Let's Open Source the World

1:20:56I had to put this together late last night.

1:20:58I've been in a whirlwind after this project went viral

1:21:01of unblocking thousands of open source developers.

1:21:05We've had VCs approaching us.

1:21:07I've been in constant meetings.

1:21:09This has been a sleepless month of activity since this happened.

1:21:14So if you're not familiar, I kind of cloned Adobe Photoshop

1:21:18and a bunch of other apps, and that went super viral.

1:21:22My phone, my inbox, everything has just been absolutely lit up.

1:21:26And so I apologize for the structure of the talk.

1:21:31This was thrown together as quickly as I can,

1:21:33but I've got a lot of really exciting stuff to talk about.

1:21:35And I think it talks about the future of not only the industry,

1:21:40but just technology as a whole.

1:21:42So we're going to open source the entire world.

1:21:45Like, why not?

1:21:46It's something that I've always wanted to do as a developer.

1:21:50That's something that a lot of us in the open source community are interested in.

1:21:54Open source powers a lot of the technology that we're familiar with,

1:22:00but it's never been at the forefront.

1:22:02Linux has always been a second-class citizen.

1:22:04We've been pulled into a whole bunch of platforms where we don't really have ownership.

1:22:09And AI kind of gives us a chance to take the web back, if you will.

1:22:15So I might have to do this manually.

1:22:20Cool.

1:22:21So a little bit about myself.

1:22:22I've been an engineer for about 15 years.

1:22:25I spent eight years at Square doing heavy-hitting infra,

1:22:29so online payments, the authentication stack,

1:22:34basically anything that could take down the company,

1:22:36my team was kind of in charge of,

1:22:37and it was eight years of carrying the pager,

1:22:39a lot of really heavy-hitting infrastructure work.

1:22:42And then I've spent seven years doing startups,

1:22:44and that's a whole different world.

1:22:46It's a lot of risk, a lot of learning what the market needs,

1:22:49a lot of correcting your own assumptions.

1:22:52And I've also been, I'm based in Atlanta.

1:22:55It's a big film state.

1:22:56I've been very big in the arts.

1:22:58So out of college, made lots of films,

1:23:01built laser projectors that play video games

1:23:03on the side of skyscrapers, that kind of thing.

1:23:04And those are the sort of customers that our startups

1:23:08or the startups that I built have catered to,

1:23:10creatives, artists, filmmakers, and so forth.

1:23:14I want to transport everybody back in time,

1:23:17back to the very beginnings of the web and the Internet,

1:23:19the 90s, the early 2000s.

1:23:23So for us millennials, we're going to feel old.

1:23:26And for everybody that's younger,

1:23:29just let me kind of set the stage for what the Internet was like.

1:23:33We had a bunch of chat applications that were completely open.

1:23:36So IRC is like Discord,

1:23:38but you can spin up your own servers that you actually own.

1:23:40You can use whatever programs to automate them and interrogate them.

1:23:44without rules. All of the commercial messaging apps were basically, they didn't care if you

1:23:50used third-party tools to connect to them. Every website had data exports. They had APIs.

1:23:56You could literally build really cool mashups with Reddit and Flickr and Twitter. And these

1:24:01were like trending on Slashdot before Hacker News was a thing. And then, you know, in the

1:24:07more modern times, these kind of trickled down to a much slower pace because the APIs

1:24:13got got removed. There's a bunch of file sharing. So not not just file sharing, you can actually

1:24:19buy mp3s and feel a sense of ownership and and everything that you were doing on the

1:24:23web. And now it's more licensing. There was so much open creativity because largely the

1:24:30people on the internet were technical. They were early adopters. So flash kids were making

1:24:35all kinds of animations. There were diffuse communities with PHP, PHP BB and envision

1:24:41and just a ton of creativity that was happening at the time.

1:24:46Wikipedia got started sort of in this ethos,

1:24:49and we don't really see this kind of stuff anymore

1:24:52in the current era and incarnation of tech.

