and write it.

And this takes us all the way up to the interview I did with Lex Fridman in the summer of last year, where I was essentially giving this spiel, right?

Like, hey, I don't like what it produces.

I'm extremely impressed.

You'd be a fool not to take this super seriously and as a directional path for where we're going.

But it's not yet.

It's not in my business yet.

And then I even, I think early December, I did another interview.

Had some revisions on it, and then from there it was like those realms could have been a decade apart because the models that dropped in late November— I think Opus 4.5 was November

27th or something— that class of intelligence suddenly switched from, I don't like what it's making, to holy shit, what?

Wow.

Okay, I'm going to merge.

So the last 3 months, I'd say, has been the most churn in my mental approach to computers in the entire time I've been using them.

More deep questions and more workflow changes have entered my personal life, my day-to-day life, my 8 hours in front of that little screen.

Than in any time before that.

And now I'm probably and thoroughly pilled.

And I don't necessarily say that with like, because that also means like I'm actually now in the herd, right?

Like the herd certainly out of Silicon Valley is like AGI is imminent.

And even between now and AGI, we're going to go through such an intelligence explosion that hold on to your socks, right?

Yeah.

I'm actually now in the consensus on that, which I'm always a little skeptical about.

I always want to try to look for the other angle, but it's just undeniable.

At some point you just gotta admit that the math maths and that the agents are incredibly capable programmers.

I thought when I talked to Lex last summer that it was gonna turn everyone into a project manager of the old type, the kind of project manager who's not really doing the things, just

telling other people what to do.

And then you got this little churn and then you get the feedback.

And I realized that's not true at all.

It is more like I've grown 18 arms and 7 more brains and I'm operating 22 screens and I'm still doing all of it, even if I'm not typing every character myself.

And my progress when the stars are just right is just unbelievable and incredibly addictive.

And that's

in some sense a little bit like that Facebook thing.

Like, do you know what?

I was wrong about how quickly this was gonna radically change everything.

Like I thought it was gonna happen.

Um, I thought like we gotta have a few years, right?

And then suddenly from December onward, we live in a completely different world.

I have to thank you for that Lex Fridman podcast.

A, great podcast, 6 hours long.

B, I bought 3 ads with Lex Fridman.

The first for my company, Hampton, the first ad went on like he doesn't tell you what it's going to go on.

The first ad went on like Iran war thing and I was like, that was not my target audience.

That's not going to work.

The second one was like a Jeffrey Epstein video or I don't know, it was like some country.

And then the third one was you and I was like, yes, got it.

And so your, your episode actually made our ad profitable.

So thank you for that.

There we go.

Are you, you know, so I read your work, I've read your books.

I think it's great.

I,

copy a lot of your stuff.

I'm inspired by your stuff because I want to build a company like you.

What about what you guys have built today do you think is foolish?

Or what have you built over the last 20 years that a lot of people admire, a lot of entrepreneurs and business owners admire, that you think is silly to replicate today?

And you would like advise someone who has like a $2 or $3 million company, if you want to be a bootstrapped $50 million or whatever you guys are company a year and be as profitable

as we are and have this life, don't actually copy what you see now because that won't work today?

I don't have a straight answer to that.

And partly it is because I can see some of the things we've historically been doing work less well than they had.

And some of it is exactly this, that being able to generate attention is not as transferable as to business outcomes as it used to be.

And there's a great, um, Gary Vee has this book, uh, Jab, Jab, Right Hook.

Love the book.

Love the concept.

In Basis is give, give, give, jab, jab, jab 100 times.

Just give, be kind, go out there, share all this stuff, and then occasionally you give it a right hook.

You call to action, buy my stuff, right?

And that's what pays the 100 jabs is that occasionally there's a right hook that lands.

What I've seen, and some of it is just a reflection of the dynamics of social media, is that the right hook doesn't travel anymore because all the main media outlets or social media

outlets, X in particular, where both Jason and I are strong, the algorithm is just never going to show your stuff.

It used to be that having a ton of followers meant that a ton of people just saw your stuff.

Now, the great irony here is that this is what people complained about on Facebook back in like 2012.

I have all these followers, but I can't reach them unless I buy ads.

Yeah, that's the demise of BuzzFeed and all these huge companies.

Exactly.

That's not actually quite what's happening now because from what I can tell, ads on X still don't really work.

We tried them a bit.

Maybe they work better now.

Never worked for us.

What works?

Is viral, right?

Like what works is sharing really interesting stuff.

