Sponsored by
Show notes
Transcript
0:00:00 Special: Sync Conf
0:00:00 I just see people both talking
about sync more on social media.
0:00:04 I see a lot of predictions that this
is the key technology for the next
0:00:08 years or more, and I suspect like
this is a one day conference, right?
0:00:13 It's a kind of maximum 300 people.
0:00:15 And my hope, and what I kind of think
is that we'll look back on it in
0:00:19 10 years time and see this as just
one of the seminal conferences that
0:00:22 actually the group of people who were
there all then went on to do amazing
0:00:26 things and you kind of look back
and it was a privilege to be there.
0:00:29 Welcome to Local-First fm. I'm your
host, Johannes Schickling, and this
0:00:33 is a special episode of the podcast.
0:00:35 We're together with Adam
Wiggins and James Arthur.
0:00:38 We're excited to announce the first Sync
Conference in San Francisco in November.
0:00:43 I hope to see you at the conference.
0:00:48 Sync Conf Announcement
0:00:49 Hey, welcome back James.
0:00:51 Adam, it's so nice to have
you back on the podcast.
0:00:53 How are you doing?
0:00:54 Great to be here.
0:00:55 Feeling good.
0:00:56 We got something to be excited about.
0:00:59 Yeah, very well, thanks.
0:01:01 Awesome.
0:01:02 Well, like Adam already mentioned, we
have quite the exciting announcement.
0:01:06 The announcement was already made
a couple of weeks ago, and in fact
0:01:11 the very first time we talked about
it publicly was at Local-First Conf.
0:01:16 Where we announced that there would
be a partner conference or sibling
0:01:20 conference, however you wanna call it,
called Sync Conf, which doesn't happen
0:01:25 in Berlin, but happens in San Francisco
, and it felt appropriate to bring on
0:01:29 the co-hosts of the event, James Adam.
0:01:33 Some of them are not here such
as Emma, Johanna, and Jans.
0:01:37 But, with you here, I think it
would be interesting to talk
0:01:41 through what the event is about.
0:01:42 Talking through the vision of the
event where it's kind of similar
0:01:47 to Local-First Conf, but also where
it's different for Local-First Conf.
0:01:52 So I think, James, you brought up
this idea initially that you just
0:01:57 saw so much demand and so much.
0:01:58 Pull and interest in sync engines, which
is what Electric Sync is all about as
0:02:04 well, that you, I think, raised the
question, would it make sense to have
0:02:08 a separate conference in partnership
that is all about sync engines?
0:02:12 So maybe you want to give your
vision for the event and what's
0:02:19 Vision and Target Audience
0:02:19 Yeah, absolutely.
0:02:20 I think a lot of us in this space, we're
seeing this concept of sync & merge as
0:02:26 a, in a way sort of core technical aspect
of Local-First that a lot of people
0:02:32 who are looking to build local-first
were interested in to deliver some
0:02:36 of the maybe sort of harder benefits
around instant modern user experience.
0:02:43 Being a good architecture to build
modern software on, whereas Local-First
0:02:47 is this sort of broader concept, which
includes really interesting motivating
0:02:52 aspects of things like data privacy,
which are not necessarily kind of on the
0:02:56 kind of critical path if you're sort of
building out a startup kind of in a hurry.
0:03:01 and so sort of focusing on sync
is a slightly more pragmatic take
0:03:05 on some of the local-first ideas.
0:03:08 I think why it's so interesting at
the moment is if you just sort of lean
0:03:13 back, it's just such an interesting time
to be alive as a software developer.
0:03:17 So much of the software stack
is just being reinvented
0:03:21 and rethought as we, speak.
0:03:24 And so you have all of this, previous
generation software, which is sort
0:03:29 of team-based SaaS, enterprise
software, and it's all essentially
0:03:33 being rewritten as agentic software.
0:03:35 And so there's this sort of huge
move to where, in a way, all software
0:03:38 is becoming agentic software.
0:03:41 or certainly it's going to be like a
large part of that software future.
0:03:44 And just one of the key aspects of
agentic software is if you start off
0:03:48 with say, users of your software.
0:03:51 One of the main drivers for
like local-first or real time
0:03:54 sync architectures in the
first place is multi-user.
0:03:57 So you introduce multiple users and if
they're both doing things on the same
0:04:01 data, you have to have some kind of
sync architecture to keep them in sync.
0:04:05 But with the agentic software.
0:04:06 The moment you introduce agents as
well as users, like all software
0:04:10 inherently becomes multi-user.
0:04:12 And so like agentic software is
naturally collaborative and if you
0:04:16 just sort of step back and think, well,
how am I gonna build these systems?
0:04:20 If you try to build those
on manual data wiring.
0:04:24 It's just crazy and you're gonna
have bugs and stale data and a
0:04:27 whole load of extra boiler plate.
0:04:29 And so this sort of core technology
of, sync or sync engine architecture,
0:04:34 which in a way has emerged
initially as like a component of
0:04:37 real time or local-first systems.
0:04:40 Is now just the key architecture
for building agentic systems.
0:04:44 And it's very hard to imagine how you
could build agent systems without it.
