WEBVTT
Kind: captions
Language: en

00:00:01.309 --> 00:00:08.309
[music]

00:00:12.500 --> 00:00:14.100
>> Hey everybody. Thanks for attending my

00:00:14.100 --> 00:00:16.200
talk. I had to put this together late

00:00:16.200 --> 00:00:18.200
last night. I've been in a whirlwind

00:00:18.200 --> 00:00:20.900
after this project went viral of

00:00:20.900 --> 00:00:22.800
unblocking thousands of open source

00:00:22.800 --> 00:00:25.800
developers. We've had VC's approaching

00:00:25.800 --> 00:00:28.000
us. I've been in constant meetings. Like

00:00:28.000 --> 00:00:29.500
this has been

00:00:29.500 --> 00:00:32.300
a sleepless month of activity since this

00:00:32.300 --> 00:00:34.800
happened. So if you're not familiar,

00:00:34.800 --> 00:00:37.000
I kind of cloned Adobe Photoshop and a

00:00:37.000 --> 00:00:39.700
bunch of other apps and that went super

00:00:39.700 --> 00:00:40.800
viral.

00:00:40.800 --> 00:00:43.100
My phone, my inbox, everything has just

00:00:43.100 --> 00:00:46.400
been absolutely lit up and so I I I

00:00:46.400 --> 00:00:48.500
apologize for the structure of the the

00:00:48.500 --> 00:00:50.400
the the talk. This was thrown together

00:00:50.400 --> 00:00:51.800
as quickly as I can, but I've got a lot

00:00:51.800 --> 00:00:54.200
of really exciting stuff to talk about

00:00:54.200 --> 00:00:57.700
and I think it talks about the future of

00:00:57.700 --> 00:00:59.300
not only the industry, but just

00:00:59.300 --> 00:01:02.000
technology as a whole. So

00:01:02.000 --> 00:01:03.600
we're going to open source the entire

00:01:03.600 --> 00:01:06.400
world. Like why not? It's it's something

00:01:06.400 --> 00:01:08.100
that I've always wanted to do as a

00:01:08.100 --> 00:01:09.600
developer. That's something that

00:01:09.600 --> 00:01:11.200
a lot of us in the open source community

00:01:11.200 --> 00:01:15.700
are interested in. Open source powers

00:01:15.700 --> 00:01:18.100
a lot of the you know, technology that

00:01:18.100 --> 00:01:19.400
we're familiar with, but it's never been

00:01:19.400 --> 00:01:21.300
at the forefront. Like Linux has always

00:01:21.300 --> 00:01:23.400
been like a second class citizen. We've

00:01:23.400 --> 00:01:26.200
been pulled into a whole bunch of

00:01:26.200 --> 00:01:27.400
platforms where we don't really have

00:01:27.400 --> 00:01:30.400
ownership and AI kind of gives us a

00:01:30.400 --> 00:01:33.300
chance to take the web back, if you

00:01:33.300 --> 00:01:36.800
will. So

00:01:36.800 --> 00:01:39.400
I might have to do this manually.

00:01:39.400 --> 00:01:41.300
Cool. So a little bit about myself. I've

00:01:41.300 --> 00:01:43.800
been an engineer for about 15 years. I

00:01:43.800 --> 00:01:47.000
spent eight years at Square doing heavy

00:01:47.000 --> 00:01:50.600
hitting infra. So online payments,

00:01:50.600 --> 00:01:52.900
the the authentication stack, basically

00:01:52.900 --> 00:01:54.000
anything that could take down the

00:01:54.000 --> 00:01:56.000
company my team was kind of in charge of

00:01:56.000 --> 00:01:57.600
and it was you know, eight years of

00:01:57.600 --> 00:01:58.700
carrying the pager, a lot of really

00:01:58.700 --> 00:02:00.892
heavy hitting infrastructure

00:02:00.892 --> 00:02:02.292
Um and then I've spent seven years doing

00:02:02.292 --> 00:02:03.892
startups, and that's a whole different

00:02:03.892 --> 00:02:05.992
world. It's a lot of risk, a lot of

00:02:05.992 --> 00:02:08.092
learning like what the market needs, a

00:02:08.092 --> 00:02:11.192
lot of correcting your own assumptions.

00:02:11.192 --> 00:02:13.092
And I've also been um I'm based in

00:02:13.092 --> 00:02:15.292
Atlanta. It's a big film film state.

00:02:15.292 --> 00:02:17.492
I've been very big in the arts. So out

00:02:17.492 --> 00:02:19.892
of college, made lots of films, built

00:02:19.892 --> 00:02:21.492
laser projectors that play video games

00:02:21.492 --> 00:02:22.892
on the side of skyscrapers, that kind of

00:02:22.892 --> 00:02:25.292
thing. And those are the sort of

00:02:25.292 --> 00:02:27.192
customers that our startups or the

00:02:27.192 --> 00:02:28.592
startups that I built have have catered

00:02:28.592 --> 00:02:30.992
to, creatives, artists, filmmakers, and

00:02:30.992 --> 00:02:32.192
so forth.