I don't mean viral actually in a negative connotation.

It also has that, but it also is something positive.

You sharing something novel with people who are interested and they engage with it.

But the trickle down has kind of stopped.

So that trickle down of like, I share a lot of interesting stuff on a daily basis, and then occasionally I'll ask you to check out my new product or something else.

Then to check out my new product, like that stopped working.

So I'm not sure our historic long run strategy of build a large audience and then good things will ha— happen is necessarily true anymore.

I think there's still something to it, but is it viable as your sole strategy being someone new?

I don't think it converts or works as well as it once did.

So that's the thing that does mean work.

Now, I still wouldn't say that.

I still think just for your own edification, you should try to share interesting, interesting things.

That's actually how you learn more.

When you try to teach what you've internalized, you make it sink in deeper.

And I also think it's just good to share for the sake of it, regardless of the outcomes.

But I can also see what happens to people who share, share, share and never see something back.

The odds of that person ending up bitter is pretty high.

One of the things that both me and Sam love is, uh, anybody who sort of plays their own game by their own rules.

And I think whenever we had Jason on or you on, like, that's one of the things that we like and respect the most.

In fact, I almost joined 37signals when I was very young.

I was applying to two jobs.

One was in San Francisco and like the tech world, and the other one was going to be with 37signals.

I went to San Francisco instead.

Did you get the interview, Sean?

Um, I did get the interview, but I got the other job first.

And so I never even took, I never even, uh, went forward with it.

That's awesome.

But it was a company I was studying.

It was like, if I'm going to join one company, I like these guys.

I like how they roll.

I like that they have their own, like, the drumbeat, right?

And so you guys have for a long time just done things your way.

I think both, you know, for you, I think you're kind of like an engineering hero for a lot of me and Sam are not developers, but like, I think a lot of developers respect you because

you, you have a strong opinion and a way you roll technically.

I think there's the 37signals part of things.

And then there's the outside of it, right?

Like doing the races and just living life a certain way.

We talked to Jason about his houses and how he's like a craftsman and he's got taste and things like that.

I'm curious, who kind of inspired you?

So who are people that you kind of liked something in their blueprint, either how they ran their company or how they had, you know, this, these side quests outside of their, their work

that, that was cool, that, you know, maybe made you want to do the races and, and, and you know, dedicating that set.

Is there anybody that you've looked up to that you, you've stolen a little bit from their blueprints?

Completely.

I mean, in fact, my entire early career was just finding people who I really liked and looked up to and said, I'm just gonna copy the motions.

Like, I don't even know if I fully understand it, but if I just move my hands in the same way and dance the same dance, maybe I'm gonna get a little bit of that magic.

One of the ones that I think maybe Jason has cited as well is Ricardo Semler.

Who wrote a book called Maverick.

I think it was published in the '90s.

Oh, was this the Brazilian guy?

Yes, I, I've seen that book title.

Can you tell us about— like, I, I know a couple— yeah, what's his story?

Amazing book, amazing story.

So Ricardo Semmler, I think, inherits or takes over this huge industrial

Brazilian company that makes pumps for, I think, oil tankers or something.

Like, the last company you would imagine being innovative on work practices and so forth.

And he just approaches that company with an irreverence to how things are done that would make both Jason or I blush.

And we try to be relatively radical, I think, in what we try to prescribe.

But he really gave us, like, certainly gave me permission to think very different ways about how to do things.

He has a, one of the examples I remember is like how to look at employees and the value that they bring over a much longer period of time.

And it gives the example of this one guy he has that on almost every week sits at his desk with a newspaper open and does nothing.

He's just there.

And you're like, why would you pay a guy to sit at his desk and read your newspaper?

Well, it just so happens to be that occasionally some of these pumps fail in really spectacular ways in far-flung places, and he's the guy having that capacity to get some oil tanker

out of some predicament immediately.

One of those will pay for his salary 7 times over.

Now, I love that anecdote because it's one I've struggled to implement at 37signals, which is the best kind, right?

I've been practicing this for literally 2 decades.

Of divorcing this sense of, I don't know, Protestant work ethic of like, why are you not moving your hands 40 hours a week with— that's not the value this individual brings to the corporation.

So I love that he has a million suggestions on how to run a company.

We took a fair bit of inspiration for both Getting Real and for Rework.

Not quite the form.

He tells more of a story.

Narrative, which is great.

Huge fan of him.

Another one is, um, Kent Beck.