0:04:48 And as all software moves to
agentic, it then really becomes the
0:04:52 key architecture for building this
next generation of modern software.
0:04:56 Right.
0:04:57 That, and that is very interesting since
when we first started talking about like
0:05:01 doing this extra conference that hasn't
even been yet at the center of it since,
0:05:06 like in many months ago, agentic Systems
agentic coding has been sort of like on
0:05:13 the horizon, but it hasn't been really
there yet as prominently as it is today.
0:05:18 So, like the topic and the core focus
of Sync Conf is now even more relevant.
0:05:24 It is not exclusively about agentic
systems, but I think they make it just.
0:05:28 Even more clear that we need
a better answer for like data
0:05:32 management, particularly in a
multi-user or multi-agent way.
0:05:36 Like if you think about users as
a user agent, now we have just
0:05:40 different ones, and that is like
multiplayer, collaborative software
0:05:45 has always been the foundation or one
of the core pillars of local-first.
0:05:49 But like you say, if you put that front
and center as the most important one.
0:05:54 That's the more pragmatic
way of getting there.
0:05:56 So Adam, you as one of the co-authors
of the local-first essay, how do
0:06:01 you feel about kind of tearing
apart those different ideals of
0:06:05 local-first software and now focusing
on one in particular with Sync Conf?
0:06:11 Yeah, well, certainly,
I'll echo what James said.
0:06:13 That's kind of a pragmatism and idealism,
not necessarily against each other, but
0:06:18 rather maybe the proportion in the mix.
0:06:20 Local-first Conf, I think is very,
and local-first, in general is very
0:06:23 driven by those ideals of user agency,
data ownership and that sort of thing.
0:06:28 And what we see is just a better
way for computers to serve
0:06:32 human needs in the long run.
0:06:34 but then something really
interesting started to happen.
0:06:37 At the conferences both years,
but especially this last year,
0:06:41 when we really saw a lot of
people coming to the community.
0:06:44 A lot more interested in those pragmatic
benefits or maybe more compelled by them.
0:06:48 And to me that's, you know, there is
a version where, I don't know, I could
0:06:52 feel, or people who, came from that more
ideological background, which obviously
0:06:56 I do, could feel like, wait a minute, you
know, we're here with these grand ideals,
0:07:00 and you just show up and say, Hey, we want
fast UIs and cheaper hosting bills, and
0:07:03 hey, this works really well with agentic
AI and we say, wait a minute, you're
0:07:07 missing out on those, those grand ideals.
0:07:09 But I actually see it as a big tent thing
that it's actually pretty interesting
0:07:14 and actually pretty awesome that, people
came to this technology or this kind of
0:07:19 architecture for many different reasons,
but we get a lot of different benefits and
0:07:26 people have different reasons to prefer.
0:07:28 I'm here for the pragmatic benefit of just
the simplified stack and the developer
0:07:32 experience, and other people are more
about the user agency and data ownership.
0:07:37 But we all can kind of, we all
come together and I think that.
0:07:39 The coal, that coalition is stronger
and it's an opportunity to expand
0:07:44 the reach and to, yeah, just to
make this way of building software
0:07:48 that we've all fallen in love with,
something that can spread even further.
0:07:52 That makes sense.
0:07:53 So in a way, Local-First has, for
me, always been about sort of like
0:07:58 aspiring to, a greater, way of building
things, to ultimately serve a human,
0:08:06 give them more agency, et cetera.
0:08:08 And it has been a huge source of
inspiration for myself, and I think this
0:08:12 is why we did the Local-first Conference
as well, to broaden that even further,
0:08:16 to give voices to many different people.
0:08:19 Working on interesting things, et cetera.
0:08:21 And, for me, like the core of it
has always been, inspiring makers,
0:08:25 building software that we can do better
and sync engines and data syncing
0:08:32 in general, I think is part of that.
0:08:34 But I think it's the.
0:08:36 biggest bang for the buck, basically.
0:08:38 like this is where you can use off the
shelf technologies today and you get
0:08:42 the, furthest with that technology.
0:08:45 Like there's, to really get
there fully with local-first,
0:08:49 I think very few apps at all.
0:08:51 Like do that full justice
to all seven ideals.
0:08:55 Like if you basically, look at
it through the eyes of, Martin
0:08:59 Kleppmann where he, like, he defines
of like, is it local-first software?
0:09:03 Yes.
0:09:03 No.
0:09:04 I think very few pieces of software
like fully are local-first,
0:09:08 but I think that's fine.
0:09:09 And Sync Conf is like very much
like an, encouragement of that
0:09:14 fact and of that trade off.
0:09:15 Like I continue to be inspired
by the local-first ideals.
0:09:20 I think it makes a lot of sense for
folks to like embrace sync engines like
0:09:25 Electric, like Jazz, like Xero, et cetera,
to build better apps faster and not having
0:09:30 the ideals stand in the way of shipping.
0:09:34 I think that said, right, and I think.
