Brian:
[00:00:00] Introduction
Brian: I see a lot of these celebrations, like our success in open source or a lot of startups are starting open source first, uh, where they're like hit their 50, 000 stars and that's great, but like, what does that mean? What we're trying to bring in OpenSauce is, like, what is the story? So I can't just stop at stars, and, like, we're trying to figure out what stars the PRs look like.
Andrew: Hello, welcome to the DevTools FM this is a podcast about developer tools and the people who make them. I'm Andrew. And this is my cohost, Justin.
Justin: Hey everyone, uh, we're really excited to have Brian Douglas on with us today. Uh, so Brian spent some time at GitHub, uh, and we're really excited to talk about that. And also we're really excited to talk about your new endeavors, uh, at OpenSauced, uh, which is just an awesome name, by the way,
Brian: Oh, well, thank you.
Justin: yeah, it's great.
Before we dig into those topics though, uh, would you like to tell our listeners a little bit more about yourself?
Brian: Yeah. Uh, Brian Douglas go by be Dougie on the internet. Um, uh, cliff notes. Uh, I spent almost five years working at GitHub. Uh, we use our handles for everything there. Uh, so most folks kind of discovered or found out about me as be Dougie. So I usually go by that. It's a SEO friendly as well. Um, but previously worked at Netlify as employee number three, um, was like before folks even knew what Netlify was, it was a.
Yeah, it was, it was quite the ride. I spent almost two years, actually over two years there, um, ramping up that, that platform, the front end dashboard, app. netlify. com, uh, and then, yeah, based out here in Oakland. And, uh, I love pizza.
Justin: It's awesome. I, I also, I'm in Brooklyn, but I also love pizza. So, you know, it's great.
Andrew: Yeah. You're pro you're probably the first like business that's not pizza related to have a dot pizza domain. What, what. what made you go with that one? It's like, it's such a very odd domain. So I have to ask the question.
Brian: Yeah. So I think it was like 2017 when the TLD came out, maybe a little bit like maybe 2016, but. Um, I was so I was working at Netlify and I was building a bunch of side projects to like prove the model of like at the time to Jamstack. I know they've kind of like moved away from that a bit, but, um, like prove that you can just do front end focus app and like connect to an API or Firebase.
And, um, I built this little tool to like manage my open source projects around that time. And I needed to add a domain. So I think I had like a, call it like the GitHub companion was like, it was kind of like a CRM for my open source PRS. Um, and that was like kind of generic. Um, so instead I was like open sauce and then that pizza came out.
It's like open sauce, that pizza does sound like it sounds funny. So I picked it up.
Andrew: Yeah, you're right there too. You can make a hot sauce for your company. The easy merch
Brian: we do, we do have a hot open sauce, that pizza, which is our. Our sort of recommendation discovery engine, uh, it's currently, uh, not maintained, but, uh, it's still out there.
Justin: Nice. Nice. I love that. Really good merch opportunities. Uh, we're, we've been thinking about that a little bit ourselves and it's like, you know, uh, fun names are also very memorable and I don't know. think that's, important
Brian: well, our number one referral for traffic is direct typing in the domain. So, like, I, I always mention them. I work for a company called open sauce that pizza because then you'll. You're going to type it. You're like, what are you talking about? Like, we're going to find this out. Also, fun fact. Uh, well maybe this, this is like the, a little bit of secret sauce.
Uh, Jack Dorsey, he actually has like jacket, um, at pizza. com. Or no, it wasn't that it was, um, something that pizza. Uh, so he's actually one of the, the, the, the people behind the pizza TLD.
Justin: Oh, interesting.
Andrew: that that's a level of wealth. I want to get to one day where I can just spin up TLDs just for the fuck of it.
Brian: Yeah, something. I mean, if, if we're successful, uh, when we're successful, I should say, um, yeah, definitely going to be, uh, trying to, I don't know, pepperoni or something. We'll get that one going.
Justin: Yeah, man, that's
awesome.
[00:03:55] Working at GitHub
Andrew: Cool. So before we, uh, dive into the sauce, uh, let's talk about your time at GitHub a little more. Uh, so what did you work on there and, uh, what was it like working at
github
Brian: Yeah. So I joined 2018. Um, so this was like, uh, I'll mention how I joined GitHub and I'll mention what I did. I get hub. So I built open sauce as a side project. And then like six months later, because I built on the graph QL API for GitHub, uh, I spoke at their GitHub universe conference in 2017. Uh, and because of that talk, I got invited to interview and then eventually work at GitHub.