1:24:55Websites were highly customizable.

1:24:58Like MySpace at the time, you could actually change the HTML and CSS

1:25:01and make it look however you want.

1:25:03So this was a totally different alien time

1:25:07compared to what we have today.

1:25:11When Web 2.0 happened, broadband, smartphones, it brought everybody into the fold.

1:25:15So the technical people got sort of pushed to the sidelines.

1:25:18And it's not a bad thing.

1:25:20Like, we needed our parents and our grandparents to be able to use these products.

1:25:23But we sort of, we rounded the edges on things.

1:25:26We took away all the technical pieces.

1:25:27We made everything easy for people to adapt because they couldn't contribute from a technical footing.

1:25:33They didn't want to unpackage things.

1:25:35They didn't know what APIs were.

1:25:36And so the platforms and the protocols got replaced by platforms,

1:25:41and they sort of grabbed as many users as they could.

1:25:46And sort of the diffuse experimental early web started to disappear.

1:25:49The APIs got taken away from us,

1:25:52and then we started seeing distribution being gatekept by the big providers.

1:25:57Like, we never once had a phone ecosystem where we could freely distribute to,

1:26:03like we could on the desktop.

1:26:06things started to cost more and more as startups began to like win market share they realized that

1:26:13they had pricing power and suddenly your Netflix actually became cable once you started subscribing

1:26:18to all the streaming services and then they started to get very extractive for companies

1:26:24that maybe maybe hit the limits of their growth they started coming up with crazy crazy gotchas

1:26:29So the one that got me is Adobe has this, you can subscribe to a monthly plan, and they don't tell you this.

1:26:37It's in like the very small print.

1:26:38If you cancel your monthly plan before the 12 months or 18 months, they actually charge you up to like $200 to unsubscribe,

1:26:46which is really inconvenient if you're going to a film festival like once a year making a short film,

1:26:51and then suddenly you've got an extra $200 bill for an application that you were just using over a weekend.

1:26:56And that's hit so many people, and it's really frustrating.

1:26:59that the web that was once really, really promising and open and free

1:27:05has turned into this extractive, owned, rented, rent-seeking world.

1:27:14So it used to be a very special place in the 90s and the 2000s,

1:27:19and more and more has been extracted.

1:27:23This is a term that's commonly used in shitification.

1:27:27I think that aptly describes everything that happened.

1:27:32Our company started with the ethos of let's build everything open source.

1:27:36Like the applications that we built in the AI space were 100% open,

1:27:41not only the desktop apps, but the actual infrastructure as well.

1:27:44So Artcraft originally started as sort of like a professionally-minded Higgs field.

1:27:49So all of the filmmakers that I know and that I've grown up with,

1:27:53we were making a product for them to use rather than a UGC-centered approach,

1:27:57like Higgs field or OpenArt or so forth have.

1:28:02And that enabled us to give them a desktop tool that lived on their machine

1:28:07where all the files that they were generating lived right next to their editor.

1:28:11And one of the things that we observed, and my advisors thought I was absolutely crazy,

1:28:16that artists, you know, they don't know what open source is.

1:28:18They don't need it. They don't talk about it.

1:28:20You know, GIMP isn't really used by artists.

1:28:22we actually observed artists vibe coding their own workflows on top of the software that we gave them.

1:28:28And that is something that was brand new to all of us.

1:28:31There were micro drama studios that were doing order fulfillment for some of the top micro drama startups.

1:28:38They would do 30 to 50 grand of fulfillment every single month.

1:28:41And they had big teams and no product out there fits their needs.

1:28:45Nobody's going to go and workshop and build that specifically for the type of delivery that they need.

1:28:50Yet they in-house have people that know cloud code and they can take the APIs and they can generate an interface that's like that

1:28:57So we started experimenting with completely open source giving it to our customers

1:29:01And they really responded to that and then we thought

1:29:05What if we could turn back time on everything like all of these things that they're consuming and building would be much more friendly and

1:29:13And give them way more capacity to create if all of the infrastructure was open source

1:29:19if all of the application shapes that they wanted to leverage were open source?