0:09:37 As there's also, there's also a kind of
a game afoot at the moment where working
0:09:43 out the best stack for software because,
so one of the big trends that's having the
0:09:47 software at the moment is agentic, and so
the type of software that we're building
0:09:52 is changing, but at the same time, how
we're building the software is changing
0:09:56 because you have these AI code editors and
LLMs and so the sort of first generation
0:10:02 of those has chosen stacks, like, for
example, Supabase which is awesome.
0:10:06 And you can kind of go a long way there
with like a backend as a service framework
0:10:10 and a database built in and sensible APIs.
0:10:14 But if you have LLMs generating
code with data fetching in it, then
0:10:20 that application that is writing
can quickly turn into spaghetti.
0:10:23 and I know from speaking to a number
of them that, a lot of these leading AI
0:10:27 code editor platforms and LLM platforms
are just looking ahead to go, well,
0:10:31 what actually is the best stack for
the software that we are generating?
0:10:35 And I think.
0:10:36 As well.
0:10:36 That's where you come back to distinctions
between, say, imperative data fetching
0:10:41 and declarative data fetching.
0:10:43 And a sync engine is just like the
core technology to unlock your ability
0:10:47 to move to a declarative model that
allows you to move these elements of
0:10:51 data transfer out of the LLM generated
code domain, which means that A, the
0:10:56 applications that, the LMS generate are
much better, but they're also much more
0:11:00 maintainable and more likely to be able
to move into production and evolve.
0:11:03 That makes perfect sense.
0:11:05 And so this brings me to my next
question, who would be the target
0:11:10 audience for the conference?
0:11:13 So if you're now listening as a
typical localfirst.fm listeners,
0:11:17 maybe you're here for the thinking,
big Pie in the Sky, inspiration.
0:11:23 And you're dreaming of a world
where like all software is purely
0:11:27 local-first, but maybe you're also here.
0:11:29 As a React developer who is currently
using some like TanStack query and
0:11:35 just like gets by building shipping
stuff quickly and you're just looking
0:11:40 for like a bigger lever to ship
stuff even more quickly and giving
0:11:44 people a better user experience.
0:11:46 So my take would be the latter one
is exactly the right target audience.
0:11:52 Someone who ships code like not how
it looks like in three years from now.
0:11:57 But today, and they need to pick something
that doesn't make their life harder but
0:12:02 easier, with ideally no strings attached.
0:12:05 Yeah, I think that's what I've been
telling folks in the Ink & Switch
0:12:08 community and the local-first
community and the sort of folks
0:12:12 that come to Local-First Conf.
0:12:13 In a way, you probably
know a lot of this already.
0:12:17 so I see it as the people who are
maybe more local-first curious.
0:12:21 They've been interested in this,
they've seen the energy in the space.
0:12:24 They hear about these practises,
you know, they use Linear and
0:12:26 they're, how is it so fast?
0:12:28 And then they see it, you know,
Linear CTO giving the talk at
0:12:31 Local-First Conf basically saying
it's because we use a sync engine.
0:12:33 And they go, okay, well I'm interested
in that, but how do I get started?
0:12:36 Where do I get started?
0:12:37 so what I've basically been looking
for is the opportunity to invite
0:12:41 people who are on that edge there.
0:12:42 They maybe are a more, what would
you call a mainstream developer or
0:12:46 just someone who's, you know, focused
on building software with proven
0:12:49 technology, which is absolutely a very
sensical, way to work your career.
0:12:54 But then, you know, you wanna be
aware of things that can yes, help
0:12:57 you build better software, help
you build that software faster.
0:13:00 And as you pointed out, there are
parts of the local-first stack, if
0:13:04 you want to call it that, that are
kind of ready for production today.
0:13:08 There's other parts that are
still more researchy, more
0:13:10 speculative and aspirational.
0:13:11 And so I think that's part of what
this conference is doing is focusing on
0:13:14 those parts that are like, Hey, you can.
0:13:16 You can really use this today.
0:13:18 and the evidence is a lot of
the speakers we have from places
0:13:20 like Figma and Notion and OpenAI.
0:13:22 In fact Notion just launched their offline
support based on a lot of this pioneering
0:13:27 work by the local-first community.
0:13:29 And we're saying, Hey, this big company
has done it for their at-scale product.
0:13:33 one of their engineers who've worked on
it is gonna come and tell you about it.
0:13:36 And this is something
you can practically use.
0:13:38 It's less about the
aspiration and more about.
0:13:41 Here's something you can go back
to your job or your side projects
0:13:44 and start using right now today.
0:13:46 Right.
0:13:46 And I think another way to look
at this as well is like the,
0:13:50 bar is constantly rising, right?
0:13:52 Like in the, baseline expectation of a
consumer using an app like that bar is
0:13:58 constantly going up, like similar to how
back in the days it would've been fine
0:14:03 to have like a mostly static website.
0:14:06 When you click, you see sort of like
the forwarding in 3, 2, 1, and at
0:14:11 some point just needed to be dynamic.
0:14:12 It needed to be fast,
and there was mobile.
0:14:15 Every app needed to have
a mobile app as well.
0:14:18 And I think now the baseline
expectations are changing that
0:14:22 Linear has sort of like normalized.
0:14:25 High quality expectations.
0:14:26 So does Figma, so does Notion, et cetera.
