Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
Welcome to rework a podcast by 37 signals about
0:02
the better way to work and run your business
0:05
we recently recorded rework live where
0:07
we took our conversation to youtube
0:09
to share some current updates from
0:12
37 signals and answer some listener
0:14
questions live here's that recording enjoy
0:18
okay i think that we are live
0:20
in the youtube i'm kimberly the host
0:22
of the rework podcast i'm
0:24
joined by jason creed and david heinemann
0:26
hansen co-founder of 37 signals i think
0:28
that we're live on youtube if
0:30
we are and you can hear us and see
0:33
us let us know in the chat thanks for
0:35
being here we know we have lots of people
0:37
joining us live we're trying something new we've never
0:39
done this before we usually record the podcast and
0:41
then push it out later but we
0:43
thought why not just hop on live and answer
0:45
some live questions we know that you guys are
0:48
here to hear from jason david so i won't
0:50
spend a lot of time talking we'll get right
0:52
to them and i think first we'll start with
0:55
a question we already saw pop up on twitter which
0:57
is you guys just launched the hey calendar what about
1:00
once when are we gonna hear about
1:02
once jason do you want to take that one yeah
1:04
i'll start i mean we're late so uh
1:06
we made a we
1:08
made a public promise essentially that we
1:10
would have this done or have the first
1:12
version first product from the once family out
1:14
by the end of 2023 now this
1:18
is more of a patting us on the back but not
1:20
really true it's true but not completely
1:23
we did release it to some customers early
1:25
at the end of last year so some
1:28
people have it there's like i think six
1:30
or something like that or eight so it is
1:32
out but not really um it's
1:35
a good example frankly of of never you
1:37
know never making public promises because you
1:39
just shoot yourself in the foot more often than
1:41
not like with the hey calendar we launched january
1:43
2nd although with the apple thing we had to
1:46
push that off a little bit but like we
1:48
were ready to go on january 2nd as we
1:50
said we would be all the way back in
1:52
march of 2023 we predicted we'd
1:54
be ready and we were but we didn't promise
1:56
that uh to the public we just
1:58
said early to And it turned
2:02
out to be true. We didn't make a public promise. It was
2:04
specific with once we did and we didn't hit it. And so
2:06
it feels kind of bad. That said,
2:09
we are very close. In fact, today
2:12
or tomorrow actually gonna launch to a
2:14
significantly larger batch of people. It
2:17
will not be public for everybody, but
2:20
it'll be, let's say wider beta and people will be able
2:22
to pay for it. The
2:24
first part again is gonna be, it's
2:27
called Campfire. It's a group chat tool,
2:29
similar to Slack or Teams, but significantly
2:31
simpler. Kind of the
2:33
way these tools used to be before
2:35
they became bloated, complicated, convoluted, massive
2:38
messes in a lot of ways. So
2:41
I expect a week or two of that
2:43
beta stuff happening before it's probably out for
2:45
wider release. So anyway, just
2:47
always a reminder that, and by
2:50
the way, the reason why it's bad to announce
2:52
release dates publicly is because
2:54
they're always far in the future and they always
2:56
sound easy to do. It's
2:58
very easy to say yes about something later and
3:01
we have to continue to learn our lesson and we learned it again
3:03
here. So I don't know. David, you wanna add anything on that? Yeah,
3:06
I think we've made this mistake several times over
3:08
the last 20 years. And every single time we
3:10
make the mistake, we say never again, never again.
3:12
And then a couple of years pass by, I'm
3:14
like, this seems so easily doable to hit this
3:16
target. I think one of the
3:18
things that added some complication here was we were trying
3:20
to do two launches at once. We have never ever
3:23
attempted anything as foolish as that. You are
3:25
a small company. So we had Hey Calendar,
3:28
which is a major new product that we've
3:30
worked on for a whole year happening
3:33
at the same time because we're a bigger, more
3:35
capable company as we were developing once. And it
3:37
felt really exciting. Hey, we have two separate teams.
3:39
We can do both things at the same time.
3:42
But there's just a gravity
3:44
of attention that goes into any
3:46
major public launch. And like
3:48
Jason and I can't split ourselves. The rest of
3:50
the team can't split ourselves when it comes to
3:52
a launch. So when we pushed out Hey Calendar,
3:55
even though I've mostly been working on the once
3:58
side of the product fence for a while. I
4:00
couldn't just sit over there and wave, oh,
4:03
looks good, have fun. I mean, the Apple
4:05
thing added a fair amount of complication to
4:07
it, and I got drawn into that. But
4:09
even if that hadn't been the case, a
4:11
launch for a company
4:13
of our size is always going to be
4:16
an all-encompassing event. And we had
4:18
put both of these expectations a little bit on
4:20
top of each other. And then
4:22
the other part was that one actually influenced
4:24
the other. So already
4:27
towards the end of the year, we
4:29
were starting to get excited about this
4:31
PWA concept, Progressive Web Applications. We were
4:34
not going to build native applications for
4:36
once. First, that started out actually being,
4:38
do you know what, our native
4:41
teams are fully engaged with, hey, calendar, hey, calendar is
4:43
the most important thing we're working on right now. We
4:45
can't also do native app for once. Then it turned
4:47
into a little bit of like, do you know what,
4:49
this is actually a little bit of a feature. Once
4:52
it's an installable product, you install it by
4:54
yourself. If we have to have native applications,
4:56
we have to also run servers, which kind
4:58
of becomes a service then. The whole point
5:00
of once is that there's no service. This
5:03
is a set of products. You install them
5:05
yourself. You operate them yourself. This is why
5:07
I think people will be pleasantly surprised when
5:09
we announce pricing, that it's not
5:11
as high as I think most people think,
5:14
because we're conceiving of this as products. As
5:16
boxes on a shelf, you take it off
5:18
the shelf and it's mainly on
5:20
you. Like there'll be a 1-800 number. I
5:23
mean, not really, it's an email. But imagine
5:25
on 1-800 number at the back of some
5:27
envelope. Like how many times did you call
5:30
Sony when you bought a Walkman? I
5:32
don't know anyone who ever called Sony about buying a Walkman, because
5:34
that was a product. And you were just like, that's on
5:36
you and you got to do it. So all
5:39
of these things came together. And then
5:41
of course, the accelerant with Apple being
5:43
just absolutely vindictive SOBs
5:46
when it came to our own
5:48
native apps. And now this Epic
5:51
Games verdict just really lit
5:53
a fire under the intention
5:55
of pushing PWAs as far as we
5:58
could. So it kind of raised the question. We
6:00
were trying to hit a bit. Then we thought
6:02
at first, you know what, let's just push this out. It's going to
6:05
be fine. And now we're like, yeah, we'd like
6:07
it to be a little better than fine. We'd
6:09
like it to be quite good, the
6:11
PWA story, because
6:14
we have to shine a light
6:16
on an alternative path for all
6:18
this App Store gatekeeper nonsense. And
6:21
no, PWAs are not going to be as good
6:23
in all the ways, but I
6:25
also don't feel like they've fully gotten a shot,
6:28
right? PWA has a concept, this idea that we
6:30
could just use web apps, especially on mobile. That's
6:32
really where the whole question is. That
6:35
that feels unresolved, even though the term
6:37
has been around for five years, because
6:39
it wasn't mainstream viable before. Now we
6:41
have Apple actually supporting PWAs. I mean,
6:44
trying to deduce like why and how
6:46
much and what they'll change in the
6:48
future. You can go crazy, but you
6:50
can also just accept what's there right
6:52
now. You can do a web
6:54
app that you can install on your phone.
6:57
It gets push notifications and it's actually kind
6:59
of quite good. Perfect. No,
7:02
just as good as peak fidelity on a
7:04
native app. No, but for a lot of
7:06
apps, including this
7:09
campfire product, it's quite good. And
7:11
we'd like to put our best foot forward when it comes
7:13
to this. We're also trying to validate sort of
7:16
a category on two different levels, right? There's
7:18
the PWAs. Want
7:20
to validate that that's a viable path and
7:22
hopefully inspire others to invest in optimizing
7:25
that as much as we can. And
7:28
then we're also trying to validate this idea
7:30
of installable commercial software again.
7:33
That used to be a thing 20 years ago,
7:35
you'd buy a CD, whatever, you'd set it up
7:38
on your own machine, then it fell
7:40
out of favor because of SaaS. Although
7:43
it kind of continued all the long
7:45
alongside, like what does it have? The
7:47
internet runs on WordPress. Every
7:50
one of those installations was an installable app
7:52
kind of thing, but that's open source. We're
7:55
trying to do commercial software in this
7:57
form. So yeah, we just took a little.
8:00
longer to get it just right and I think we're
8:02
almost there. Okay, let me ask a quick
8:04
question and then also if you guys have questions put them in
8:06
the chat. I can see them. I will make sure that we
8:08
get Jason and David to answer them. David and Jason, tell
8:11
us with the ONCE products you said we're starting with Campfire.