00:02:32.192 --> 00:02:33.092
Um

00:02:33.092 --> 00:02:34.892
I want to transport everybody back in

00:02:34.892 --> 00:02:37.392
time, back to the very beginnings of the

00:02:37.392 --> 00:02:40.492
web and the internet, the '90s, the the

00:02:40.492 --> 00:02:42.992
the early 2000s. So for us millennials,

00:02:42.992 --> 00:02:46.292
we're going to feel old. And for

00:02:46.292 --> 00:02:48.092
um everybody that's that's younger, this

00:02:48.092 --> 00:02:49.792
is just let me kind of set the stage for

00:02:49.792 --> 00:02:51.592
what the internet was like.

00:02:51.592 --> 00:02:53.692
Um we had a bunch of chat applications

00:02:53.692 --> 00:02:56.192
that were completely open. So IRC is

00:02:56.192 --> 00:02:57.892
like Discord, but you could spin up your

00:02:57.892 --> 00:02:59.592
own servers that you actually own. You

00:02:59.592 --> 00:03:01.892
can use whatever programs to automate

00:03:01.892 --> 00:03:03.892
them and interrogate them without rules.

00:03:03.892 --> 00:03:05.992
All of the commercial

00:03:05.992 --> 00:03:07.892
messaging apps were basically they

00:03:07.892 --> 00:03:09.392
didn't care if you used third-party

00:03:09.392 --> 00:03:11.292
tools to connect to them.

00:03:11.292 --> 00:03:13.892
Every website had data exports. They had

00:03:13.892 --> 00:03:15.992
APIs. You could literally build really

00:03:15.992 --> 00:03:18.992
cool mashups with Reddit and Flickr and

00:03:18.992 --> 00:03:21.392
Twitter. And these were like trending on

00:03:21.392 --> 00:03:23.492
slashdot before Hacker News was a thing.

00:03:23.492 --> 00:03:25.492
And then, you know, in in the more

00:03:25.492 --> 00:03:27.892
modern times, these kind of trickle down

00:03:27.892 --> 00:03:29.492
to

00:03:29.492 --> 00:03:31.892
a much slower pace because the APIs got

00:03:31.892 --> 00:03:33.292
got removed.

00:03:33.292 --> 00:03:35.892
There was a bunch of file sharing. So

00:03:35.892 --> 00:03:37.492
not not just file sharing, you could

00:03:37.492 --> 00:03:39.692
actually buy MP3s and feel a sense of

00:03:39.692 --> 00:03:41.792
ownership in in everything that you were

00:03:41.792 --> 00:03:43.592
doing on the web. And now it's more

00:03:43.592 --> 00:03:45.292
licensing.

00:03:45.292 --> 00:03:46.092
Um

00:03:46.092 --> 00:03:48.092
there was so much open creativity

00:03:48.092 --> 00:03:49.392
because largely the people on the

00:03:49.392 --> 00:03:51.092
internet were technical. They were early

00:03:51.092 --> 00:03:53.592
adopters. So Flash kids were making all

00:03:53.592 --> 00:03:56.292
kinds of animations. There were diffuse

00:03:56.292 --> 00:03:58.492
communities with PHP

00:03:58.492 --> 00:04:01.292
PHPBB and Envision and just just a ton

00:04:01.292 --> 00:04:02.994
of creativity that was happening at that

00:04:02.994 --> 00:04:04.594
time.

00:04:04.594 --> 00:04:06.794
Wikipedia got started sort of in this

00:04:06.794 --> 00:04:09.694
ethos, and we don't really see this kind

00:04:09.694 --> 00:04:12.194
of stuff anymore in the current era and

00:04:12.194 --> 00:04:14.294
incarnation of tech.

00:04:14.294 --> 00:04:16.694
Websites were highly customizable, like

00:04:16.694 --> 00:04:18.294
MySpace at the time. You could actually

00:04:18.294 --> 00:04:20.594
change the HTML and CSS and make it look

00:04:20.594 --> 00:04:23.394
however you want. So this was a totally

00:04:23.394 --> 00:04:26.594
different alien time compared to what we

00:04:26.594 --> 00:04:30.094
have today.

00:04:30.094 --> 00:04:31.694
When Web 2.0 happened, broadband,

00:04:31.694 --> 00:04:33.494
smartphones, it brought everybody into

00:04:33.494 --> 00:04:35.394
the fold. So the technical people got

00:04:35.394 --> 00:04:37.394
sort of pushed to the sidelines. And

00:04:37.394 --> 00:04:38.794
it's not a bad thing. Like we needed

00:04:38.794 --> 00:04:39.694
our, you know, our parents and our

00:04:39.694 --> 00:04:40.894
grandparents to be able to use these

00:04:40.894 --> 00:04:43.494
products. But we sort of we rounded the

00:04:43.494 --> 00:04:44.994
edges on things. We took away all the

00:04:44.994 --> 00:04:46.994
technical pieces. We made everything

00:04:46.994 --> 00:04:49.094
easy for people to adapt because they

00:04:49.094 --> 00:04:51.594
couldn't contribute from a technical

00:04:51.594 --> 00:04:53.294
footing. They didn't want to unpackaged

00:04:53.294 --> 00:04:55.394
things. They didn't know what APIs were.

00:04:55.394 --> 00:04:57.994
And so the platforms and the the the

00:04:57.994 --> 00:04:59.694
protocol like protocols got replaced by

00:04:59.694 --> 00:05:03.294
platforms. And they sort of grabbed as

00:05:03.294 --> 00:05:04.994
many users as they could. And sort of

00:05:04.994 --> 00:05:06.994
the diffuse, experimental early web

00:05:06.994 --> 00:05:08.594
started to disappear.