0:14:28 So the next generation of apps now
have to like the entry barrier is now
0:14:34 already that high quality expectation.
0:14:37 And I think the only way how we can
get there, at least that I'm aware of,
0:14:41 is to use sync engine technologies.
0:14:44 And I think that's another way for
people to look at why they should care
0:14:48 about this conference is like, if they
want to build applications that have a
0:14:54 competitive, rational chance of being
adopted, then I think this can be a
0:14:59 critical, component to just like make
faster apps that work collaboratively
0:15:05 since if you don't use that and you try
to build collaborative apps, good luck.
0:15:10 I think, one thing is why we wanted
to, host in San Francisco is that a
0:15:17 lot of this, new stack for agentic
systems is still being figured out and
0:15:20 it's kind of being figured out on the
ground by the people who are building
0:15:24 that software 'cause you have to
almost just dive in to kind of figure
0:15:27 out what the actual problems are.
0:15:29 What do you need to do to
make these things work?
0:15:32 so for Sync Conf, moving from Berlin to
San Francisco is partly because that's
0:15:36 where the people building this next
generation of agentic software are.
0:15:40 And then for the people that we
wanted to come to the conference.
0:15:45 Was almost deliberately to combine
product builders and some of the kind
0:15:50 of leading product builders or people
defining those stacks like Tanner
0:15:53 Linsley from Tan Stack, for example.
0:15:56 with developers actually
building kind of new software
0:15:59 and agentic system on the ground.
0:16:01 And I think if we can sort of have that
cross pollination from this sort of
0:16:05 world of kind of sync engine local-first
technologists and developer tooling.
0:16:09 With practical experience of building
agentic systems, then it may be
0:16:14 that actually it's not necessarily
come to the conference to sort
0:16:16 of find out how it's all done.
0:16:18 It's like, come to the conference
and let's figure it out together
0:16:21 That's a great point.
0:16:21 One of the things I've loved so much
about Local-first Conf last two years
0:16:25 is it really feels like being part of
a community that is in the process of
0:16:29 figuring itself out both what it stands
for, but also what the technology is gonna
0:16:34 be and what the hard parts of the problem
are, and how to, you know, monetize
0:16:38 it or make a, make a living out of it.
0:16:40 How it relates to the current
software that exists today, whether
0:16:44 it's sits side by side or it's more
revolutionary or more incremental.
0:16:48 All of that is, feels like it's
happening very live, and we're hopefully
0:16:51 creating a venue where that can happen.
0:16:53 the three of us here, as well as our
other co-organizers, have our own opinions
0:16:57 about that and maybe their own directions.
0:16:59 We'd like to see things going, but it's
community and it's evolving and in an
0:17:03 organic way that's wandering in all
kinds of interesting directions and
0:17:07 exploring interesting nooks and crannies.
0:17:09 And I see this as a another chance,
again a different place with a different,
0:17:13 slightly different demographic of
speakers, but probably also attendees.
0:17:17 And it's another, which is sort of
exploring another canyon in the idea
0:17:20 maze, another path down that I think
will give kind of fresh perspective
0:17:25 and it's hardly exactly, it's
hardly about here's how you do it.
0:17:31 Speakers Overview
0:17:31 Right.
0:17:32 and I think people are in different,
parts of the spectrum of like how
0:17:36 much they have figured this out.
0:17:38 I think we have like
some real pioneers here.
0:17:40 We have someone from Figma coming
when Figma launched, I think
0:17:44 it launched out of the gate.
0:17:46 Being collaborative, like this is
similar to how it blew me away when
0:17:51 Google Docs launched and I saw like
that blinking cursor and I could
0:17:54 share it with a friend and like we, we
could collaboratively like that magic.
0:17:58 I just recall it so strongly and that
was maybe even stronger for Figma.
0:18:03 Like it blew me away how nice the software
was, but then I could like, share it
0:18:08 with someone and saw their cursor.
0:18:10 Figma has had that for
now, like many, many years.
0:18:13 I, maybe even a decade.
0:18:15 And so Figma has certly figured out
a lot of pieces there, and I think
0:18:20 they also, they have a much richer
vocabulary around that already.
0:18:24 And they've like blogged a lot about
this, with their life graph system.
0:18:28 And so, I think they have some really,
really interesting stories to share there.
0:18:33 so I'm really looking forward
to, that talk in particular.
0:18:37 But we also hear from like many
other speakers, how they're already
0:18:41 using some of those technologies in
their applications and, their stacks.
0:18:46 So I'm, I'm curious whether
there's any kind of particular
0:18:49 talks that you're already looking
forward to and what you're, what
0:18:52 you're particularly looking for.
0:18:55 Yeah, well, I think we mentioned
Figma and Notion already.
0:18:57 They're both such notable examples of
high craft software, but also at scale,
0:19:03 at successful scale, in our industry.
0:19:06 And so hearing from software
engineers who have been working
0:19:08 for a long time very directly on
their CRDT based or sync engine
0:19:12 based systems is gonna be fantastic.
0:19:15 I'm also looking forward
to hearing from Sunil Pai.