Um, so I joined their DevRel team as the first advocate. Um, so like previously it was always engineers went and spoke at conferences and managed community. Uh, they wanted to establish our actual DevRel team at GitHub. So I joined as a, as the first person on it. And then, uh, I was focused on really this ecosystem engagement.
So API, GraphQL, uh, and that, that led into marketplace integrations to then. They ship get up actions during my first six months at GitHub or the it was like alpha And then I was like axe injection basically for the first couple years at GitHub Would do a lot of examples and building CI and building pipelines, but also fun like little automation tools and then we built a sort of program like the maintainers program which we engage All open source projects, top open source projects to find out what was missing on GitHub.
So I would meet with them pretty regularly. So I feel like at GitHub, DevRel is very different than most companies. And, um, I know we'll get in spicy takes later, but like, I just don't think DevRel is approached properly in most companies. And I think that's why we're seeing a lot of like layoffs. But I say that with a lot of hubris because GitHub didn't have an awareness problem.
Like we could just walk in, I can email github. com emails. Out to people and people will respond and I can get meetings and like, so for that reason, like we were just trying to continue to like, build a relationship with community members and show off cool features to the folks before they launched.
Justin: I bet joining during the time when actions was coming out was pretty exciting because that was a pretty big to me. It felt like a big paradigm shift in the ecosystem for sure.
Brian: Yeah, it's a, it's um, so I just talked to Kyle Daigle on my, my, my podcast, the secret sauce. Kyle's the one who invited me to work. I get up. He's not chief operating officer. So he started as engineer and now he's like CEO, which is like a pretty come up. It's an amazing come up. Um, but he was a team that built the platform ecosystem that set up actions.
Uh, I know Sam Lambert who runs planet scale is the one actually got actions off the ground infrastructure wise. Um, so there's a lot of like X hubber or github mafia that's out and other other other companies. Um, but no, it was really interesting because like the idea and I think a lot of this, Uh, like Jason Warner is like former CTO during that time as well.
He's running poolside, which is like an ML AI startup. But um, the, the goal was to build a platform that people could build on top. So the idea was like glitch was very popular. All these like building like code spaces was another idea that eventually shipped. That was the actual goal for actions. But what they did is they like took off like they scoped it down to like what's the first piece like what's the first principle and it was like people build integrations like GitHub.
How do you make the easier GitHub actions? Um, so that and it's actually the catalyst for GitHub getting acquired by Microsoft because the folks who are running those teams were trying to find like partnerships with compute. So obviously you're talking to Google and talking to Amazon and talking to You to Microsoft, uh, and that's what got Microsoft excited about, Hey, y'all have no CEO right now.
Why don't you, uh, why don't you get acquired? And that's, that's basically the story.
Andrew: Yeah, I've talked to a few people from GitHub now and it's very interesting to me how, like, features come about. Cause it seems like the product briefs for features before they're built are very open ended. Like you said, let's make integrations easier. I talked to Some of the people that are behind GitHub discussions and the prompt for that, I think, was make developers lives easier.
So, the culture there around developing these products seems very interesting to me.
Brian: Yeah. And this is like, this is the tweet that went out early. So Monday GitHub announced, um, it's been refounded on, on, on AI. Um, so the, the conversation right now is like, Oh, does GitHub care about developers? Like what's going on? Git is the foundation of like all collaboration. Like why are they not focused on that?
And I think the truth is like. The, the home of all open source, a home for developers is GitHub. Like we could probably just say that's mostly true. Um, other people who use like Garrett and other random stuff, um, might have another stake in the game. But I think what's happening is, um, and like the way we approach problems, like how eventually there's enough developers on the platform, like 30 million based on stack overflow, a hundred million based on GitHub.
Well, that difference of 70 million is like, well, what's a developer? And, um, so like if you just manage markdown or if you only write docs, are you a developer? You're just a technical writer. Or if you only like put up Figma links and screenshots inside of a GitHub issues, like are you a developer? Who knows?
But like, I think if you have competency within the platform of GitHub, like we consider you as a developer, maybe not an engineer. Uh, so like now with the AI, the Admin AI, like we're seeing more people build random Shopify integrations based on AI or use like some random GPT plugin to traverse some, some web flow screenshots or whatever, like the world, like it's expanding a bit, um, or I guess the universe is expanding, I should say.
Um, so, but like the approach for, for GitHub is like first principles, like find out, okay, folks. Need to collaborate. And I think that one of the biggest things that happened like during the pandemic is like GitHub was already remote friendly and remote first. So like, we doubled down on collaborations. So like things like discussions, like when that shipped, um, it was shipped right before we had a, I think our 2020 GitHub universe, which was only remote, which was the first remote only conference for GitHub.