8:13
Why do you guys pick that one as the first one
8:15
to start with? So one of
8:17
the premises behind ONCE is that besides
8:19
it being installable software and not subscription
8:21
based is to look at categories of
8:23
products that exist that are
8:26
widely adopted, that everybody already knows
8:28
how to use but are
8:30
still being priced like luxury items. And
8:33
group chat is something that has permeated
8:35
pretty much every business or every organization.
8:38
Yeah, some of these tools do some more things
8:41
but fundamentally the core of it is like I
8:43
need to talk to my co-workers in real time
8:45
by chatting, right? And
8:48
the thing is that this has been around now
8:50
for 15 years. It
8:52
will actually longer if you go back to IRC or whatever
8:54
but it's been around for a long time yet
8:57
it's still priced as if it's brand new. And
8:59
so we asked in fact on Twitter how much
9:01
do you pay for Slack and the bills
9:04
that we're coming back were like thousands of
9:06
months, some people tens of thousands a month,
9:08
even just hundreds a month, even
9:10
a hundred a month. And we're like that's
9:13
just obscene. It's obscene to spend that
9:15
kind of money on essentially a
9:18
tool to chat with your co-workers when
9:20
things like WhatsApp or Signal are free.
9:23
Like what is it about Slack
9:25
or Teams that makes it worth thousands a month? So we're
9:27
like, okay, we know how to build this. We actually built
9:29
this way back in 2006. It was
9:31
actually called Campfire originally. We know how to make
9:33
a simple thing. We haven't already in Basecamp. This
9:37
would be a perfect category where we don't have
9:39
to explain how to use it. It's a known
9:41
entity. You can actually build a
9:43
UI with almost no words in it. It's that
9:45
clear. It's like transcript box to bottom where
9:48
you type list of rooms or channels on
9:50
the side, very well known. And
9:52
we can price this and it can be hosted
9:54
on someone else's box. It
9:56
doesn't need to require email or any
9:59
complicated interactions. with outside systems
10:01
essentially. It's very self-contained and
10:04
it can be radically cheaper and
10:06
just as good at the 80% level.
10:09
So we're not going to have video conferencing in it.
10:11
It's not going to have audio conferencing in it for
10:13
V1 or any of that stuff. It's just basic simple
10:15
group chat. But it just felt
10:17
like the right thing to do, the right
10:19
category to start in and the right place
10:21
to prove this idea. And so that's essentially
10:24
why we chose that. There's
10:26
just so many angles of this too that are
10:29
exciting to me. Not only this idea that we're
10:31
taking something old that was good, like
10:34
the installable software and we're bringing it
10:36
into the present, I was going to
10:38
say the future, but it really is
10:40
the present. This fact that the fundamental
10:42
of change. There's a reason why everyone
10:44
moved to SaaS in the early mid
10:46
2000s. Because it was such a pain
10:48
in the ass to install and update
10:50
and keep in sync if you
10:53
were running your own servers. That was just
10:56
very poor ergonomics. Now the
10:58
fundamental underlying technologies have improved leaps and
11:00
bounds. This whole thing of containerization where
11:02
you can take an entire system, not
11:04
just a piece of software, but the
11:06
entire system, put it in a
11:08
box that is turnkey. You
11:11
just turn the thing on and then it
11:13
just runs as it was designed. It's very
11:15
different from how it used to be to
11:17
distribute server-based software. You'd have to worry about
11:19
what machine are you installing it on? Did
11:21
you have the right things? Can you get
11:23
it wet? Can you get it out of
11:25
shape? All of those questions have
11:28
largely been solved. It's
11:30
one of those things that the fundamental
11:32
improvement or advancement with containerization, I mean
11:34
at this point it's like whatever, 10
11:36
years old. We haven't
11:38
fully digested what the possibilities are.
11:40
I just find those things so
11:42
fascinating. Where you have a
11:45
fundamental technology that is
11:47
done, but people haven't still caught
11:49
up with what does that make
11:51
possible. This makes it possible to
11:54
distribute these installable apps without having
11:56
to have service calls, without having
11:59
to send someone out and look at
12:01
your server without having to get access to your
12:03
server. All these things that used to be true
12:05
if you wanted to distribute that kind of software,
12:07
boom, gone. The
12:10
complexity has just been compressed down to
12:13
a thin sliver. We just add our
12:15
little special sauce on top to make
12:17
it even easier. In fact, I'm about
12:19
to record an installed video. We just
12:21
went through on a walkthrough, Jason and
12:23
I and a couple of people from
12:25
the team yesterday, like what does
12:27
it take to set up this kind
12:29
of installable software on your own host? We just
12:31
used DigitalOcean. I've been testing with Hasner. If you
12:33
use any of those Cloud VMs, you can do
12:35
the same thing on your own server. If you
12:37
have an actual office,
12:39
I would totally recommend that someone just put
12:41
like a little box in the corner and
12:44
you can run chat on your internal network.
12:46
But it's so easy to set up. It's literally
12:49
you turn on a fresh box,
12:51
you paste in one command and
12:54
it's going to run through its thing. You're up
12:56
and running with auto update and all these other
12:58
things. Just a lot of
13:01
technology enabling things. It's not about the technology.
13:03
I don't think there's a lot of people
13:06
who will actually care. Oh,
13:08
it runs Docker underneath that. Whatever. Can I do the
13:11
chat thing? Can you save me like $10,000 a month
13:13
on this thing? Can
13:16
we own the data? This is something
13:18
that's also huge. I mean, I just spent three
13:20
years in Copenhagen living there. The
13:23
way the GDPR, this
13:25
privacy framework, has just
13:28
infiltrated in some senses good and
13:30
other senses. I don't know about
13:32
that, but infiltrated European business mentality.
13:34
Like how do you store data?
13:36
Where can you keep it? I
13:39
think these kinds of software is just like
13:41
tailor made for the moment. I
13:43
can imagine a Danish
13:46
public school installing this campfire
13:48
tool and running it like that. Inconceivable
13:50
that they would at this point in
13:53
the GDPR's advancement would use a tool
13:55
from us as a SaaS product as
13:57
an American company. I
14:00
just find that just really exciting. But I
14:03
think what's important to say here is all this is speculative. Jason
14:05
is so excited about a new product category. We
14:07
want to push it out there. We've thought a
14:10
lot about it. We put our hearts, sweat and
14:12
soul into it. But until
14:14
people actually buy, we don't know
14:16
what we got. Actually, I see
14:18
some things from, I want to respond to a couple of
14:20
quick things. We move on. So Justin
14:23
is asking, uh, will we eat our
14:25
own dog food with one chat? You already use
14:27
chat and basically, so we've converted the entire company
14:29
over to using campfire chat
14:31
for now. Um, and,
14:34
uh, so we're not really chatting in base camp or chatting
14:36
in this. We have, you have to dog food your own
14:38
product. You have to use your own thing as you're building
14:40
it to really start to understand what it's missing, what it's
14:43
good at, what it's not. So we've been doing that. Um,
14:45
and you're also, it says that you're surprised to see
14:48
that we're, we're doing chat considering we've been semi anti
14:50
chat for many years. We're not
14:52
against chat. We're against chat
14:54
as a primary method of communication.
14:57
It's a terrible default method
15:00
for all communication across the company. And so
15:02
most companies who've switched to something like Slack
15:04
all in or teams, like
15:06
it sucks, you're bombarded with, with,
15:08
with notifications all day. You're bouncing
15:10
between dozens of different channels. It's
15:13
chaotic. We don't support that at all,
15:15
but as a tool, it's a good thing to have
15:17
in the tool belt. We do go to chat when
15:19
we need to discuss something or hash something out real
15:21
time. It's very good for social interactions and just kind
15:23
of screwing around, shooting the shit kind of
15:25
thing. Um, and it's also really
15:27
nice for really small teams occasionally to make some
15:30
quick progress. So it's a
15:32
tool just like in base camp. It's one
15:34
of eight different tools we offer within a
15:36
project. It's not the primary method. So we
15:38
primarily talk long form, but we do go
15:41
to chat occasionally. So part of what's
15:43
interesting about this whole thing is that,
15:45
you know, while this is technically
15:47
could be a Slack replacement for many
15:49
companies, I actually see it as also
15:52
sort of Slack adjacent or teams adjacent.
15:55
You can imagine many of the people in the company
15:57
still using Slack, but you set up a separate corner
15:59
for just. executive team or just a
16:01
smaller team who wants to keep things completely
16:03
separate. Like a separate
16:05
space, a private meeting room that's really truly
16:07
private and that only six people have been
16:10
invited to this thing. It's
16:12
so low cost that it's like
16:14
a no brainer to have also.