0:19:18 Who's at CloudFlare now, but he's
previously was creator of Party Kit and I
0:19:22 think of him as both a local-first og, but
also someone who's very forward thinking.
0:19:26 And I think he is gonna, speak on
the building agentic apps topic.
0:19:30 And, you know, he, he's better qualified
than almost anyone I know to tell, to
0:19:34 talk about that thing of, that we kind
of touched on earlier, which is okay.
0:19:38 Here's how, here's how users are
expecting software to work now.
0:19:42 And it turns out that the sync engine
foundation that we have created already
0:19:46 is actually a really, in addition to all
these other benefits, it has, like the
0:19:49 multiplayer, collaboration by default
or the better developer experience
0:19:53 also actually is the best foundation
for doing this kind of software.
0:19:57 I think for me, the one I'm most
looking forward to is Frank
0:20:00 McSherry from Materialized.
0:20:02 So personally right back in the
beginning, before starting Electric,
0:20:06 when I was first digging into this
space, discovering some of his work
0:20:10 on differential data flow, timely data
flow, building that into Materialize
0:20:15 as like a streaming database product
was one of the big inspirations behind
0:20:19 starting what we are doing with Electric.
0:20:21 And so he has personally invented
and then pioneered and evolved a
0:20:25 lot of the core underlying tech,
but behind, sync technology.
0:20:30 But also I think with the journey
with Materialize, which is now quite a
0:20:34 mature company serving larger customers.
0:20:37 They have evolved from being a streaming
analytics database, to a platform
0:20:44 where you can also build applications
on top of that streaming data.
0:20:48 And so I think his insights both into
the development of the technology.
0:20:52 But also the use cases of the demand
that's sort of pulling them towards
0:20:56 that application development use case
is really telling because rather than
0:21:00 just being like focusing, doubling
down on analytics use case, they're
0:21:04 seeing that their customers want them
to build applications on this pattern.
0:21:08 And so I think understanding about
that would be really interesting.
0:21:10 Yeah, for sure.
0:21:11 I remember looking into Materialize
many years ago, and I've been like,
0:21:15 blown away by like the, similar premise
basically, which is to think about data
0:21:21 more declaratively and less imperatively
so that you can basically, from
0:21:26 your data sources have like a always
up-to-date view, which is kind of like
0:21:31 what sync engines are about as well.
0:21:34 Like, yes, Materialize started out
earlier, more in the analytical space,
0:21:38 but now like, it's kind of like a blurry
line where, where like your analytical
0:21:43 workload and your like experiments
stop and where your application starts,
0:21:48 particularly now, where you kind of have
not just the high quality expectation
0:21:53 for software in terms of craft and like
being collaborative, but also for like
0:21:58 how smart your software should be, right.
0:22:01 I think the days are over where the
expectation can stop by software
0:22:06 being like fully deterministic and
like just a hard coded functions.
0:22:10 But now you kinda, like you, you've
used chat GPT, so like why not, why
0:22:15 shouldn't your other tool like be able.
0:22:18 to operate in that way and fulfill
your, hopes and visions in this way.
0:22:23 And so a lot of what
makes it happen is data.
0:22:27 And I think this is like also the
main theme of the conference is like,
0:22:32 okay, to build awesome apps, we gotta
figure out the data thing and we, we
0:22:36 shouldn't do it like in the 1990s.
0:22:39 But we should do it like
how it's done in the future.
0:22:42 And that starts right now.
0:22:43 And I think this is a super exciting
like case study and perspective
0:22:49 through the lens of Materialize that
I think we can apply on like pretty
0:22:54 much all technological systems.
0:22:56 can take something away from this.
0:22:58 So very excited about this one.
0:23:01 another one I'm personally also
really excited about is having
0:23:04 Carl Sverre, on the conference.
0:23:07 he has in the past built a system called
SQL Sync, which is a really, really
0:23:13 interesting sync engine based on SQLite,
which is like as there are multiple
0:23:19 technologies in the local-first or sync
engine space hovering around SQLite, me
0:23:24 building also one of them I think he has
a particularly interesting approach by
0:23:29 tackling this in a very low level way.
0:23:32 he's worked in the past on different
database engines or he's like
0:23:37 really like a distributed systems
engineer through and through.
0:23:41 And he's been working on some new
systems as well called Graft, which is
0:23:46 basically a sync engine for memory blocks.
0:23:50 That can, for example, empower
a system like SQL Sync.
0:23:54 So I think he's cooking some
really, really interesting
0:23:57 building blocks here that other
systems could be built on top of.
0:24:01 And you can build arbitrary collaborative
applications on top of this.
0:24:06 So if you build the next Figma,
maybe you build it on top of that.
0:24:10 So also really curious to, hear
more from, from Carl on this.
0:24:15 I think with that you see some
cross-fertilization between some of the
0:24:18 concepts where, so one of things that's
really interesting about Graft is a lot of
0:24:23 sync engines are syncing logical changes
where it's more like physical replication.
0:24:28 But of course, that's kind of the same
technology underpinning serverless
0:24:31 databases, things like Neon, for
example, which has obviously recently
0:24:36 been bought by Databricks, kind of
because they were brought into this
0:24:38 lake-base kind of concept, which
is almost hybrid sort of analytics.