00:05:08.594 --> 00:05:11.394
The APIs got taken away from us.

00:05:11.394 --> 00:05:13.494
And then we started seeing distribution

00:05:13.494 --> 00:05:16.394
being gatekept by the big providers.

00:05:16.394 --> 00:05:19.694
Like we never once had a phone ecosystem

00:05:19.694 --> 00:05:21.694
where we could freely distribute to like

00:05:21.694 --> 00:05:25.094
like we could on the desktop.

00:05:25.094 --> 00:05:27.794
Things started to cost more and more as

00:05:27.794 --> 00:05:30.494
as startups began to like win market

00:05:30.494 --> 00:05:31.694
share. They realized that they had

00:05:31.694 --> 00:05:33.394
pricing power and

00:05:33.394 --> 00:05:34.994
suddenly your Netflix actually became

00:05:34.994 --> 00:05:36.694
cable once you started subscribing to

00:05:36.694 --> 00:05:38.794
all the streaming services.

00:05:38.794 --> 00:05:40.894
And then they started to get very

00:05:40.894 --> 00:05:43.194
extractive. For companies that maybe

00:05:43.194 --> 00:05:45.094
maybe hit the limits of their growth,

00:05:45.094 --> 00:05:47.594
they started coming up with crazy crazy

00:05:47.594 --> 00:05:50.094
gotchas. So the one that got me is Adobe

00:05:50.094 --> 00:05:51.394
has this

00:05:51.394 --> 00:05:54.694
you can subscribe to a monthly plan and

00:05:54.694 --> 00:05:55.994
they don't tell you this. It's in like

00:05:55.994 --> 00:05:58.194
the very small print. If you cancel your

00:05:58.194 --> 00:06:00.594
monthly plan before the 12 months or 18

00:06:00.594 --> 00:06:02.594
months, they actually charge you up to

00:06:02.594 --> 00:06:05.194
like $200 to unsubscribe, which is

00:06:05.194 --> 00:06:06.695
really inconvenient If you're going to a

00:06:06.695 --> 00:06:09.195
film festival like once a year, making a

00:06:09.195 --> 00:06:11.095
short film, and then suddenly you've got

00:06:11.095 --> 00:06:13.295
an extra $200 bill for an application

00:06:13.295 --> 00:06:14.695
that you were just using over a weekend.

00:06:14.695 --> 00:06:16.695
And that's that's hit so many people and

00:06:16.695 --> 00:06:18.595
it's really frustrating

00:06:18.595 --> 00:06:21.395
that the web that was once really really

00:06:21.395 --> 00:06:24.095
promising and open and free

00:06:24.095 --> 00:06:26.995
has turned into this

00:06:26.995 --> 00:06:28.595
extractive

00:06:28.595 --> 00:06:33.195
owned rented rent-seeking world.

00:06:33.195 --> 00:06:35.695
So it it used to be a very special place

00:06:35.695 --> 00:06:37.895
in the '90s in the 2000s

00:06:37.895 --> 00:06:39.195
and

00:06:39.195 --> 00:06:42.495
more and more has been extracted.

00:06:42.495 --> 00:06:45.095
This is a term that's commonly used in

00:06:45.095 --> 00:06:47.195
shitification. I think that aptly

00:06:47.195 --> 00:06:51.095
describes everything that happened.

00:06:51.095 --> 00:06:53.195
Our company started with the ethos of

00:06:53.195 --> 00:06:54.995
let's build everything open source. Like

00:06:54.995 --> 00:06:56.695
the applications that we built in the AI

00:06:56.695 --> 00:06:58.395
space were

00:06:58.395 --> 00:07:01.295
100% open, not only the desktop apps but

00:07:01.295 --> 00:07:03.795
the actual infrastructure as well. So

00:07:03.795 --> 00:07:05.595
Artcraft originally started as sort of

00:07:05.595 --> 00:07:07.295
like a professionally minded

00:07:07.295 --> 00:07:08.995
Higuysfield. So all of the the

00:07:08.995 --> 00:07:10.595
filmmakers that I know and that I've

00:07:10.595 --> 00:07:12.495
I've I've grown up with, we were making

00:07:12.495 --> 00:07:14.695
a product for them to use rather than a

00:07:14.695 --> 00:07:16.995
UGC centered approach like like

00:07:16.995 --> 00:07:19.895
Higuysfield or or Open Art or so forth

00:07:19.895 --> 00:07:22.895
have. And that enabled us to give them a

00:07:22.895 --> 00:07:25.795
desktop tool that lived on their machine

00:07:25.795 --> 00:07:26.995
where all the files that they were

00:07:26.995 --> 00:07:28.695
generating lived right next to their

00:07:28.695 --> 00:07:29.795
editor.