16:17
So this idea of also software that
16:19
you already have something primarily for this,
16:21
but you want another similar version
16:23
also for a different use case. The other
16:25
thing is that we are giving
16:27
the code away. So not only
16:29
do you get this thing, but you get to see
16:32
how this product was built all the way through, which
16:34
is something you never ever get with commercial software. So
16:37
you might just buy it simply because you're curious about how
16:39
it was built, how a team like us put this product
16:41
together. And again, the price is going
16:43
to be, it's less than a conference ticket in most cases.
16:45
So like it becomes another
16:47
no brainer purchase to learn and to have
16:49
a team dissect this stuff and look through
16:52
it. So anyway, that's just a couple of quick
16:54
things I wanted to get to. On that note, Kevin
16:56
asked, who do you see as the target demo
16:58
initially for campfire? You've talked about that a little
17:00
bit, but maybe Jason, who isn't the target audience
17:02
for this new product? I would
17:05
say this is a really interesting open question,
17:07
to be honest. I'm not really entirely sure
17:09
who's going to, like, is
17:12
it a company size? Is
17:14
it a type of company? There
17:17
is some basic technical knowledge required, very
17:19
low, but some to kind of get
17:21
this going. If you're using Slack
17:23
and it's free, why would you pay for anything at
17:25
all? Like there's some of that. But
17:28
I think that there's a couple things. There's a privacy
17:30
aspect of it. There's a control on your own data aspect
17:32
of it. There's just the ability to
17:34
look at the product all the way through and
17:36
know exactly what it's doing, knowing that your data
17:39
is not being sent to someone else's servers and
17:41
someone else's location. There's the
17:43
ability to potentially customize this because you get the code and
17:45
you can mess around with it to some degree. There's
17:48
this idea of having a separate, segregated
17:50
place for certain discussions that are separate
17:52
from the rest. Companies do this all
17:54
the time. They have, like, file
17:57
system and there's only certain people who have certain
17:59
access to specific folder like that's
18:02
and you know, of course, you can do that
18:04
and other tools, but to actually thoroughly separate them
18:06
is even safer. But I think we're
18:08
gonna see a lot of mid-size companies jump into this.
18:10
I think we're gonna see a lot of small teams
18:12
who are product development teams who are curious about how
18:14
this works by this. I think we're gonna see a
18:17
lot of large companies like with do with Basecamp, the
18:19
whole company doesn't use it. But
18:21
a team within it uses it. They want their own
18:23
thing. They want their own place.
18:25
They want their own rules that they want to set
18:27
up. So I think we'll see it in
18:30
a lot of different places. And I think
18:32
ultimately, if we can get the pricing right in
18:34
other markets, we're gonna see this adopted all over the
18:36
world in markets where, you know,
18:38
typical SaaS software is way too expensive
18:41
to get going. So what we'll have to
18:43
see, I'm very curious. We typically, to be honest, we
18:46
don't think about markets so much ahead
18:49
of time for anything we build. We think
18:51
like, what do we want to make for ourselves? And
18:54
let's put it out there and then we'll find out what the market has
18:56
to say about it. So with
18:58
Campfire, everyone's loading it themselves using their
19:00
own servers. This kind of answers Chris's questions.
19:02
Maybe you guys can address that. Will
19:04
Campfire have a mechanism for preserving messages for
19:07
legal purposes? We don't have the data.
19:09
In the sense that you
19:12
have the data, you have the database,
19:14
the system is built for you to
19:16
be able to take backups of that
19:18
database. There are no sort of enterprise
19:21
retention controls in the traditional sense that
19:23
you would with a Slack or something.
19:26
But in some ways, it's not
19:28
needed in the same way because it's literally
19:30
your installation. Like it's literally running on your
19:32
stuff, either your physical server or
19:34
a cloud VM that you control. You have
19:37
direct access to the database. We've built a
19:39
really nice system in where you can take
19:41
backups of not just the database, but all
19:43
the files that's been uploaded. So say, for
19:45
example, you have a preservation request, you're in
19:47
a lawsuit and they say, hey, you can't
19:49
anything in the past three months, you got
19:51
to just store, right? You can do it
19:53
back up and take that back up
19:55
and put it somewhere safe. Just
19:57
that you are in compliance with that subpoena or whatever.
20:00
whatever it is. So,
20:02
yeah, I think in that sense, it's quite
20:04
well set up, but it's not the same
20:06
way. You really have to shift your mindset
20:09
a little bit when you go from multi-tenancy,
20:11
which is what systems like Slack and Basecamp
20:13
is when someone else runs on your behalf,
20:15
to single-tenancy where you own the whole thing.
20:18
Like you're not renting. You
20:20
are buying a product here, and
20:22
it's yours. This is the other
20:24
thing here. You're not making a license
20:26
agreement with you that we can revoke your
20:28
right to run the software. There's a license
20:30
agreement on what you can do with the
20:33
code. Like, it's still copyrighted. We trademark the
20:35
name. And so you can't just buy our
20:37
code and then give it away for free
20:39
or distribute it or whatever, but it is
20:42
yours to run the way you see fit.
20:45
You don't have to get any of our updates.
20:47
You can buy, can't file once, continue to run
20:49
it on your services, use please, and you can
20:51
run it in the next thousand years.
20:54
I mean, I'm making a sort of
20:56
broad proclamation here. But the software is
20:58
yours. The software is open source. I
21:00
guarantee, well, I shouldn't say guarantee, who
21:03
knows? Civilization is around in a few
21:05
hundred years. But if it is, if
21:07
civilization is still around, there will be
21:10
people who are capable of running this,
21:12
which is absolutely not true for commercial
21:14
SaaS companies like Slack or even
21:16
like Basecamp, right? Like you can't
21:19
make declarations about whether this is
21:21
going to be around in a hundred years or
21:23
two hundred years. Now, is this actually a practical
21:25
concern? I don't know. Not so
21:27
much, but there's a philosophical attraction to this
21:29
that really appeals to me. One of the
21:31
reasons it does is because I've gotten really
21:33
into retro gaming. There's these
21:35
emulators that allow me to
21:37
play the video games I first played
21:39
when I was like five years old.
21:41
It's incredible. You can emulate a Commodore
21:43
64, which was the
21:46
first computer I really had exposure to.
21:48
All those games on a modern sort
21:50
of handheld gaming device, were you actually
21:52
playing that real software? That
21:56
to me is magic. We have a
21:58
history here of computers. and software
22:00
that we should be cherishing. I do
22:02
worry a little bit that the
22:05
sassification of everything destroys that. This
22:07
is one of the other reasons
22:09
I'm so thrilled that we still
22:11
run silly things like Tadao is,
22:13
this free service we made like 18
22:15
years ago. Because it's trying to give
22:18
friends to this idea that software should have
22:20
a legacy and a history. All the major
22:23
material things that I truly enjoy,
22:25
whether it's watches or cars or
22:27
cameras, they all have these long
22:29
legacies and you can take
22:32
a watch. Actually, I mean, this
22:34
is a watch from I think 1968. It
22:37
is from before I was born. I can
22:39
wear it on my wrist and it tells
22:41
time. Like that's amazing. Software
22:44
should be more like that. I think
22:46
with this kind of software, we're trying to put our
22:49
chips in and say like, all right, it's going to
22:51
be that. I think that's just
22:53
fun. Okay. I want to
22:55
go- You're addressed. I'm sorry, Kimberly. Go ahead. No, no. I
22:58
was going to go to SHAPE UP. So if you want to stay on
23:00
the- One more thing about- Let's do one more thing about
23:02
once. First, let's come to SHAPE UP. I saw, David, this
23:04
is really more of a question for you. It says like,
23:06
if you're giving the code away, does that make it open
23:08
source? Can people modify it? How are
23:10
we thinking about allowing people to mess with the
23:12
code and see the code? Yeah, this is
23:15
where I think the parallels are so great too,
23:17
to physical stuff. This watch or
23:19
you take a Leica camera M3, or
23:22
you take a Corvette from 1968. Those
23:25
are your things. Like, I
23:28
didn't sign a end user agreement
23:30
with Rolex, or General
23:33
Motors, or Leica, to be able to
23:35
own those things and do whatever the
23:37
hell I please with those things. Now,
23:39
the question is whether you can. If
23:42
you buy a mechanical watch from
23:45
the sixties, do you know what?
23:47
You can take it to any independent service provider.
23:49
They may be good, they may be bad, but they
23:51
can work on it. They can open the case and
23:53
they can rinse it out and clean it out, whatever.
23:55
You can take your Leica M3 to an independent shop
23:57
and they can fix it. You can take your car.
24:00
If you're good enough, interested enough in this stuff, you
24:02
can learn it yourself. You can change the, I
24:04
don't know, corporate or battery,
24:06
whatever. Do you know what
24:08
you can't do if you buy a goddamn Tesla? Do anything.