So Kent Beck is the author of Extreme Programming, one of the heretical software methodologies back in the '90s when everyone was stuck on Rational or the Waterfall method, big upfront

designs, like large manuals before you write a piece of code.

Kent Beck was one of the absolute pioneers on a completely different way of working, the agile way of working.

Now that word has stopped meaning almost anything because today it just means software development.

No one is gonna tell you they're not agile except for the really edgy people who've now started to turn towards agile, which is fun to watch.

But the mainstream, agile got mainstream.

It wasn't in '99 or early 2000s, and I saw Kent Beck on stage in 2001 at a Danish conference have complete and utter command of the room as he delivered his sermon on software methodology.

And I was just in awe.

By the way, his LinkedIn, his LinkedIn is hilarious.

His bio says, of the handful of people who have shaped how we build software, Kent is the raised middle finger.

And he's giving a, it looks like, it looks like he's giving a university graduation speech and he's wearing a tie-dyed Oxford button-down shirt.

Amazing.

I love Kent and everything he's contributed to software development, both in terms of his methodology, but also in terms of his craft.

My favorite book on the nitty-gritties of programming is a small book he wrote in the '90s called Smalltalk Best Practices.

It's a quite short book and it is the most influential book on how I write software that I've ever read.

Now, it's funny because I've mentioned this book many times over my career, and at one point I think it was not available yet as an ebook.

You could only find the print versions and managed to create so much demand that I think at one point it went for $4,500 for print edition.

Now it is It's available online.

You can just buy the PDF.

It still totally holds up in this day and age from a language most people are not programming in.

Smalltalk is incredible.

And Kent wrote not just that, like a craftsman's book about craftsmanship, but then also reshaped how all of software methodology is done.

And then finally, I'd say, uh, Matz, who created Ruby back in the '90s.

And opened my eyes to rearranging the priority list.

When I got into programming, most programming languages were sold on like, well, I'm the fastest, I'm the most correct, I have the least memory usage, I'm the most like C, I'm the most

safe.

All those adjectives.

And I think, yeah, yeah, yeah, that makes sense.

I should pick the fastest, the least amount of errors.

And then Matz comes in and says like, no, no, Ruby is designed to make the programmer feel good.

Like, what?

Wait, what did you just say?

It's about programmer happiness?

What hippie dippy bullshit is this?

How are we gonna get create or correct programs that makes us lots of money if we're focused on programmer happiness?

That was a real mind-blown explosion.

And it was also the inflection point of my professional career as a programmer.

I'd been using a bunch of other things before as tools, looking at like what's the most efficient and how do I get the fastest?

And then suddenly I discover Ruby and realize that writing it can be this intensely satisfying experience, just getting it beautiful.

We talked a little bit about those aesthetics earlier and why I didn't like AI autocompletion for so long.

Was because it interfered with that massaging of the Ruby poem.

And I've now spent well over 20 years just absolutely loving the Ruby programming language because some guy in Japan rearranged the priorities of what was permissible to care about

as a software developer.

We have this thing, um, David, uh, I'm now going to call it the MFM Lean.

Sean, do you want to show them the MFM Lean?

You had us both.

We are— whatever.

We sit like this and we're just like, you have got us to fall down.

If you're listening on Spotify, you got to go on YouTube and hear this part.

But if you've gotten us to fall down your slippery slope,

you, you made a comment about longevity, which is pretty cool.

You talked about, you know, all these people tried all these tactics.

It didn't work.

But, you know, we didn't— we weren't too trendy.

And now we're sitting here and we still have lasted.

And you've built this amazing company that has

made,

I don't know how much, hundreds of millions, billions in cumulative revenue.

Is there anything that you do on a day-to-day basis within your company where you are making decisions to last 5 or 10 years out or being, you know, like longevity decisions versus

short-term decisions?

Like, do you actively think about that on a regular basis?

Or maybe I could even ask you the other way around, whichever is easier.

What decisions would you make differently if you were just trying to like, you know, get in and get out?

Good question.

I think on the longevity front, the number one objective, both Jason and I, is to ensure that 37signals is still a place we want to work.

Not just want to work, but like to work, is our favorite place to work.

I see and talk to a lot of entrepreneurs who are in a hurry because they managed to build a company they don't like.

And they would love to get out of it.

And I find that tremendously sad, not because they can't achieve other objectives in their life.

Like maybe they have a tremendous exit and then they

grit their teeth for the 2 to 5 year exit period when they're on holdup and earnout, and then they go on living merry good lives.

Great, lovely.

I am glad for you.

That's not what I want.