00:07:29.795 --> 00:07:32.595
And one of the things that we observed

00:07:32.595 --> 00:07:34.195
and my advisors thought I was absolutely

00:07:34.195 --> 00:07:35.595
crazy that artists, you know, they don't

00:07:35.595 --> 00:07:36.895
know what open source is. They don't

00:07:36.895 --> 00:07:38.195
they don't need it. They don't talk

00:07:38.195 --> 00:07:40.095
about it. You know, isn't really

00:07:40.095 --> 00:07:42.395
used by artists. We actually observed

00:07:42.395 --> 00:07:45.095
artists vibe coding their own workflows

00:07:45.095 --> 00:07:46.495
on top of the software that we gave

00:07:46.495 --> 00:07:47.995
them. And that is something that like

00:07:47.995 --> 00:07:50.195
was brand new to all of us. Like

00:07:50.195 --> 00:07:52.495
there were micro drama studios that were

00:07:52.495 --> 00:07:54.195
doing order fulfillment for some of the

00:07:54.195 --> 00:07:57.195
top micro drama startups. They would do

00:07:57.195 --> 00:07:59.495
30 to 50 grand of fulfillment every

00:07:59.495 --> 00:08:01.695
single month and they had big teams

00:08:01.695 --> 00:08:03.395
and no product out there fits their

00:08:03.395 --> 00:08:04.595
needs. Like nobody's going to go and

00:08:04.595 --> 00:08:07.295
workshop and build that specifically for

00:08:07.295 --> 00:08:09.497
the type of delivery that they need. Yet

00:08:09.497 --> 00:08:11.697
they in house have people that know

00:08:11.697 --> 00:08:13.897
cloud code and they can take the APIs

00:08:13.897 --> 00:08:15.397
and they can generate an interface

00:08:15.397 --> 00:08:16.897
that's like that. So we started

00:08:16.897 --> 00:08:18.597
experimenting with completely open

00:08:18.597 --> 00:08:20.897
source, giving it to our customers, and

00:08:20.897 --> 00:08:23.397
they really responded to that. And then

00:08:23.397 --> 00:08:24.597
we thought,

00:08:24.597 --> 00:08:26.197
what if we could turn back time on

00:08:26.197 --> 00:08:28.097
everything? Like all of these things

00:08:28.097 --> 00:08:29.897
that they're consuming and building

00:08:29.897 --> 00:08:32.697
would be much more friendly and and and

00:08:32.697 --> 00:08:35.697
give them way more capacity to create if

00:08:35.697 --> 00:08:37.697
all of the infrastructure was open

00:08:37.697 --> 00:08:40.097
source, if all of the application shapes

00:08:40.097 --> 00:08:41.897
that they wanted to leverage were open

00:08:41.897 --> 00:08:44.197
source. Like they use Premiere, they use

00:08:44.197 --> 00:08:46.097
Adobe. Like what could what could we do

00:08:46.097 --> 00:08:47.297
to give them something that they could

00:08:47.297 --> 00:08:49.497
actually take away all of the surface

00:08:49.497 --> 00:08:51.597
area that they don't use and constrain

00:08:51.597 --> 00:08:54.197
the apps to their specific workflow?

00:08:54.197 --> 00:08:55.397
What would it look like if they could do

00:08:55.397 --> 00:08:58.097
that, you know, five or 10 times for

00:08:58.097 --> 00:08:59.597
each different artist on their team

00:08:59.597 --> 00:09:01.597
that's doing a particular part of that

00:09:01.597 --> 00:09:04.597
workflow? And so we wanted to kind of

00:09:04.597 --> 00:09:07.197
undo what had happened during Web 2.0

00:09:07.197 --> 00:09:09.397
and everything became owned and and

00:09:09.397 --> 00:09:11.797
consolidated and and

00:09:11.797 --> 00:09:13.897
gate kept and just give people back

00:09:13.897 --> 00:09:17.197
their applications. And AI really is the

00:09:17.197 --> 00:09:20.497
means for us to do that. So we took a

00:09:20.497 --> 00:09:23.997
stab at the Adobe Creative Suite.

00:09:23.997 --> 00:09:25.697
We looked at Photoshop first. We

00:09:25.697 --> 00:09:27.897
basically prompted, you know,

00:09:27.897 --> 00:09:29.897
this is the basic shape of a photo

00:09:29.897 --> 00:09:31.797
editing app. You you've consumed and

00:09:31.797 --> 00:09:33.497
trained on all of the tutorials about

00:09:33.497 --> 00:09:35.297
what Photoshop is. That's in your

00:09:35.297 --> 00:09:36.697
training data.

00:09:36.697 --> 00:09:39.797
We we laid out a large large prompt to

00:09:39.797 --> 00:09:42.097
say let's let's build out the the

00:09:42.097 --> 00:09:44.197
application. Let's separate the UI UX

00:09:44.197 --> 00:09:46.697
from that. Let's make everything MCP and

00:09:46.697 --> 00:09:49.297
CLI compatible. Let's build out plans

00:09:49.297 --> 00:09:51.297
for every single tool, how all of those

00:09:51.297 --> 00:09:52.697
work. Let's talk about the data

00:09:52.697 --> 00:09:54.697
structures, the file format importers.

00:09:54.697 --> 00:09:56.397
We basically built these large prompts

00:09:56.397 --> 00:10:00.397
and within 24 hours we were looking at a

00:10:00.397 --> 00:10:02.097
pretty good approximation of what

00:10:02.097 --> 00:10:04.897
Photoshop was. And it it kind of blew

00:10:04.897 --> 00:10:06.497
our minds away.