24:12
Now, I love Tesla. So, I'm
24:14
not taking a dig at them
24:16
here. I'm saying there's been a shift here
24:18
where everything has become software and all software
24:20
has become closed when it comes to commercial
24:22
systems. So, this is
24:25
not that. This is a Mustang 68 that
24:27
you can buy. And if you can teach yourself
24:30
how to operate it and maintain it and change
24:32
it, you can. You can install a different stock
24:34
and do all those things. That's
24:36
not the same as open source as people
24:38
traditionally understand it. People understand open source in
24:40
the same way. We give away Ruby on
24:43
Rails, right? You can distribute it. You
24:45
can build it into something else. You can build a product
24:47
with it. You can sell that product. There's a lot of
24:49
rights that come with that kind of open source. We're
24:52
keeping some of those rights. So, this
24:54
is somewhere in the in-between land. This
24:56
is commercial software. The source
24:58
code is trademarked. Sorry,
25:02
copyrighted. You can't just take it and
25:04
distribute it. Just like you can't just buy a book, take
25:06
photocopies of all the pages, and then hand
25:09
that book to someone else. Now, you know
25:11
what? We have some limitations here to ensure
25:13
that commerce endures. But
25:15
you can still get into the machine room.
25:18
You can change it. You can adapt it,
25:20
modify it, do whatever the hell you want.
25:24
And I think that's what... It's really an
25:26
intersection of two things I care about. I
25:28
care about commercially viable software where companies can
25:30
actually make profits and pay employees and all
25:32
those other things. And then I also care
25:34
about open source. And this exists somewhere, I
25:37
don't know, in the middle. Sorry,
25:39
Jumba. No, no, no. There's a couple of... Since
25:41
people are asking about one stuff since we're on
25:43
the topic, let me just knock out a couple
25:45
of quick ones. Nick, how do you order the
25:48
economics? How are you going to charge for it
25:50
again if you're going to update it? So
25:53
basically, it would seem to work like traditional software has
25:55
always worked. Every major version. So if you have 1.0
25:57
and 1.1 and 1.2... and
26:00
1.3. These are all included for free,
26:02
free updates. If we ever launch a
26:04
2.0 or like a massive redo or
26:06
a massive upgrade, we will sell that
26:08
again as an optional upgrade. That's
26:11
how that'll work. Also, we're going to do
26:13
what I think we're going to guarantee three years of
26:15
security updates regardless of what version you're on. So we
26:17
will do that. Which let me just
26:19
take one point in that because I've gotten some feedback
26:21
on that too. Oh, okay. So if I install my
26:23
own software, like, how am I even going to get
26:25
updates? We built all that in.
26:28
So actually, this is something we did recently. When
26:30
we started out, it was like a manual process. If
26:32
you wanted to get an update, you'd go in and
26:34
update yourself. Now we build an auto updater, which is
26:36
not like something new. Your
26:39
Mac has an auto updater. It's not SaaS.
26:41
Your Mac is not SaaS. It'll pull an
26:43
update from Apple, or your computer will ask
26:45
like once a day, oh, Apple, is there
26:47
any new updates here? We've done the same
26:49
thing. So when you install this campfire through
26:51
the once command, it'll set up an auto
26:53
updater. Every night, I think it's at
26:56
2am in the morning, the system will ask our servers,
26:58
hey, is there any update? And it will auto apply.
27:00
You can turn that off. Not everyone wants auto updates.
27:02
But if you want the auto updates, it's going to
27:04
be on by default, and it's going to be super
27:06
duper easy. Okay, another one
27:09
question. Will you try to have
27:11
similar products of once products play
27:13
together? Are they completely separate? Are they going to?
27:15
Don't know, don't care. Not thinking about that
27:17
right now. Yeah, there's a lot of, I
27:19
saw some other questions about it. It will
27:21
not do so many things. It will not
27:23
do almost everything. Let's do a
27:25
few things. Each one of these tools will do
27:28
a few things really well. What
27:30
we're really aiming for here are
27:32
high quality generics. If
27:35
you think about like what is the canonical
27:37
version of a group chat? It's
27:39
a transcript, it's a place to talk.
27:41
It's a list of rooms, maybe some
27:44
direct messages. It's like that. That is
27:46
what this is. It is not more
27:48
than that. It's not more sophisticated than
27:50
that intentionally. This is like the core
27:53
nugget, the 80 or maybe 90% of
27:56
what these tools really stand for.
27:59
And then all the others. stuff around it is not
28:01
part of it. So these are
28:03
radically simplified, straightforward versions of
28:07
popular product categories that
28:09
we've just stripped down. And
28:11
that's why they're just, you know, let's call it a few
28:13
hundred bucks once. I
28:15
think this is one of the things Jason and
28:17
I were talking about yesterday. Like, I've been so
28:20
excited about like the prospect, the new distribution forms,
28:22
everything we're bringing back and I mean, I think
28:24
it's fair to say we've been hyping that. I
28:26
mean, I've certainly been hyping that and we need
28:29
to sort of take down the expectations like two
28:31
levels from like what is the actual software underneath?
28:33
Because if you have the expectation that this is
28:35
like exactly the same as Slack with all the
28:37
features, bells and whistles that Slack does just massively
28:40
cheaper and you get to own it and we
28:42
build it all in like the amount of time
28:44
we did. I mean, I'm sorry, but
28:46
that's delusional. Like, that's just not
28:48
how things work. You can't
28:51
get all of it plus like 500 times
28:53
extra. So this is how it
28:55
always is when we make new products when
28:57
hey, email launched, right? It included a ton
28:59
of really novel, interesting things. And then there
29:01
was a bunch of baseline stuff that like
29:03
wasn't there on day one. Now,
29:05
some of that baseline stuff still isn't there.
29:08
We've covered the majority of it, but some
29:10
of it isn't there. We're not trying to
29:12
replicate everything that single thing that Gmail does.
29:14
I think there's like 500 different items or
29:17
configuration points and Slack has grown into like
29:19
a Gmail like monster in that regard to
29:21
Jason and I a few weeks back went
29:23
through like the sign up process for Slack
29:25
and was like, holy shit, like I don't remember
29:27
it just being this
29:30
much stuff. And you can tell me, I mean,
29:32
that's not a critique per se, like every
29:35
single little checkbox that it has in
29:37
its 500 million configuration screens like is
29:40
there because it helped make a sale
29:42
to some enterprise the organization somewhere. That's
29:44
their business model. Totally fine. Great. Actually,
29:46
that should exist. It should be good.
29:49
We're making something else. And I think it
29:51
connects to this idea of like, that
29:53
watch I held up 1968 to
29:56
have something endure for like
29:58
not just two years. for 10
30:00
years, for 50 years. You
30:03
actually kind of want it simple. I think
30:05
that's the recipe behind the
30:08
magic of a Leica camera from 1954 that still
30:10
works. Like a
30:12
Corvette from 1968, you
30:14
can still make run because they're actually quite
30:16
simple products. I
30:19
do not think that my Tesla Model S is
30:21
going to run in 50 years. That's
30:24
just unlikely. There
30:26
are so highly complicated systems built with
30:28
computers and software that it just seems
30:30
like, you know what, this is a
30:32
more of a disposable item. Again,
30:35
that doesn't mean it's bad. It means it's
30:37
actually good for a lot of things. We get a lot of progress out
30:39
of that. But it also means there
30:41
should exist a category of software that's not
30:43
that. That is more like trying
30:45
to be a Leica M3, more like trying to be a
30:47
68 Corvette, more like a Rolex
30:50
GMT from 1972. We're
30:53
trying to make that. Okay, last one's
30:55
question before we go to shape up. I doubt it.
30:58
This one is, how do you distinguish products
31:00
suitable and not suitable for the ONCE model?
31:02
That's from Nick who's live with us. The
31:05
way I think about it is, can you boil
31:07
this down to a few things and be done?
31:11
For example, people have been asking for, like,
31:13
can we do a CRM, a
31:15
ONCE CRM? Like, I'm
31:17
not so sure because CRM requires a
31:19
lot of integrations with email and some
31:22
other sophisticated external systems. There's
31:25
a lot to that that is not as
31:27
simple as, like, literally group
31:30
chat. It's obscene that anyone pays
31:32
more than 50 bucks
31:34
a month for group. It's just obscene. Like,
31:37
it's a transcript in a box where you type.
31:40
That is all it is. That
31:43
is it. I know there's other things
31:45
it does like threading. You
31:47
don't need any of that. You don't. It's
31:49
just like, can we find ideas when we have a
31:52
handful of them that are
31:54
so simple in their concept, conceptually
31:56
simple, that we can deliver a
31:58
very high quality, simple version. Another
32:00
example of this, which we might do next,
32:02
I don't know, is Kanban. The
32:04
concept of Kanban, columns and tickets
32:07
and moving call tickets between columns,
32:09
that is 99% of what Kanban
32:11
is. Simple,
32:14
conceptually simple. You
32:17
can squint and draw it
32:20
and cover all the bases. I think this is another
32:22
way I've thought about it. It's like, if I could
32:24
sketch the UI and cover 95% of the whole product
32:27
with a single sketch with eight
32:30
lines, that's the one's product.