1:29:24Like, they use Premiere. They use Adobe.

1:29:26Like, what could we do to give them something that they could actually take away

1:29:30all of the surface area that they don't use

1:29:32and constrain the apps to their specific workflow?

1:29:35What would it look like if they could do that, you know,

1:29:38five or ten times for each different artist on their team

1:29:41that's doing a particular part of that workflow?

1:29:43And so we wanted to kind of undo what had happened during Web 2.0

1:29:48and everything became owned and consolidated and gatekept

1:29:54and just give people back their applications.

1:29:56And AI really is the means for us to

1:30:00do that.

1:30:00So we took a stab at the Adobe Creative Suite. We looked at Photoshop first. We basically

1:30:07prompted, you know, this is the basic shape of a photo editing app. You've consumed and

1:30:13trained on all of the tutorials about what Photoshop is. That's in your training data.

1:30:18laid out a large, large prompt to say, let's build out the application. Let's separate

1:30:24the UI UX from that.

1:30:26Let's make everything MCP and CLI compatible.

1:30:29Let's build out plans for every single tool,

1:30:32how all of those work.

1:30:33Let's talk about the data structures,

1:30:35the file format importers.

1:30:36We basically built these large prompts.

1:30:38And within 24 hours, we were looking

1:30:40at a pretty good approximation of what Photoshop was.

1:30:44And it kind of blew our minds away.

1:30:47It was functional.

1:30:49It was very familiar to us.

1:30:50It started 10 times faster than the original Photoshop.

1:30:53Now, we didn't actually cover the entire surface area of Photoshop.

1:30:57There's a lot of work. We're still not at 100% parity.

1:31:00But we're measuring ourselves against what needs to be done,

1:31:03and we're getting closer and closer.

1:31:05We were at 50%. We've just passed 60%.

1:31:08And in the fullness of time, like Pareto principle,

1:31:11we could reach 99.99% parity.

1:31:15I'm going to take a little bit of a diatribe.

1:31:18Our team has talked about whether we take this further now

1:31:20or whether we just go for parity.

1:31:21And right now, we want to fulfill the muscle memory of all the artists that use Adobe.

1:31:28So we are going for Adobe, like their sort of interface and look and feel directly.

1:31:32In the fullness of time, we will build net new apps that have new interfaces that are more agentic.

1:31:37But we wanted to do that also as sort of like a proving function,

1:31:40that we have actually accomplished something real, completely agentically.

1:31:44So we did Photoshop. We did basically all of their apps.

1:31:50We did InDesign.

1:31:51We did Premiere.

1:31:54And this, I did not expect this to go viral.

1:31:58I was withholding posting it on Hacker News.

1:31:59It got out on Reddit.

1:32:01And then, like, two days later,

1:32:02like, New York Times was asking me for an interview.

1:32:05And our Discord went from 3,000 people to,

1:32:07it's now got more people in our Discord

1:32:09than Jev, Suno, Higgsfield.

1:32:13I'm still struggling to bear with what has happened.

1:32:15This was, it struck a nerve with people.

1:32:17They've felt sort of maligned by Adobe

1:32:19be in other tech companies.

1:32:21And what we've built has really resonated with them.

1:32:24And it's sort of built up this mission to build more.

1:32:28So all of the anti-AI artists who have been pushing back

1:32:31against AI all of this time, they've sort of turned around.

1:32:35We actually love what you're doing.

1:32:38They've started using these apps.

1:32:39They've started giving us feature requests and bug reports.

1:32:42And we're engaging them.

1:32:44And I think we're going to be able to teach them

1:32:46this is what an MCP is.

1:32:47This is how you agentically create things.

1:32:49And maybe we'll start pulling away some of these bloated interfaces

1:32:52and giving them things that are faster, more fluid,

1:32:55and let them do their work faster.

1:32:57So we didn't stop with those apps.