00:10:06.497 --> 00:10:08.197
It was functional. It was very familiar

00:10:08.197 --> 00:10:11.198
to us. It started 10 times

00:10:11.198 --> 00:10:12.698
than the original Photoshop. Now we

00:10:12.698 --> 00:10:14.798
didn't like actually cover the entire

00:10:14.798 --> 00:10:16.198
surface area of Photoshop. There's a lot

00:10:16.198 --> 00:10:18.698
of work. We're still not at 100% parity,

00:10:18.698 --> 00:10:20.598
but we're measuring ourselves against

00:10:20.598 --> 00:10:22.598
what has what needs to be done and we're

00:10:22.598 --> 00:10:24.398
getting closer and closer. We were at

00:10:24.398 --> 00:10:27.298
50%. We've just passed 60% and in the

00:10:27.298 --> 00:10:29.598
fullness of time like Pareto principle,

00:10:29.598 --> 00:10:32.098
we could reach 99.99%

00:10:32.098 --> 00:10:34.498
parity. Um

00:10:34.498 --> 00:10:35.898
I'm going to take a little bit of a sort

00:10:35.898 --> 00:10:37.398
of a diatribe. Our team has talked about

00:10:37.398 --> 00:10:38.998
whether we take this further now or

00:10:38.998 --> 00:10:41.098
whether we just go for parity. And right

00:10:41.098 --> 00:10:43.198
now we're just we want to have we want

00:10:43.198 --> 00:10:45.698
to fulfill the muscle memory of all the

00:10:45.698 --> 00:10:47.198
artists that use Adobe. So we are going

00:10:47.198 --> 00:10:49.598
for Adobe like their sort of interface

00:10:49.598 --> 00:10:51.198
and look and feel directly. In the

00:10:51.198 --> 00:10:53.298
fullness of time we will build net new

00:10:53.298 --> 00:10:54.898
apps that have new interfaces that are

00:10:54.898 --> 00:10:56.998
more agentic. Um but we wanted to do

00:10:56.998 --> 00:10:58.398
that also as sort of like a proving

00:10:58.398 --> 00:10:59.798
function that we have actually

00:10:59.798 --> 00:11:01.698
accomplished something real completely

00:11:01.698 --> 00:11:02.998
agentically.

00:11:02.998 --> 00:11:06.398
Uh so we did Photoshop. We did um

00:11:06.398 --> 00:11:09.098
basically all of their apps. We did

00:11:09.098 --> 00:11:13.098
InDesign. We did Premiere. Um and uh

00:11:13.098 --> 00:11:14.198
this

00:11:14.198 --> 00:11:16.698
I I did not expect this to go viral. I

00:11:16.698 --> 00:11:17.998
was withholding posting it on Hacker

00:11:17.998 --> 00:11:20.098
News. It got out on Reddit and then like

00:11:20.098 --> 00:11:22.398
two days later like the like New York

00:11:22.398 --> 00:11:23.898
Times was asking me for an interview and

00:11:23.898 --> 00:11:26.398
our Discord went from 3,000 people to

00:11:26.398 --> 00:11:27.698
It's now got more people in our Discord

00:11:27.698 --> 00:11:31.298
than Jev, Suno, Higgfield. It

00:11:31.298 --> 00:11:33.298
I I'm still struggling to bear with what

00:11:33.298 --> 00:11:34.798
has happened. This was a It struck a

00:11:34.798 --> 00:11:37.198
nerve with people. Uh they've felt sort

00:11:37.198 --> 00:11:38.698
of maligned by Adobe and other tech

00:11:38.698 --> 00:11:41.498
companies and uh what we've built has

00:11:41.498 --> 00:11:43.098
has really resonated with them and it's

00:11:43.098 --> 00:11:44.198
sort of

00:11:44.198 --> 00:11:45.698
built up sort of this mission to build

00:11:45.698 --> 00:11:49.098
more. Uh so all of the anti-AI artists

00:11:49.098 --> 00:11:50.698
who have been pushing back against AI

00:11:50.698 --> 00:11:52.898
all of this time have you know they've

00:11:52.898 --> 00:11:54.498
sort of turned around and they're like

00:11:54.498 --> 00:11:56.598
we actually love what you're doing.

00:11:56.598 --> 00:11:57.798
They've started using these apps.

00:11:57.798 --> 00:11:59.198
They've started giving us like feature

00:11:59.198 --> 00:12:01.498
requests and bug reports and we're

00:12:01.498 --> 00:12:03.698
engaging them and I think we're going to

00:12:03.698 --> 00:12:05.198
be able to teach them this is what an

00:12:05.198 --> 00:12:07.298
MCP is. This is how you agentically

00:12:07.298 --> 00:12:08.698
create things and maybe we'll start

00:12:08.698 --> 00:12:10.498
pulling away some of these bloated

00:12:10.498 --> 00:12:12.098
interfaces and giving them things that

00:12:12.098 --> 00:12:13.599
are faster and more fluid, and let them

00:12:13.599 --> 00:12:16.799
do their work faster. Um, so we didn't

00:12:16.799 --> 00:12:18.299
stop with those apps. We've done all

00:12:18.299 --> 00:12:20.399
kinds of stuff. We've done CAD, we've

00:12:20.399 --> 00:12:25.199
done um uh uh uh uh Flash, we've done uh

00:12:25.199 --> 00:12:27.799
a bunch of digital audio workstations.

00:12:27.799 --> 00:12:29.499
It's not just us, it's the community

00:12:29.499 --> 00:12:30.999
doing these things as well. And there's

00:12:30.999 --> 00:12:33.399
really no limit in terms of what can be

00:12:33.399 --> 00:12:35.299
done. Uh so we've done the entire

00:12:35.299 --> 00:12:36.899
Microsoft well the core apps within the

00:12:36.899 --> 00:12:38.899
Microsoft Office suite. There's been

00:12:38.899 --> 00:12:41.899
Google Docs. We have uh companies and

00:12:41.899 --> 00:12:43.499
universities and all sorts of people

00:12:43.499 --> 00:12:45.299
approaching us wanting to use these

00:12:45.299 --> 00:12:47.799
things. And we're we're basically trying

00:12:47.799 --> 00:12:50.199
to fulfill all of these different asks.