32:33
Kanban is like that, chat is like that, there's a
32:35
number of other things like that. Things
32:37
that people already know how to use, already understand that
32:40
are simple concepts that you can do on paper. You
32:43
can make a Kanban board for your office by
32:46
using this tape on your wall
32:48
and post it notes. That's Kanban, right?
32:52
You can do that digitally, of course. It does need
32:54
to do all the things that something like Trello does
32:56
and all the other things that other
32:58
Kanban systems do. What's the simplest version?
33:00
Is there one that's attainable very quickly that we
33:02
could build in a few months? Absolutely. That's sort
33:05
of the bar for us. I
33:07
think it's good to contrast this with
33:09
what doesn't fall in that category, for
33:11
example. Hey is a perfect
33:13
example. Hey is an email
33:16
service, not an
33:18
email client. Running an email service
33:20
today is actually kind
33:22
of jokingly, depressingly
33:24
difficult. Ensuring deliverability
33:26
of email has turned into
33:28
a freaking PhD project.
33:33
I'm trying to remember all the standards that you
33:35
have to keep in your head, DKIM, DISS and
33:38
SC. I can't
33:40
even keep it in my head on an
33:42
operating basis. That's how complicated
33:44
it is. It's not a good system to try
33:46
to run yourself. It used to be, I used
33:49
to run my own mail server, back in the
33:51
late 90s, I think it was. I'd run my
33:53
own mail server. Lots of people did. But today,
33:56
you really have to be a
33:58
very dedicated hobbyist to attempt to running
34:00
your own email service, if you care whether
34:02
your emails reach the recipients. So that's the
34:04
kind of service that I think, in
34:07
the word actually, service. Anything
34:09
that's a service where someone has to
34:12
tend to the operations of this thing
34:14
on a running, ongoing basis, otherwise it
34:16
falls apart, it's a poor fit
34:18
for this. Now, there are people who tried. I actually
34:20
bought like a physical product. I forget what it was
34:22
called, but someone did a physical product that was a
34:25
mail server in a box. And
34:27
it was just, it was complicated. It was not
34:29
a single sketch. It was not actually set up,
34:31
even though it actually came as a physical thing.
34:33
I'm like, I don't think this
34:35
is a good fit for it. And that's the other
34:37
important part of this. When we talk about ones, we're
34:39
not saying all SaaS needs to
34:41
go away. Every SaaS product needs to
34:43
be a installable product. That
34:46
is unrealistic. And in fact, bad.
34:48
There are lots of services that are best
34:50
of services. Email is a great example of
34:52
that. Anything that requires a bunch of integrations
34:56
or updates or continuous
34:58
refinement doesn't really fit this
35:01
model all that well. I
35:03
love that we're recording this live. You guys
35:05
can actually see what happens. It's me just
35:07
trying to not interrupt them, basically. It's how
35:09
it goes. Oh, here comes it. No, quick.
35:11
Here's a question that I actually a little
35:13
off, I thought was a great one. Jason,
35:15
you used to do lots of customer support
35:17
work, which would have helped you stay in
35:19
the loop with existing and new customers. Now
35:21
that you delegate the work, how do you
35:23
maintain your customer muscles? I'm
35:26
a little bit more involved in new
35:28
product development than maintaining existing products these
35:30
days. So because
35:33
there are no new customers for new products that
35:35
we don't have yet, it's a little bit different.
35:37
It's more about gut intuition, what do we want
35:39
to build for ourselves? But
35:41
we have someone named Brian, for example, Brian's our head
35:43
of strategy. He's keeping track of
35:45
all the things that are on customers' minds. He's
35:48
doing a lot of the shaping work for Basecamp
35:50
and continuing to improve Basecamp. So he's very much
35:53
in touch and
35:56
in the weeds there. And if I was doing that, then
35:58
I'd be in the weeds there as well. since
36:00
I'm more on the leading edge of things that
36:02
we're doing right now, like the Hey! Calendar, of
36:04
course, we had to request for Calendar in Hey!,
36:06
but like the thing we built was not what
36:08
people told us to build. That's sort of
36:10
my focus. So it's a little bit different these days. Although
36:13
I will say that when we do launch a new product,
36:15
as we just did, we step into the
36:19
river of customer feedback. I
36:21
mean, both Jason and I have fielded
36:24
absolutely hundreds of emails from customers giving
36:26
us feedback on Hey! Calendar and
36:28
whether that feature requests, bug reports, or
36:30
what they'd like to see, or just
36:32
appreciation for the fact that they found the
36:34
product so likable, all that
36:37
stuff. Like we get the firehose
36:39
sort of occasionally, but yeah,
36:41
I think at some point, I
36:43
think we get, I mean, many hundreds of emails a
36:45
day on customer support. Gone are
36:47
the day where Jason could just answer them all
36:50
by himself, which he did. And I think the
36:52
first three years, I think, I think Jason was
36:54
doing 160 emails. A
36:57
day on customer support while also running
36:59
everything else. And just for sense of
37:01
scale, what we asked or
37:03
what we benchmarked our customer support team
37:06
is 50 to 60 tickets a day.
37:08
So Jason was working three customer support
37:10
jobs while also being CEO and running
37:12
things in the early days. Like, yeah,
37:15
good times. And we walked by the way,
37:17
uphill, uphill, both directions. And it was always
37:19
snowing and we were bare feet. And I
37:22
had a snow machine indoors just to make
37:24
sure of the, but the thing is, is
37:26
like, and I would recommend every founder or
37:28
whatever, anyone making something new for the first
37:30
time does that. You must do that. You
37:33
want to step into the firehose. You,
37:36
but after you launch the
37:38
thing. So I would not go ask people what they
37:40
want. I would, I would build something, then you put
37:43
it out there, then you step into the firehose and
37:45
you learn. There's no better way to learn than that.
37:47
So all right. Okay.
37:50
I've been saying shape ups for a while now because
37:52
some people mentioned it on Twitter. And
37:54
then Parker, I do see you. I haven't
37:56
forgotten about you. The question is, can you
37:58
explain use of shape up? when building
38:01
new product lines? Yes, so
38:03
I was just on a podcast, Hackers
38:06
Incorporated with Adam Wotham yesterday,
38:08
two years ago, when that launched. And
38:11
he asked me about how we built Hey Calendar, and I
38:13
talked all about it. And then there was this one part
38:16
in there where I'm like, well, we don't really follow all
38:18
the traditional shape-up methodologies when
38:20
building a new product. And
38:22
that caught a few people's eyes or attention. They're like,
38:24
gotcha, gotcha. I knew you guys didn't use shape-up. I
38:27
know you invented it, but you don't use it. We
38:30
do use it. What
38:32
we've discovered, though, and this is a constant
38:35
process discovery and evolution. This whole system
38:37
evolved over 20 years. Which
38:40
are now six weeks, used to be three months. And
38:42
before that, we didn't have cycles. So this is an
38:44
evolution. Any
38:47
process that's stuck is dead. Like
38:50
Latin as a language is dead because it doesn't
38:52
change. Things that don't change are
38:54
dead. This process continues to change. What
38:56
we've discovered is that for new product
38:58
development, while we stack up
39:01
work back to back to back to back like
39:03
you would in a cycle, the cycles
39:05
are a little bit more fluid to some degree.
39:08
We don't shape everything ahead of time. Some
39:10
stuff that's especially small is just sort of
39:12
done on the fly. But
39:14
what we don't do is we don't let things take
39:16
as long as they want. So that's like also
39:20
consistently shape-up. Shape-up has an appetite, maximum
39:22
six weeks. We're not working on one
39:24
feature for eight months this way either.
39:26
There are limits. We
39:29
don't have cool downs typically during the new
39:31
product development process where we do with existing
39:33
product development. So typically we do
39:35
six weeks and then a two-week cool
39:37
down and then six weeks. That's what we do on
39:40
existing products. For this we go back to back
39:42
to back to back to back. It's a little bit more of a
39:44
constant jog without
39:46
rest. We're not sprinting because you can't do that for
39:48
months and months and months. But we do that maybe
39:50
a little bit more towards the end. But
39:52
it's a little bit more chaotic, a little bit
39:54
more on the fly, a little bit more feeling
39:56
things out because we're going to be using this thing as
39:58
we're building it. We don't... really know what
40:00
we're making yet. And you can't
40:02
be so rigid ahead of time to know what
40:04
you're going to do over the next year when
40:07
you're exploring something new. But
40:09
there are time boxes, there are limits.