0:24:43 So LTP workloads for app building,
which is exactly what we just
0:24:46 discussed around Materialize.
0:24:48 So actually some of these sort of
like sync concepts are actually
0:24:52 sort of fusing with some of the
developments in cloud and analytics.
0:24:57 But they also represent potentially
very different approaches.
0:25:00 My chat with Carl was basically
he, he's kinda like, ah, the
0:25:04 CRDT application level thing.
0:25:06 No, no, no.
0:25:06 Do it at the, you know,
the lower physical level.
0:25:08 And obviously there can be different
technologies for different use cases.
0:25:12 So it's not necessarily that they're
in competition, but it to that point of
0:25:15 like, what is the right way to accomplish?
0:25:18 What we wanna accomplish
in our applications.
0:25:20 We're all figuring it out together.
0:25:21 And here are some different people
who are on the cutting edge of
0:25:24 trying each possible way of doing it.
0:25:28 And let's kind of compare and contrast
and, learn from each other and get
0:25:31 inspired and even compete a little bit.
0:25:33 I think that's part of
what makes it exciting.
0:25:36 Yeah, exactly.
0:25:37 And I think about data, it's always
the ultimate, like it depends.
0:25:41 So you're building your system, you have
like your trade-offs that you should
0:25:45 leverage and then around those trade-offs,
pick the best data architecture, the best
0:25:50 implementation, the best off-the-shelf
sync engine or roll your own.
0:25:54 And I think someone like
Carl has like over the years.
0:25:58 Deeply studied those trade-offs and then
has picked one that particularly spoke
0:26:04 to him and started going all in on
this to, a level of like excellence and
0:26:09 like focus on performance, et cetera,
that I think is like hard to match.
0:26:14 So anyone who's coming to the conference
and you're trying to figure out like, how
0:26:18 do we build maybe our own sync engine?
0:26:20 Should we build our own sync engine?
0:26:22 Carl is definitely a person, to, speak to.
0:26:25 And I think that's also one of the
main questions to figure out, like,
0:26:28 Hey, should we, for our product,
should we build our own sync engine?
0:26:32 The answer might be yes, or should we
use an off-the-shelf technology talking
0:26:36 to other practitioners is it a stupid
idea to build your own sync engine or
0:26:41 in which cases is it actually necessary.
0:26:44 So I think that's one of the, key
questions that technical decision
0:26:48 makers can find the answer for than in
no other place than at this conference.
0:26:52 I think.
0:26:53 So I'm, pretty excited about this
and another person I'm very thrilled
0:26:58 to have is my, friend Lee Byron, who
is one of the co-creators of GraphQL.
0:27:03 This is how I got to know him and spent
a lot of time with him in my past life.
0:27:08 But these days he's working at
OpenAI and he's actually shipped
0:27:13 some notable parts of ChatGPT
particularly the Canvas feature.
0:27:18 And so the canvas feature is, like already
exhibiting one of the things that you've
0:27:24 mentioned, James, which is basically that
the AI is collaborating with us humans.
0:27:30 And so when you're using chat
GPT, you're collaborating on
0:27:35 an artifact on this canvas.
0:27:37 And so this is, to my understanding,
actually driven by a either a CRDT
0:27:43 or OT based system, like the details.
0:27:46 We'll get to learn at the conference talk,
but this is where OpenAI had with least
0:27:52 help shipped their own implementation of
this, that powers this experience that
0:27:57 many of us use on a day-to-day basis.
0:28:00 And I'm really excited to hear his
perspective since he is obviously a
0:28:05 connoisseur when it comes to good data
management and tasteful data management.
0:28:10 So I'm very curious what,
what he's crafted there.
0:28:14 I think OpenAI as a product is also
really interesting in showing that
0:28:20 just this evolution of like how LLM
models are being productized, where
0:28:25 earlier generations you had a kind
of fundamental model and an API, but
0:28:29 the actual software architecture now
in front of them with routing layers
0:28:33 and complex parts of that software
is just increasing and increasing.
0:28:38 And that is a domain, which
is like those are some of the
0:28:41 most complex agentic systems.
0:28:42 And so understanding the architecture
of how to build those is then really,
0:28:46 interesting to kind of inform how to build
other systems on similar architecture.
0:28:51 And I think as we're seeing, a lot of
it is about being able to, have different
0:28:55 parts of the system agents, services,
users kept in sync with the same data.
0:29:00 So another speaker I'm really
looking forward to is you, Adam.
0:29:05 Adam, you are giving a talk as well.
0:29:08 No one told me about this.
0:29:11 So what can we look
forward to in your talk?
0:29:14 Yeah, I think the way I'm thinking
of it is I do think it's still
0:29:17 important to represent kind of those
local-first ideals and again, people
0:29:21 that come to this may be from a more.
0:29:24 How do I build something
today, pragmatic perspective.
0:29:26 Maybe it's a chance for them to also
see those more aspirational things.
0:29:30 So I think I'll probably focus on
some of those local-first fundamentals
0:29:35 for kind of, for this audience and
perhaps somewhat updated to 2025.