00:12:50.199 --> 00:12:51.899
Um they're being used on Linux where

00:12:51.899 --> 00:12:54.599
they've never been used before. Um PDF

00:12:54.599 --> 00:12:55.899
Craft is probably the best app. You

00:12:55.899 --> 00:12:56.899
should all go download it right now.

00:12:56.899 --> 00:12:58.899
You'll never have to use Acrobat again.

00:12:58.899 --> 00:13:01.199
This is like really fast,

00:13:01.199 --> 00:13:03.388
really smooth.

00:13:03.388 --> 00:13:06.099
[laughter]

00:13:06.099 --> 00:13:08.699
It works on Linux, it works on NetBSD.

00:13:08.699 --> 00:13:10.299
It's it's brilliant. Like this one's

00:13:10.299 --> 00:13:11.699
ready to go. So go ahead and get this

00:13:11.699 --> 00:13:14.899
one. Um we did Figma. We've actually

00:13:14.899 --> 00:13:16.999
done the server component for this too

00:13:16.999 --> 00:13:19.599
and you can basically have your team

00:13:19.599 --> 00:13:21.499
collaborate on on documents. It's got

00:13:21.499 --> 00:13:23.999
like the raft paxos like multi-user

00:13:23.999 --> 00:13:25.799
setup. So we're starting to go into

00:13:25.799 --> 00:13:27.299
infrastructure too. And you know I the

00:13:27.299 --> 00:13:29.199
entire infrastructure and data model all

00:13:29.199 --> 00:13:30.599
of it is open source. We're not keeping

00:13:30.599 --> 00:13:32.399
any of it proprietary.

00:13:32.399 --> 00:13:33.899
Um

00:13:33.899 --> 00:13:35.299
there's a community that's working on

00:13:35.299 --> 00:13:37.199
Solid Craft that's led by engineers.

00:13:37.199 --> 00:13:40.099
They've actually got uh an open

00:13:40.099 --> 00:13:41.999
uh engineering kernel that manages all

00:13:41.999 --> 00:13:44.199
of this. Uh and it's it's really solid.

00:13:44.199 --> 00:13:46.299
Like the MCP you can actually do 3D

00:13:46.299 --> 00:13:48.299
printing projects really quickly with

00:13:48.299 --> 00:13:49.499
Solid Craft.

00:13:49.499 --> 00:13:50.999
Um

00:13:50.999 --> 00:13:53.299
we've had so so much happen. We've got

00:13:53.299 --> 00:13:56.499
over 3,000 300,000 uh GitHub stars.

00:13:56.499 --> 00:13:58.899
4,000 different people have been

00:13:58.899 --> 00:14:00.199
contributing to this project and it's

00:14:00.199 --> 00:14:02.299
picking up. It's it's it's actually

00:14:02.299 --> 00:14:04.499
become sort of a movement if you will.

00:14:04.499 --> 00:14:06.399
And we're we're trying to adapt to how

00:14:06.399 --> 00:14:08.599
much influx of of people trying to

00:14:08.599 --> 00:14:09.699
contribute to this. I'm going to pick up

00:14:09.699 --> 00:14:11.299
the pace cuz I've been kind of lagging

00:14:11.299 --> 00:14:14.299
on this. But um uh we're We're about 1.7

00:14:14.299 --> 00:14:16.001
million downloads a day and this is

00:14:16.001 --> 00:14:18.101
accelerating. We've got a launcher app

00:14:18.101 --> 00:14:19.901
that puts us on the desktop, and like

00:14:19.901 --> 00:14:21.301
you can you can basically manage

00:14:21.301 --> 00:14:22.601
everything here. It works on all

00:14:22.601 --> 00:14:24.901
operating systems.

00:14:24.901 --> 00:14:26.901
A quarter of our users are on Linux,

00:14:26.901 --> 00:14:28.901
which is really fascinating. Like Linux

00:14:28.901 --> 00:14:30.501
has been historically really, really

00:14:30.501 --> 00:14:33.101
underserved, and now they have super

00:14:33.101 --> 00:14:36.001
high quality professional apps for

00:14:36.001 --> 00:14:39.201
creative and and office work, which is

00:14:39.201 --> 00:14:41.601
it's really nice. So

00:14:41.601 --> 00:14:43.901
we've you know, we we love and respect

00:14:43.901 --> 00:14:45.301
all of the work that's been done there

00:14:45.301 --> 00:14:46.801
in the past, but people are really

00:14:46.801 --> 00:14:48.201
excited to use the apps that they didn't

00:14:48.201 --> 00:14:51.601
have access to.

00:14:51.601 --> 00:14:53.701
All of these things are MCP and CLI like

00:14:53.701 --> 00:14:55.601
first class. You can totally command

00:14:55.601 --> 00:14:57.401
these, and you can watch them build and

00:14:57.401 --> 00:14:58.901
assemble creative projects right in

00:14:58.901 --> 00:15:00.601
front of you.