40:11
There is this urgency not to spend too
40:13
much time on something. We want to make
40:15
the thing that we're using. And
40:17
we want to move on for things. We want to finish things
40:19
up and move on. Sometimes you leave something sort of half done,
40:22
you know, you're gonna get back to it, but you also know
40:24
you got to move things on. So
40:26
it's an evolution. It's a different context. It's
40:28
a slightly different variation. And we should update
40:30
the book and update the website and talk
40:32
more about our latest discoveries and how
40:34
to pull this off for new product development. I
40:37
think one thing I've realized is that where
40:40
Shapr3D really helps us at this stage of
40:42
the company is it allows a delegation
40:46
of decision making power on
40:48
an interval where Jason and I don't have to be
40:50
involved on like a day to day or weekly basis.
40:53
So with Brian, for example, shaping
40:56
most of the pitches, taking a lot of
40:58
input from customers and feedback, what we otherwise
41:00
want to do, he can like go off
41:02
and do that work for several weeks, maybe
41:04
he'll sort of bob with Jason a couple
41:07
of times. And then like every cycle, we
41:09
can look at what's on the menu.
41:12
And when you look at it in that
41:14
way, you end up with a certain course
41:16
decision, your decision making power, we cannot make
41:18
decisions in that forum, unlike an hour by
41:20
hour basis, or even a day by day
41:23
basis, the minimum block we can actually operate
41:25
with is a week, and more
41:27
likely it's two weeks than most projects that get
41:29
shaped, they have like a two week block. That's
41:32
fine. When you're evolving an existing product,
41:34
and you can set a couple things
41:37
to go when you're doing R&D, that's
41:39
inherently explorative, you don't
41:41
even know what you want to build, right? Because
41:43
you don't know exactly the shape of
41:45
the product itself, you have to be
41:47
working on a day to day basis. And
41:49
the decision making power to
41:51
change direction should be embedded in
41:53
the team. So you don't have
41:55
this delegation in the same way
41:57
with both hey calendar and with
42:00
with once, Jason and I have
42:02
been in it, him
42:04
on the calendar side, at least on the design part
42:06
of it, and I've been on
42:08
the technical side with once, like really in it
42:10
all the time to constantly like, all right, we
42:13
hit a dead wall here, let's change direction, let's
42:15
go over here. So you're operating almost like a
42:17
different scale, like if you think frog view versus
42:19
bird side view, that the
42:21
more you get up into the level of, I
42:24
hate the fucking word, but stakeholders,
42:26
like people who want to see an outcome
42:28
of something, but they're not involved in the
42:30
day to day of it, shape up becomes
42:33
more and more important, more and more critical
42:35
that you have the shaping and you have
42:37
the repeating ways of doing
42:39
it. But I'll be perfectly honest, like
42:41
if I was starting a new company tomorrow, and we
42:44
were four people, we would
42:46
not use any process, barely.
42:49
I mean, we wouldn't even have we wouldn't have kickoffs,
42:51
we wouldn't have heartbeats, because it would be no one
42:53
to communicate with. The smaller
42:55
you are, and the more isolated that
42:57
group is from the rest of an
42:59
organization, the less methodology, the less
43:01
process you need, and the more
43:04
liberating that is, this is why I think a
43:06
lot of people, a lot of entrepreneurs look back
43:08
wistfully on the early days, oh, remember when we
43:10
were just five people, like, there wasn't
43:12
any of the complexity? Yeah, because the scale
43:14
was so small. It's just, it isn't hard
43:17
to organize more people, it just isn't. Organizing
43:20
a company of like, whatever, 72,
43:23
75, that's a more difficult problem, let
43:25
alone organizing company of a few hundred
43:27
or a few thousand or 10s of
43:29
thousands, right? They're just different scales. And
43:31
if there's one thing I think we've
43:33
been consistent in saying is you have
43:35
to pick tools that are appropriate for
43:37
the scale you're at. This is
43:40
why I'm so adamant that microservices
43:42
is a terrible idea for most
43:44
small teams because these are techniques
43:47
built by and formed
43:49
in huge organizations. What works
43:51
for Amazon at that level?
43:53
It's not what's going to work for you. So I
43:56
think ShapeUp has a little bit of the same thing. I don't actually
43:58
think most people go to the same level. looking for
44:00
shape up until they run into
44:02
process pains. You're not going to
44:05
have process pains with four people. You're going to have
44:07
process pains maybe 10, 12. Now you're going
44:09
to have to have multiple people and they have to have
44:11
an eye on five different things. They can't be involved in
44:13
everything all the time. Shape ups makes
44:15
a ton of sense. Do you know how
44:18
big the ones team was? We had, yeah,
44:20
how big was the ones team? For the
44:22
longest time, two developers, one designer, Jason, me.
44:25
That's five. You don't need any process with
44:27
five people. Okay, here's another
44:29
shape up question before we move on to just
44:31
the random questions. How do you handle betting on
44:33
smaller projects like say three weeks?
44:35
During the six weeks, the engineers focus on
44:37
two shorter projects rather than a big one.
44:39
Just a general shape up question for us
44:41
here. Yeah, so before a cycle starts,
44:44
we set up the work that we're going to do
44:46
and not the tasks, but
44:48
the big ideas. And
44:50
we say, we're going to pick, for
44:52
this example, I'm going to say we're going to
44:54
pick five things to do. Two
44:56
of these things are going to be full cycle
44:58
projects. They're going to take the whole time. They're
45:01
bigger things. I'll give you
45:03
an example from Hey Calendar, if
45:05
we're doing this in shape up. Habit
45:08
tracking and time tracking are both full cycle
45:10
projects. They're going to take the full six
45:12
weeks. But there's a bunch of smaller
45:14
things that are going to take a week or two
45:16
to do. And so we set the appetites
45:18
for each one of these projects. And
45:21
usually, we have a team work on a
45:25
big thing, a teamwork and a big thing. And then
45:27
one team might work on four smaller things in a
45:29
given cycle. So there's like the small
45:31
batch team, which will do a bunch of
45:33
things, and the big batch teams, which will
45:35
do one thing. And that's kind of how
45:37
that comes together. And like, the
45:40
other important thing here is that this isn't really,
45:42
we're not aiming
45:44
for rigidity here. There is a framework
45:46
and a concept that's supposed to help
45:48
guide things and prevent the worst
45:50
parts of human nature from taking hold, like
45:53
just giving things an unlimited amount of time
45:55
and perfectionism and all that stuff. So there's
45:57
discipline here that's built into the system, but
45:59
there's all also flexibility. So
46:02
if someone needs to jump over from one project to another,
46:04
they need help, it's of course, the answer is of course.
46:07
It's all within reason. But
46:09
that's typically how we would assign the work
46:11
out. And then again, once the work is
46:13
assigned, the work that's assigned is the big
46:15
idea. The actual bits that need to get
46:18
done are defined by the team doing the
46:20
work. There's no architect or taskmaster
46:22
or scrum master or whatever making
46:24
all the individual tasks and tickets
46:27
for each person. The people
46:29
doing the work make their own work. That's how that
46:31
whole thing works. And
46:33
I think it's key here to contrast
46:35
that with, in my
46:37
opinion, the ways agile, big A
46:40
agile, mythologies went astray, because they
46:43
absolutely did try to codify way
46:45
too many things in too
46:47
high a level of rigidity, ironically,
46:49
given the fact that it's called agile, but everything's
46:52
called agile this day. No one is gonna say,
46:54
I'm not agile, right? Everyone say they're agile about
46:56
something. And then they present
46:58
a very rigid definition of what
47:01
that agility actually looks like. And
47:04
it's far away from the agile manifesto, which
47:06
was this idea of
47:08
it's still online, if you Google agile
47:11
manifesto, you can see these
47:13
ideas of people or processes, for example,
47:15
like if something's gonna bend or give, we're
47:17
gonna give it to people, we're not going
47:19
to stay rigid in the process. And this
47:21
is where the context setting is so important
47:23
that you can't just take even shape
47:26
up. I don't think you can't
47:28
just apply shape up as a whole to an
47:30
entire organization of 100,000 people without making
47:33
serious modifications. I also
47:35
don't think you can apply it as we wrote the
47:37
original book to a team of four people. It
47:40
fits somewhere within the processes starting to hurt.
47:42
And we're not a mega company yet. So I
47:44
don't know if I was going to say something,
47:46
I'd say from 10 people to maybe 500. Something
47:51
in there is the sweet spot for
47:53
it. The further you get away from
47:55
that, either on the small side or
47:57
on the big side, you're gonna have
47:59
to make modifications, you're gonna have to
48:01
make a change. changes to it. And
48:03
that should not be seen as a
48:05
refutation of the ideas that are in
48:07
there. You can totally take a concept
48:09
of, for example, budgets over estimates, right?
48:11
I think that is a universally applicable
48:13
concept that goes directly to Jason says,
48:15
like the worst parts of human nature.