0:29:39 So that's probably the way I'll go.
0:29:41 I mean, given that amazing lineup
you just mentioned, and by the way,
0:29:44 there's some we haven't even mentioned
already, not to mention our CFP, which
0:29:47 is I think, closing in a few days.
0:29:49 I don't know how we're gonna fit it all
in, but certainly, I will be just one
0:29:54 voice among many, but I hope to again,
at least be a, a chance to plant that seed
0:29:59 for those that want to go a little deeper
in some of the idealistic, aspirational,
0:30:03 kind of long-term community pieces.
0:30:06 Yeah.
0:30:06 And just to go through the list
that we have of like speakers
0:30:09 that are already announced.
0:30:11 we also have Tanner Linsley who I
think, Electric actually has been
0:30:16 collaborating with more closely lately.
0:30:19 So at Electric we've been working
with Tanner and the TanStack team on
0:30:23 TanStack db, which is a new reactive
client store built into TanStack
0:30:27 for building applications on sync.
0:30:30 so TanStack is a popular framework
for building web and mobile apps,
0:30:34 and it's evolved out of React Query.
0:30:37 Originally a kind of layer for.
0:30:39 doing some of the sort of shared
boilerplates around query based
0:30:42 apps, talking to APIs where you
have query caching and validation
0:30:46 and then mutation primitives.
0:30:48 and Tanner and my co-founder Kyle, have
both been kind of thinking about the
0:30:55 developer experience of building on React
Query and TanStack for a long while.
0:31:01 And kind of envisaging a more local-first
approach with a more comprehensive,
0:31:07 transactional, optimistic mutation APIs as
a kind of ideal way of building software.
0:31:13 So I think like, like a lot of Tanner's
perspective on this is as somebody
0:31:17 who has great taste around software
should be built and has built one of
0:31:21 the best frameworks for doing that,
and with TanStack db, the collaboration
0:31:25 has been a sort of way to go, can we
build sync into that framework in a
0:31:28 way that still adheres to his sort
of quality of developer experience?
0:31:33 And I know that both Kyle and Tanner just
we're all very excited about what we've
0:31:37 been able to create with TanStack db.
0:31:38 So I think he's going to be talking
through some of the specifics
0:31:41 there around what we've built.
0:31:43 Also just his view generally around
how software should be built, ideally
0:31:47 on a sync engine architecture.
0:31:49 Right.
0:31:49 Yeah.
0:31:49 I'm really excited for that.
0:31:51 I mean, I had the honor to have Tanner on
the podcast a little while ago where we
0:31:56 talked about TanStack DB in more depth,
and I think Tanner is really like, for
0:32:02 me, representative of the target audience
that I envision for the conference.
0:32:08 Like the pragmatists with taste and the
aspiration for building high quality
0:32:15 products and doing so in a way unlike
me how I'm building Overtone, which
0:32:20 still hasn't shipped, but I'm trying to
like get out the last bit of quality.
0:32:26 And for me, it's about the journey as
well, but I think TanStack as a ecosystem
0:32:32 that's for pragmatists, that's for people
who wanna ship, not tomorrow, but today.
0:32:36 And I'm really looking forward
to like seeing how we gonna
0:32:41 get the best of both worlds.
0:32:42 That's like pragmatic software.
0:32:44 With a new approach of sync engine,
so really excited for Tanner's
0:32:48 contribution there as well.
0:32:50 we also have a slightly different
perspective, probably less from
0:32:53 a practitioner's take here.
0:32:56 we have Martin Casado from Andreessen
Horowitz, who's in the past been
0:33:01 like absolutely pivotal for many
layers of the stack in like from
0:33:06 software defined networking.
0:33:08 these days he's a VC at Andreesen
Horowitz, where he has a pretty
0:33:12 good perspective on the larger
software ecosystem and like
0:33:16 infrastructure tools in particular.
0:33:18 So he has been looking at sync engines
for quite a while and that broader space.
0:33:24 And so you're gonna also hear
his take on that ecosystem, which
0:33:29 I'm really excited about as well.
0:33:31 At Local-First Conf, we always try to make
sure we represent at least a little bit
0:33:34 the business side, whether it's bringing
up companies who have been successful
0:33:38 at this, like Ditto or someone who can
speak about the, essentially the different
0:33:43 business models and monetization.
0:33:44 'cause that's definitely one of the
biggest questions people have is that
0:33:48 SaaS and cloud and kind of data hoarding
you might call it, has been such an
0:33:53 incredible business model in the last
decade or so, and it's just really
0:33:56 clear how you make money from that.
0:33:58 but people have more maybe concerns
or reservations about Local-First.
0:34:02 I think in the end, actually the business
model looks pretty similar to SaaS.
0:34:05 Realistically.
0:34:06 but like, so having someone come in
who can come at it from a hardnosed
0:34:10 investor perspective, again, they're
less, in this case, they're less
0:34:13 motivated by, I want a better developer
experience or by any kind of idealism,
0:34:17 but more just thinking where is a good
place to make money, where successful,
0:34:21 the next generation of successful
software business is gonna be built.