00:15:00.601 --> 00:15:02.401
They've been localized into languages

00:15:02.401 --> 00:15:04.101
that were never like officially

00:15:04.101 --> 00:15:06.501
supported by Adobe or Microsoft. Like we

00:15:06.501 --> 00:15:08.701
were told that Arabic was not supported

00:15:08.701 --> 00:15:11.101
in InDesign. We've also supported these

00:15:11.101 --> 00:15:13.701
apps past their support lifecycle. Like

00:15:13.701 --> 00:15:16.101
Adobe Animate was being

00:15:16.101 --> 00:15:18.001
end of life or it was being like put on

00:15:18.001 --> 00:15:19.401
life support. We've we've picked that

00:15:19.401 --> 00:15:20.801
up. We've had basically animation

00:15:20.801 --> 00:15:23.001
studios come to us saying, we use this

00:15:23.001 --> 00:15:25.301
for our jobs. Like this is important.

00:15:25.301 --> 00:15:28.201
Please please support this. So a lot of

00:15:28.201 --> 00:15:29.801
a lot of really fascinating things are

00:15:29.801 --> 00:15:31.101
happening where we're getting this into

00:15:31.101 --> 00:15:32.201
more places than they've ever been

00:15:32.201 --> 00:15:34.201
before. It's it's definitely not me.

00:15:34.201 --> 00:15:36.001
Like I I started this and now we've got

00:15:36.001 --> 00:15:38.801
a community of like 4,000 developers.

00:15:38.801 --> 00:15:40.901
This is taken on a life of its own. They

00:15:40.901 --> 00:15:43.401
are doing everything from CI/CD. It's

00:15:43.401 --> 00:15:45.001
it's incredible. This this is way bigger

00:15:45.001 --> 00:15:46.301
than just me.

00:15:46.301 --> 00:15:47.601
I really got to pick up the pace here.

00:15:47.601 --> 00:15:51.001
I've made way too much content. So

00:15:51.001 --> 00:15:52.601
there are 300 apps being developed by

00:15:52.601 --> 00:15:54.801
the community. We've developed an app

00:15:54.801 --> 00:15:56.501
where non-technical people can start

00:15:56.501 --> 00:15:58.301
contributing and use the basis of the

00:15:58.301 --> 00:16:00.001
apps that we've created like the UI/UX

00:16:00.001 --> 00:16:02.001
libraries. We're decomposing this into

00:16:02.001 --> 00:16:03.901
all of the the components and primitives

00:16:03.901 --> 00:16:05.601
so that you can reuse them. So you'll

00:16:05.601 --> 00:16:07.201
have nice

00:16:07.201 --> 00:16:09.301
3D libraries, nice 2D libraries. It's

00:16:09.301 --> 00:16:11.101
all great. I'll get into the technical

00:16:11.101 --> 00:16:13.401
stuff. So

00:16:13.401 --> 00:16:15.301
Rust was probably essential in this

00:16:15.301 --> 00:16:16.701
product's product success. We'd seen

00:16:16.701 --> 00:16:18.502
people doing it in Electron before, but

00:16:18.502 --> 00:16:20.602
for high-quality creative apps, you

00:16:20.602 --> 00:16:23.102
really need something that's native. Um,

00:16:23.102 --> 00:16:24.602
there's just a lot of really, really

00:16:24.602 --> 00:16:26.902
good quality things about Rust.

00:16:26.902 --> 00:16:28.802
Google had papers about like the low

00:16:28.802 --> 00:16:31.102
defect rate. It starts really fast. It's

00:16:31.102 --> 00:16:33.902
it's the best language for this.

00:16:33.902 --> 00:16:35.802
Uh,

00:16:35.802 --> 00:16:37.502
yeah. Um,

00:16:37.502 --> 00:16:40.602
it supports all all different platforms

00:16:40.602 --> 00:16:42.102
and operating systems. We've got this on

00:16:42.102 --> 00:16:44.602
NetBSD. We've got this on tablets. It

00:16:44.602 --> 00:16:47.902
natively deploys to the web via Wasm, so

00:16:47.902 --> 00:16:50.802
it can get anywhere and everywhere.

00:16:50.802 --> 00:16:52.902
Um, do you need to know Rust to start

00:16:52.902 --> 00:16:55.502
using Rust today as an engineer? Uh,

00:16:55.502 --> 00:16:57.502
definitely not. Uh, I've been using Rust

00:16:57.502 --> 00:16:59.402
since pre-1.0 times, but everybody on

00:16:59.402 --> 00:17:01.402
the team uh it's a mix of people that

00:17:01.402 --> 00:17:02.502
have never experienced the language

00:17:02.502 --> 00:17:04.402
before. Your agents will drive it just

00:17:04.402 --> 00:17:05.302
fine. You're not going to have any

00:17:05.302 --> 00:17:07.002
trouble. It is a fantastic language. The

00:17:07.002 --> 00:17:08.302
one big thing you're going to run into

00:17:08.302 --> 00:17:11.002
is compile times. Um,

00:17:11.002 --> 00:17:12.502
is it GUI yet? There's this famous

00:17:12.502 --> 00:17:14.302
website or series of websites in the

00:17:14.302 --> 00:17:16.302
Rust Rust world. Does it support like

00:17:16.302 --> 00:17:18.602
modern features? Yes. You can build GUI

00:17:18.602 --> 00:17:21.102
rich GUI apps in in Rust, UI UX. Uh,

00:17:21.102 --> 00:17:22.502
we've run into a lot of gotchas like

00:17:22.502 --> 00:17:24.502
permissions, GitHub, scaling, CI. A lot

00:17:24.502 --> 00:17:25.402
of people have talked about that. I'm

00:17:25.402 --> 00:17:26.702
going to just speed through this cuz I'm

00:17:26.702 --> 00:17:29.002
kind of running out of time. Uh, CI is

00:17:29.002 --> 00:17:30.632
like falling over.