1:32:59We've done all kinds of stuff.

1:33:00We've done CAD.

1:33:01We've done Flash.

1:33:05We've done a bunch of digital audio workstations.

1:33:09It's not just us.

1:33:10It's the community doing these things as well.

1:33:12And there's really no limit in terms of what can be done.

1:33:15So we've done the entire Microsoft,

1:33:17well, the core apps within the Microsoft Office,

1:33:19Suite, there's been Google Docs.

1:33:21We have companies and universities and all sorts of people

1:33:25approaching us wanting to use these things.

1:33:27And we're basically trying to fulfill

1:33:30all of these different asks.

1:33:31They're being used on Linux, where they've never

1:33:34been used before.

1:33:36PDFcraft is probably the best app.

1:33:37You should all go download it right now.

1:33:38You'll never have to use Acrobat again.

1:33:40This is really fast, really smooth.

1:33:47It works on Linux.

1:33:49It works on NetBSD.

1:33:50It's brilliant.

1:33:51This one's ready to go.

1:33:52So go ahead and get this one.

1:33:55We did Figma.

1:33:56We've actually done the server component for this, too.

1:33:58And you can basically have your team collaborate on documents.

1:34:02It's got the wrapped Paxos multi-user setup.

1:34:06So we're starting to go into infrastructure, too.

1:34:08And the entire infrastructure and data model, all of it, is open source.

1:34:11We're not keeping any of it proprietary.

1:34:15There's a community that's working on SolidCraft that's led by engineers.

1:34:18They've actually got an open engineering kernel that manages all of this.

1:34:24And it's really solid.

1:34:26Like the MCP, you can actually do 3D printing projects really quickly with SolidCraft.

1:34:32We've had so, so much happen.

1:34:34We've got over 300,000 GitHub stars.

1:34:384,000 different people have been contributing to this project, and it's picking up.

1:34:43It's actually become sort of a movement, if you will.

1:34:46And we're trying to adapt to how much influx of people

1:34:50trying to contribute to this.

1:34:51I'm going to pick up the pace, because I've

1:34:52been kind of lagging on this.

1:34:53But we're getting about 1.7 million downloads a day.

1:34:57And this is accelerating.

1:34:58We've got a launcher app that puts us on the desktop.

1:35:00And you can basically manage everything here.

1:35:03It works on all operating systems.

1:35:06A quarter of our users are on Linux, which is really

1:35:09fascinating.

1:35:09Linux has been historically really, really underserved,

1:35:12and now they have super high-quality professional apps

1:35:17for creative and office work, which is really nice.

1:35:22So we love and respect all of the work

1:35:26that's been done there in the past,

1:35:27but people are really excited to use the apps

1:35:29that they didn't have access to.

1:35:32All of these things are MCP and CLI first class.

1:35:35You can totally command these,

1:35:37and you can watch them build and assemble creative projects right in front of you.

1:35:41They've been localized into languages that were never officially supported by Adobe or Microsoft.

1:35:47We were told that Arabic was not supported in NDesign.

1:35:51We've also supported these apps past their support lifecycle.

1:35:55Adobe Animate was being end of life, or it was being put on life support.

1:36:00We've picked that up.

1:36:01We've had basically animation studios come to us saying,

1:36:03we use this for our jobs.

1:36:05This is important.

1:36:06Please, please support this.

1:36:08So a lot of really fascinating things are happening

1:36:11where we're getting this into more places

1:36:13than they've ever been before.

1:36:14And it's definitely not me.

1:36:16Like, I started this, and now we've got a community

1:36:18of, like, 4,000 developers.

1:36:20This has taken on a life of its own.

1:36:22They are doing everything from CI, CD.

1:36:24It's incredible.

1:36:25This is way bigger than just me.

1:36:27I really got to pick up the pace here.

1:36:29I've made way too much content.

1:36:30So there are over 300 apps being developed by the community.

1:36:34We've developed an app where non-technical people can start contributing and use the

1:36:39basis of the apps that we've created, like the UI, UX libraries.