48:17
If you go with estimates, humans suck
48:19
at estimation. Even if they're foreign people, they will still suck
48:21
at estimation. So you can take that idea and go like,
48:23
yeah, you know what, we're going to apply that here. We're
48:26
going to set up some appetites. We're going to do that.
48:28
We're not going to have the whole thing. We're not going
48:30
to do the formal pitches and
48:32
they're not going to be at that level. But we're going to
48:34
think about estimates. Okay, I'm
48:36
going to move on to this question about
48:38
new products from Julian. When you start something
48:41
new, how much time do you spend on
48:43
it before saying it's a good idea or
48:45
not? So if we're talking about a
48:47
brand new product, it's typically, it depends.
48:49
I guess like, if we're, for example, this once
48:51
campfire, like we've already built chat a few times.
48:53
So we kind of knew what this was going
48:55
to be. But let's say that the hey calendar,
48:58
let's take that as a, I always want
49:00
to get practical. So by the way, if
49:02
we're talking too high level,
49:05
tell us in the chat, I want
49:07
to get practical and real. Okay. So
49:11
hey, calendar, we spent a few months, typically
49:14
no more than a couple months max
49:17
exploring in the early days, like is there something
49:19
here, we knew we were going to build a
49:21
calendar, we decided we were going to do
49:23
that. But what was it going to
49:25
look like? Well, it wasn't gonna look like
49:28
a traditional calendar. But what does that mean? So there
49:30
could be a million other ways to do it. So we spent
49:32
a couple months playing, it's
49:34
really play. And then at
49:37
some point, like I think it was early March, basically,
49:39
we started this, I think in January, I don't know,
49:41
something like that early March, we're like,
49:43
okay, we kind of have
49:45
a direction here. This doesn't mean
49:47
everything's finalized, basically, nothing's finalized. But
49:49
we've landed on a few interesting
49:52
ideas that form enough of a foundation
49:54
from which we can then build the walls and build
49:56
the roof and build another layer and all the whole
49:58
thing. I'd say
50:00
a couple months of exploration and play
50:02
before settling in. If you're spending more
50:05
time than that, I'm granted,
50:07
look, there's always exceptions. If
50:09
you're building an electric car from scratch and
50:11
you've never done that before, for example, you spend more
50:13
than a few months. Software,
50:16
conceptually, a couple months should get you far enough to
50:18
go, if there's something here or there's something not. That's
50:21
kind of how we basically do it because otherwise, you
50:24
will keep going. And if you feel like
50:26
you need to explore every possible avenue before you
50:28
decide to commit to something, you're
50:30
never going to get this thing done. So you've
50:32
got to go on faith at some point after you've
50:34
determined that there's enough of foundation here on which to
50:36
build future this. Okay,
50:39
I'm going to go to a question that
50:41
came earlier from Victor, which is, would your
50:43
business structure be any different if you guys
50:45
made products that you didn't use internally? We've
50:47
talked about it all the time. We are
50:49
our customers. We use Basecamp. We
50:51
use Hayes. Would the business look different if that works in the
50:53
case? I wouldn't be here. I
50:57
don't know. I'm not
50:59
interested in building products for imaginary
51:02
people. And to be honest, I know it
51:04
sounds really... I don't know
51:06
what it sounds like, but I don't mean
51:08
it in a negative way, really. It just means I
51:10
don't want to imagine someone else's problems. I've got my
51:13
own problems. It's hard enough for me to understand my
51:15
own problems. I'd rather solve my own
51:17
problems that I'm close to, that I can really understand
51:19
if this is a solution or not. I know what
51:21
my own struggles truly are. I can
51:23
just decide to build something. If the struggle
51:26
goes away, we've nailed it. At
51:28
some level, once you've got something out there
51:30
in the world, you do build for other people. But
51:33
initially, we build for us. For example,
51:36
there is no amount of money someone could pay us right now
51:38
to build a patient
51:40
scheduling software for a dental practice. It's
51:43
just not going to happen. We don't care.
51:46
You can pay us a billion dollars. I just don't
51:48
care. Not going to do it. It's not worth
51:50
doing. So, we're going to find things that
51:52
we know how to do that we struggle with, that
51:54
we want. And that's the kind
51:56
of business we build. I think if you build another kind of
51:59
business where you're doing things... things, you're imagining other
52:01
people's struggles and doing it for other people, you don't
52:03
really know where to stop. You're probably gonna have a
52:05
lot of sales people because you're gonna
52:07
need to sell this thing. You
52:09
can't just make something that's gonna be really good on
52:12
its own. And
52:14
it just doesn't sound interesting to me. I'm not really interested in
52:16
spending my time. I like to make tools
52:19
that I'm gonna use. So that's my missive
52:21
on that. I'm completely on board
52:24
with that too. You could not motivate me. It's
52:26
not even that you couldn't pay me. Yeah, of
52:28
course. It makes me to do that. And
52:31
I just know this is one of the reasons I was
52:33
such a terrible employee. Because whenever I
52:35
had to work on something for other people that
52:37
I didn't care about, I literally could not make
52:39
myself do it. So there's
52:41
that. And then the other part is,
52:43
holy shit, I'm glad not everyone feels this way. We
52:46
need someone to make the dental
52:48
practice booking software. And
52:51
this is where Jason and I are
52:53
these oddities. And we're all odd and
52:55
weird in our own different ways. We
52:57
just have accepted our own oddities and
52:59
built a business to facilitate those oddities
53:01
not getting terribly in the way. But
53:03
we need other people to do this.
53:06
Jason and I, terrible folks to ask for
53:08
advice on that because we literally could not even
53:10
motivate ourselves to barely think about it. You need
53:13
to talk to someone who's done that, who builds
53:15
software for other people. And there are lots of
53:17
people who do this. And many
53:19
of them do it well. And I have
53:21
tremendous respect for the fact that they are
53:23
able to do this. And I think the world is a better
53:25
place because they do it, not
53:28
Jason and I. Can I
53:30
take Eric and Ilan, both have a
53:32
similar question. That was what I was going to cue
53:34
up for you. So you're good. It's a pricing
53:36
question. So it's about pricing because we've
53:38
long had a policy that we don't increase
53:41
prices on existing customers. And in fact, for
53:43
the past 10 plus
53:45
years, we've changed prices a bunch
53:48
only on new customers. So we have a new pricing
53:50
model we experiment with. We put that on the website.
53:52
Anyone signing up will pay the new price existing customers
53:54
continue to pay their old price. And
53:57
we've done this for basically the tires distance
53:59
of the company. early on, we had a
54:01
little bit of price changes with a
54:03
couple of different products. But it's been a
54:05
long time. In fact, some customers that we still have,
54:07
we've had for 20 years, they're still paying
54:10
the exact same price they paid 20 years
54:12
ago. However, just
54:14
recently, we decided that
54:16
it's finally time, our
54:18
costs have gone up, as everyone's costs have gone up.
54:22
And we've instituted a price
54:24
increase on some
54:26
legacy products, some existing products. It's
54:30
about 20%, roughly, give or take
54:33
on existing customers, we've given them about 90 days notice.
54:38
And the existing prices on the current site
54:40
will not change moving forward. But we are
54:42
bringing people a bit more up to date.
54:45
Basically, our cost structure has just changed. We have a lot more
54:47
employees than we used to have. We've
54:49
saved a lot of money in the cloud, but our
54:51
payroll has gone up significantly. As everyone knows, payrolls have
54:53
gone up significantly for a variety of reasons. There's inflation,
54:55
there's a whole bunch of other things. Costs
54:58
are up. And at some
55:00
point, in order to maintain the
55:02
kind of business we want to run, we have to pass some of that
55:05
on. So we
55:07
decided, again, after a long, long period
55:09
of time, that over 10 to 20 years,
55:11
only to raise prices once,
55:13
about 20% seemed very fair,
55:16
very reasonable, and necessary for
55:18
us to continue the business the way we want
55:20
to run it. So that's how
55:22
we did that. What to me is kind
55:24
of fun to look back upon is we launched Space Camp 2004. We
55:27
still had some customers paying essentially those
55:29
2004 prices. In
55:33
2004, we were paying developers, including myself and
55:35
I think Jason as well. I think it was $42,000
55:37
or $45,000 a year. That
55:41
was the salaries. Do you know how much
55:43
programmer you get these days for $42,000?
55:47
You don't get a whole person. You
55:49
get some slice of a person somewhere. I don't know if
55:51
you want a leg or you want a head or an
55:53
arm. You're not going to get a whole programmer, at
55:56
least not at the cost basis that we're looking
55:58
at in our jurisdiction. if you want to call it that.
56:01
So if you compare that, like
56:03
what is the average salary for Silicon
56:07
Valley type levels?
56:10
It's not 20% that's gone up in 2004. It is many
56:12
times that. So
56:16
at some point, it just seems like out of sync.