00:17:30.632 --> 00:17:32.202
[laughter] We've been spinning up our

00:17:32.202 --> 00:17:34.002
own. We found a good place to get RAM on

00:17:34.002 --> 00:17:35.902
co-located servers. We're made of that.

00:17:35.902 --> 00:17:38.002
Um, this is a good one. So like agentic

00:17:38.002 --> 00:17:40.002
develop versus human uh review. If

00:17:40.002 --> 00:17:41.702
you're doing backend work or things that

00:17:41.702 --> 00:17:44.002
require schema migrations, API design,

00:17:44.002 --> 00:17:45.702
let your humans do that, but for like

00:17:45.702 --> 00:17:47.702
desktop apps and games and anything

00:17:47.702 --> 00:17:49.702
that's basically immutable, throw your

00:17:49.702 --> 00:17:50.902
agents at it and give them like full

00:17:50.902 --> 00:17:52.402
confidence. You don't need to be human

00:17:52.402 --> 00:17:54.102
in the loop fully. Like the upstream

00:17:54.102 --> 00:17:55.502
like directing where things go, just

00:17:55.502 --> 00:17:57.002
just let it run. Tell it to build

00:17:57.002 --> 00:17:58.702
things. Somebody in our community is is

00:17:58.702 --> 00:18:00.902
building Uh, they've got 40 machines

00:18:00.902 --> 00:18:04.602
rebuilding Unreal Engine in Rust. So you

00:18:04.602 --> 00:18:05.902
you don't even have to mind these

00:18:05.902 --> 00:18:09.502
things. Um, Autocraft is a completely

00:18:09.502 --> 00:18:11.002
autonomous software factory that we're

00:18:11.002 --> 00:18:13.102
going to open source. It ingests issues,

00:18:13.102 --> 00:18:14.602
it triages them, it does agentic

00:18:14.602 --> 00:18:16.502
development, it post-trains any vowels

00:18:16.502 --> 00:18:17.902
on all of the decisions. This will be

00:18:17.902 --> 00:18:20.502
open source soon. Um, basically we've

00:18:20.502 --> 00:18:22.003
got an entire app factory that's

00:18:22.003 --> 00:18:24.203
designed in code. You can express this.

00:18:24.203 --> 00:18:25.603
We'll open source this soon. I'm just

00:18:25.603 --> 00:18:27.303
going to have to hurry cuz I'm overtime.

00:18:27.303 --> 00:18:28.103
Uh

00:18:28.103 --> 00:18:31.003
injustice issues.

00:18:31.003 --> 00:18:33.003
This is open source code rabbit open

00:18:33.003 --> 00:18:34.403
source all the things. We're we're doing

00:18:34.403 --> 00:18:36.503
a lot here and

00:18:36.503 --> 00:18:37.903
I I thought I would save more time for

00:18:37.903 --> 00:18:39.203
the technical, but this is really good

00:18:39.203 --> 00:18:41.503
and please please look out for this. Um

00:18:41.503 --> 00:18:43.003
it will support all of your agentic

00:18:43.003 --> 00:18:45.003
development. It supports all of the sort

00:18:45.003 --> 00:18:47.403
of crafting and curating how you uh

00:18:47.403 --> 00:18:48.803
approach a project. It keeps your

00:18:48.803 --> 00:18:51.503
engineers on task. Um again fully open

00:18:51.503 --> 00:18:53.703
source. All of the decisions made by it

00:18:53.703 --> 00:18:55.003
are training data that we're going to

00:18:55.003 --> 00:18:56.803
use for evals and post training on the

00:18:56.803 --> 00:18:59.303
decision trees. Um

00:18:59.303 --> 00:19:00.403
let's take back the entire internet.

00:19:00.403 --> 00:19:03.203
Let's rebuild everything. All software,

00:19:03.203 --> 00:19:10.203
all platforms.

00:19:10.803 --> 00:19:12.803
We can we can turn all of these things

00:19:12.803 --> 00:19:15.403
into proper protocols. Um we can open

00:19:15.403 --> 00:19:17.503
source the entire world. Let's own our

00:19:17.503 --> 00:19:19.503
apps, our infra, our protocols, our

00:19:19.503 --> 00:19:20.903
models, all of it.

00:19:20.903 --> 00:19:22.603
Um

00:19:22.603 --> 00:19:24.303
we are hiring. So if anybody is

00:19:24.303 --> 00:19:26.103
interested and wants to work on 100%

00:19:26.103 --> 00:19:27.703
open source, none of our code is closed

00:19:27.703 --> 00:19:30.603
source. All of it is in the open. Um uh

00:19:30.603 --> 00:19:31.903
you get time to work on your own

00:19:31.903 --> 00:19:33.903
projects. Just reach out to me at hello

00:19:33.903 --> 00:19:36.303
at storyteller.ai and thank you so much

00:19:36.303 --> 00:19:40.303
everybody.