1:36:42We're decomposing this into all of the components and primitives so that you can reuse them.

1:36:46So you'll have nice 3D libraries, nice 2D libraries.

1:36:50It's all great.

1:36:51I'll get into the technical stuff.

1:36:53So Rust was probably essential in this product's success.

1:36:58We'd seen people do it at Electron before, but for high-quality, creative apps, you really

1:37:02need something that's native.

1:37:04There's just a lot of really, really good quality things

1:37:07about Rust.

1:37:08Google had papers about the low defect rate.

1:37:11It starts really fast.

1:37:12It's the best language for this.

1:37:18It supports all different platforms and operating systems.

1:37:23We've got this on NetBSD.

1:37:24We've got this on tablets.

1:37:25It natively deploys to the web via Wasm.

1:37:29so it can get anywhere and everywhere.

1:37:32Do you need to know Rust to start using Rust today as an engineer?

1:37:37Definitely not.

1:37:38I've been using Rust since pre-1.0 times,

1:37:40but everybody on the team,

1:37:41it's a mix of people that have never experienced the language before.

1:37:44Your agents will drive it just fine.

1:37:46You're not going to have any trouble.

1:37:47It is a fantastic language.

1:37:48The one big thing you're going to run into is compile times.

1:37:52Is it GUI yet?

1:37:53There's this famous series of websites in the Rust world.

1:37:57Does it support modern features?

1:37:58Yes, you can build rich GUI apps in Rust, UI, UX.

1:38:02We've run into a lot of gotchas, like permissions,

1:38:04GitHub, scaling, CI.

1:38:06A lot of people talked about that.

1:38:07I'm going to just speed through this,

1:38:08because I'm kind of running out of time.

1:38:09CI is falling over.

1:38:12We've been spinning up our own.

1:38:14We found a good place to get RAM on co-located servers.

1:38:16We're moving to that.

1:38:17This is a good one.

1:38:18So agentic develop versus human review.

1:38:21If you're doing back-end work or things that require schema

1:38:24migrations, API design, let your humans do that.

1:38:26But for like desktop apps and games and anything that's basically immutable, throw your agents at it and give them like full confidence

1:38:33You don't need to be human in the loop fully like the upstream like directing where things go

1:38:36Just just let it run tell it to build things somebody in our community is is building

1:38:40They've got 40 machines rebuilding unreal engine in rust. So you don't even have to mind these things

1:38:49Autocraft is a completely autonomous software factory that we're going to open source it ingest issues it triages them

1:38:55It does agentic development.

1:38:56It post trains and evals on all of the decisions.

1:38:59This will be open source soon.

1:39:01Basically, we've got an entire app factory that's designed in code.

1:39:04You can express this.

1:39:06We'll open source this soon.

1:39:07I'm just going to have to hurry because I'm over time.

1:39:09Injust issues.

1:39:12This is open source code rabbit, open source all the things.

1:39:15We're doing a lot here.

1:39:16And I thought I would save more time for the technical.

1:39:20But this is really good.

1:39:20And please, please look out for this.

1:39:23It will support all of your agentic development.

1:39:25It supports all of the crafting and curating

1:39:28how you approach a project.

1:39:29It keeps your engineers on task.

1:39:32Again, fully open source.

1:39:33All of the decisions made by it are training data

1:39:36that we're going to use for evals and post-training

1:39:38on the decision trees.

1:39:40Let's take back the entire internet.

1:39:42Let's rebuild everything.

1:39:43All software, all platforms.

1:39:52We can turn all of these things into proper protocols.

1:39:56We can open source the entire world.

1:39:58Let's own our apps, our infra,

1:40:00our protocols, our models, all of it.

1:40:04We are hiring, so if anybody's interested and wants to work on 100% open source,

1:40:08none of our code is closed source. All of it is in the open.

1:40:12You get time to work on your own projects, just reach out to me at hello at storyteller.ai.

1:40:16And thank you so much, everybody.