0:34:25 Yeah, so there will be more talks, but
I don't think they're yet announced.
0:34:31 And there's still an open CFP as
well, which may be this your last
0:34:35 chance to try to speak at the
conference by applying, there will
0:34:40 be a selection committee, et cetera.
0:34:42 we already got way more
submissions than we have space for.
0:34:45 But that being said, please take
your chance to apply as I think
0:34:50 there is a quite an interesting
range of different topics that should
0:34:54 be represented at the conference.
0:34:56 So, there's more to talk about, but
that's probably at a separate time.
0:35:01 So maybe.
0:35:02 rounding out.
0:35:03 there's still some tickets
available to my knowledge.
0:35:07 so for people interested coming to the
conference, check out the conference
0:35:11 website on syncconf.dev, where I think the
early bird tickets are already sold out.
0:35:18 But they're still the normal bird and late
bird, and they're, I think they're, the
0:35:22 birds are flying out pretty quickly, so
you should get your chance to pick yours.
0:35:28 and yeah, with that, maybe some
last words from James and Adam on
0:35:33 what you're looking forward most.
0:35:35 Yeah, I'll just add onto the call to
action there to say, you know, not
0:35:38 just submit your talk if you want,
or buy a ticket if you want to come.
0:35:41 But also, if you know someone that,
again, is kind of in that demographic
0:35:45 of local-first curious, but has never
really taken the plunge you wanna
0:35:49 invite them, you know, maybe this is
the right opportunity for them to come
0:35:53 into this world we all love so much.
0:35:56 and I'll also note that I think
we're almost full on sponsors, but
0:35:59 we maybe have one more space left.
0:36:01 So if you wanna do that,
that's a possibility as well.
0:36:04 You can just email us.
0:36:05 yeah, and as for your question of,
yeah, what am I looking forward to
0:36:09 most, that's, that's really tough.
0:36:10 I'll tell you what I'm not looking
forward to is that Johannes, typically
0:36:14 you and I have done the blocking out of
the schedule and figuring out the time
0:36:18 slots and the breaks and all that stuff
and all of these great speakers and
0:36:22 cramming all of this into one day in a
way that also leaves room for hallway
0:36:26 track and, you know, breathing space to
think about what you've just learned.
0:36:30 That is a challenge I am
not looking forward to.
0:36:33 But once we once we conquer that little
milestone, I'll be very much looking
0:36:38 forward to attending and I think just
like a local-first comp being there.
0:36:42 Not only hearing the talks, but the
ideas that are provoked from that
0:36:46 and hearing what people are building
and how the talks kind of change
0:36:50 or inspire their, change, their
perspective or inspire them on that.
0:36:54 I just have learned so much at
our conferences these last two
0:36:57 years, and I feel like this one.
0:36:59 That is in a new place with a slightly
different type of person is also gonna
0:37:03 be a similar kind of, you know, mind
blow emoji, expand the perspective new,
0:37:08 new way to think about this space.
0:37:10 That even though I've been thinking
about it for over a decade somehow,
0:37:14 I always feel like I learned
something new at these conferences.
0:37:16 I think you kinda looking at a space
like this, you get a sense of just the
0:37:20 kind of timing of the ideas emerging.
0:37:23 And certainly from my perspective,
I just see people both talking
0:37:28 about sync more on social media.
0:37:30 I see a lot of predictions that this
is the key technology for the next
0:37:34 years or more, and I suspect like
this is a one day conference, right?
0:37:39 It's a kind of maximum 300 people.
0:37:41 And my hope, and what I kind of think
is that we'll look back on it in
0:37:45 10 years time and see this as just
one of the seminal conferences that
0:37:48 actually the group of people who were
there all then went on to do amazing
0:37:52 things and you kind of look back
and it was a privilege to be there.
0:37:55 So I think that's the kind of
event that we're trying to create.
0:37:59 and it kind of just feels like
the timing is right there.
0:38:02 It's such an exciting time in software
development and I think that this is
0:38:05 just gonna be one of the most, sort
of special occasions to be at that
0:38:09 we'll look back on fondly in time.
0:38:12 Outro
0:38:12 I couldn't agree more.
0:38:14 Did we actually say actually when it
is November 12th, 2025 in San Francisco
0:38:22 and the website is syncconf.dev.
0:38:24 That's great.
0:38:25 Awesome.
0:38:26 Well, thank you so much for coming on
the show to discuss this, and I'm looking
0:38:31 forward to seeing the two of you at the
conference in November in San Francisco.
0:38:37 But I also hope to see many of
our listeners at the conference.
0:38:42 Can't wait to reconnect
with this community.
0:38:45 Awesome.
0:38:46 All right.
0:38:46 See you.
0:38:46 See
0:38:47 you then.
0:38:48 Thank you for listening to
the localfirst.fm podcast.
0:38:50 If you've enjoyed this episode and
haven't done so already, please
0:38:54 subscribe and leave a review.
0:38:55 Please also share this episode
with your friends and colleagues.
0:38:58 Spreading the word about the
podcast is a great way to support
0:39:01 it and to help me keep it going.
0:39:04 A special thanks again to Jazz
for supporting this podcast.