56:19
Because to maintain those legacy products,
56:22
actually, sorry, let me restate. To
56:24
maintain those legacy services, that's the
56:26
key point here. We
56:29
need to continue to pay talented people
56:31
to work on it. That's
56:33
to make sure that those things are updated, is that
56:35
they're still running, or everything that goes into
56:37
offering a service sort of
56:40
requires human input. And
56:42
that is the biggest cost that we have
56:44
in our company. So even for example, as
56:46
Jason said, we've saved some money on the
56:49
cloud, very happy about that. In the grand
56:51
scheme of like payroll, it's substantial, but payroll
56:53
is so much larger than whatever those statements were.
56:55
So I also think it's just one of those
56:57
things where I
57:00
hate getting the drip drip. So
57:03
I use ADT for my security
57:05
system, whatever, they freaking send me
57:07
a well, our prices have gone
57:09
up like two, three times
57:12
a year. And it's always like, oh,
57:14
we're raising our prices 3.7 to 4%.
57:16
And I always go like, that
57:19
sounds bullshit. Like if you have that many different
57:21
models, like something is just fishy here, I'd
57:23
rather just go like, you know what, we're going to do
57:25
it rarely. But then we're going to do like a chunk.
57:27
20% first time in 20 years.
57:31
Do you know what, I don't find that
57:33
a reasonably. And that's so funny. Like what
57:35
is reasonable? It is such an individual conception.
57:39
And I've had the same thing.
57:41
I just had this thing with Disney+. So I
57:44
was on a Disney Plus plan that was $99 a year. I don't know,
57:46
maybe that was an introductory
57:48
pricing or whatever. And then I
57:50
get this email saying, hey, cool,
57:53
it's going to be 139 now. So
57:56
a 40% increase. And I just went no, can
57:59
I afford? $140 for Disney Push. Yes,
58:02
I can afford that, but I don't want to.
58:05
And I was trying to look at myself
58:07
in third person, like, why am I being
58:09
petty about that increase? It just felt
58:11
like it didn't feel fair. And I
58:13
think this is the key aspect of pricing
58:16
that you should internalize is this sense of
58:18
fairness. Now, you're not going to make anything
58:20
fair for everyone. We got some pointed
58:23
feedback on the 20% people saying that's not
58:25
fair. But broadly, the feedback
58:27
was, you know what, sounds about right.
58:30
And I think that you've got to know
58:32
that you're dealing with human emotions, you're
58:35
dealing with a sense of fairness or unfairness.
58:37
How often can you do this? How much
58:39
can you do this? And
58:41
you really have to respect those things
58:43
or otherwise you you could get yourself
58:45
in trouble. I mean, we
58:49
we root over this for some time. It wasn't just like,
58:51
all right, that's what it is. Exactly. Because we hadn't done
58:53
it for a long time. But I think I mean, given
58:55
the feedback, it seems like it was sort
58:58
of well received. But proof is always in the
59:00
pudding. And the pudding is
59:02
that the price increase will hit the first
59:04
invoices, I think, February 1st. So
59:06
that's when we'll see the moment of truth. Are people going to
59:08
pay? OK, last question that we're going
59:11
to take before we wrap it up, we've been with you
59:13
almost an hour is about A.I. What
59:16
are your thoughts on generative A.I. any plans for
59:18
doing something? And this is a scam for any
59:20
of our products. I
59:22
think A.I. is the most exciting,
59:24
interesting development in technology
59:27
that I've probably ever seen. And I
59:29
include the Internet in that, which is
59:31
saying something. So I'm hugely bullish on the
59:34
future of A.I. I think it's amazing. At
59:37
the same time, I don't feel like I
59:39
have to be the first in the pool
59:41
swimming the hardest, because
59:43
if you've paid any attention to startups
59:45
doing A.I. Stuff
59:47
over the last 18 months,
59:50
you've seen that like two or
59:52
three generations have already been steamrolled. You
59:55
think, oh, I came up with this new
59:57
idea and then the open A.I. steamroller just.
1:00:00
rolls right over you, you're no longer a thing,
1:00:02
you're a feature in their setup. Most
1:00:04
everyone right now, at least, they're
1:00:07
not doing AI. They're doing
1:00:09
API AI. That's different. You're not
1:00:11
doing fundamental research that's giving you
1:00:14
a lasting advantage here. You're
1:00:17
using the API of OpenAI or some of
1:00:19
the other tools, which, great. I'm not saying
1:00:21
that's bad. I'm just saying that, to
1:00:23
me, it seems too early. I'm
1:00:26
not actually an early adopter. Or
1:00:29
maybe I'm an early adopter. I forget how the chart
1:00:31
goes. There's a segment of people who jump in on
1:00:33
something as soon as it comes out. I
1:00:35
don't like to do that. I like to give it five minutes.
1:00:38
Can the landscape settle just
1:00:40
somewhat? So instead, we don't spend a bunch
1:00:42
of time looking into something and then it just goes and
1:00:45
then it eats all that work and swallows
1:00:48
it up inside the OpenAI machine. I don't
1:00:50
want to waste time doing that. I
1:00:53
don't think it's clear at all where AI
1:00:55
is going to land. Even a year from now,
1:00:58
is AI going to end up in a
1:01:00
place where we're even using graphical interfaces? Is
1:01:03
everything just going to be spoken? Where is
1:01:05
this actually going to be? I
1:01:09
sit with this paradox in my head.
1:01:11
I think this is the most exciting
1:01:13
thing that's happening in tech. Also, I
1:01:15
don't mind being a spectator right now.
1:01:17
Eventually, it will settle to whatever degree
1:01:20
it settles and we will go
1:01:22
right in on it. We've done a handful of
1:01:24
experiments internally. Like the one
1:01:26
everyone was excited about for a hot minute
1:01:28
was like, oh, it can summarize your boring
1:01:30
emails. So we tried
1:01:33
that. I was like, yeah, okay, cool-ish.
1:01:36
That's not the part of the Institute of AI that
1:01:38
actually gets me excited is that we take shitty writing
1:01:40
and turn it into another shitty version of the writing.
1:01:44
That doesn't get me fired up. A lot of
1:01:46
the other things do, especially around media. I'm
1:01:48
not sure it's clear where
1:01:50
we land yet, but it hopefully
1:01:53
will be in like whatever a year. Hopefully
1:01:55
not. Maybe this says it's exciting for
1:01:57
another year or two or three and we just go like.
1:02:00
I don't know what's happening. Is anyone even going
1:02:02
to work tomorrow? Or software developers
1:02:04
are all out of a job. I think
1:02:06
we've never had that amount of uncertainty in
1:02:08
plain sight. There's always been uncertainty about
1:02:11
what's new technology going to do, but it
1:02:13
always felt like something you had to dig
1:02:15
for, even the internet. I mean,
1:02:17
both Jason and I were there, like
1:02:19
95, when the internet started
1:02:21
taking off. And you have to kind of dig. Not
1:02:24
everyone understood that the internet was going to be what
1:02:26
it is. I
1:02:28
think everyone understands today that AI is
1:02:30
going to be a major shift, but
1:02:33
we don't know exactly what. Okay,
1:02:37
well, I think that is the perfect way to
1:02:39
end. Before we wrap it up, I have two
1:02:42
requests. One, if you were just coming to our
1:02:44
YouTube channel for the first time ever, please follow
1:02:46
us. That's 37signals. I know we
1:02:48
are putting this live on YouTube, but we have
1:02:50
a channel that we have videos that come out
1:02:52
all the time. So make sure to give us
1:02:54
a like and follow there. And
1:02:56
let us know in the chat my second request. What
1:02:59
was that, David? Smash the like
1:03:01
button, ring the bell. Do
1:03:04
it on YouTube and then like get all the way up on
1:03:06
the camera. You
1:03:09
guys know what to do. My other request in
1:03:11
the chat, let us know what you thought of this live.
1:03:13
This is the first time we've done it. We've never done
1:03:15
this kind of live situation with
1:03:18
Jason and David, kind of branding it as rework. So
1:03:20
let us know if it's something that you like
1:03:22
and we should do it again. But I didn't
1:03:24
get to your question. I'm sorry. There was a
1:03:27
lot going on. We will get to your question
1:03:29
hopefully the next time if we do a live
1:03:31
session again. You also can email us at rework
1:03:33
at 37signals.com or you can leave us a voicemail.
1:03:35
And that's 708-628-7850. Do you want to leave
1:03:37
us a voicemail message? We
1:03:43
often replay those in the podcast. We
1:03:45
get those from listeners. Jason, David, anything
1:03:47
else to add before we close it up
1:03:49
and get back to work? No,
1:03:52
I mean, we'd like to do these again. This is fun.
1:03:54
So yeah, I'm curious to see what people want to know. Thumbs
1:03:57
up. Yeah. Enjoy the energy of
1:03:59
live. questions that's great.
1:04:03
Okay thanks guys thanks for joining us and we will
1:04:05
see you soon. Thank you.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More