And all conversations happen in discussion. So like getting massive adoption for like, here's a talk. Here's a discussion. Like I love that you guys are using discussions for the show notes for this podcast. Like that was how all conversation in the conference was happening. And like we had so many other, that was like the first piece.
We had so many other pieces after that. That was going to like build a full on like live streaming on the platform and etc, but, um, not not all plans come to fruition. So, um, now we have discussions and you could still collaborate on GitHub.
Andrew: yeah, it's interesting how AI has, like, changed the game of, like, who is a developer, like, I think at the start of, like, ChatGPT, everybody, rightly so, was afraid, oh, my God, my job's going to go away, but I've always joked that my job, I, I am an engineer, but I'm really more of a plumber, like, I know how to put the, put the pipes together, I know how to, like, look, look at a website and be like, that, that's the bad parts, here's how I fix them.
I think, Our jobs won't go away. There will just be lots more plumbers. Cause if an everyday person can generate a full app, they're, unless we have full AGI, they're not going to be able to go in there and like that, like surgically fix that app that that's, there's still a person for that. And I think that's going to be us going forward.
Brian: yeah, 100 percent because like, I think that so a lot of the folks like heavy hitters in the open source space were kind of like, what's going on GitHub? Like what? Why you refocused off of Git? But I think the question is like, The developers, the whole like, what, 10 years ago is developers are ruling the world.
And they, people wrote books on that and had talks on it. And uh, we're always coming to not to is like developers are still important, but now that designer that can like be a whiz and web flow or in figma or in like other places are now just as important as the developer or that product manager that can like now have justification of like what happened here.
How did we get to this bug? How do we get back to. What happened here or like get blame inside of A. I. It's like absolutely magical. And I think I think folks are spending way too much time trying to generate test and generate descriptions of commits and stuff like that. When I think if someone could generate get blame in a story from there, like that's a better GPT.
By far, that's what actually what get up is going to be what they're shipping in February. So, um, but with that, with that being said, like the world again, the universe continues to expand and I think, I think it's, it's, we're in a weird place where we're seeing like the industry, like we saw a lot of layoffs early this year.
Um, and folks who are making like 200 K and like upstate New York or some random city in Kansas, nothing, nothing against people in upstate New York and Kansas, but like that world is not as realistic, uh, when I can get that same sort of support and some for folks who already have knowledge of the, of the code base with this new copilot chat thing that they, that they should be launching.
Justin: Yeah, it's, it's, it's still an interesting time. I mean, I can definitely see the play from GitHub for sure, because it's like you, you were kind of saying this earlier. You're like, GitHub's a platform for collaboration. Uh, and you know, I mean, I think when I think about open source and I kind of want to talk a little bit more about your project, but before we go to that, like, Open source itself is just like this beautiful thing of like collaboration, iteration, iteration in particular, you know, it's like the people who come in and it's like, I love your project.
There's just something else I would like to do with it. And like, I would like it to do a little bit more or like, what about this? And I think a lot of times we sort of thrive as like creatives in those moments of iteration of like building off something of taking things one next step. And I feel like the really cool thing about AI, the thing that like, I think GitHub is, is.
Leveraging right now is that it gives you something to build off of, you know, it like can generate some things or gives you that next idea or like, you know, plugs into that place where you're stopped. So I was like, it does make sense. It's like, Gives you a, some edgewise automation for, you know, just iteration and that's super powerful.
Uh,
Brian: Yeah, this is the, um, so I was part of the, um, I was on the, like the, the later stage of the copilot launch squad. Um, so like it was like Microsoft said, Hey, GitHub, we invest in open AI, build a thing. A few folks on the team built a prototype, uh, a lot of folks in the, in the company got wind of it and like a, uh, all hands or like this, like a demo that was shared.
Uh, and then they started collaborating, like, Oh, Deverell come help out here and do this things. And like, we were like trying to think of messaging and I think it was kind of like in the final hours of like trying to get that messaging again. And I wasn't like fully in the team, but I was like part of the extended squad to like Help out and the, the story was like, you're, you're, you're basically your coding assistant, like your pair programmer.
And I think like we see a lot of generated art or like I can go to mid journey, but like, give me a picture of me holding a bear, a bear or me holding a beer, like whatever generated picture. Like we think that we can just generate code and that's like, that's the end all, but like, that's not the goal. The goal is like, how can you get back like when we ship actions?
It was like always, how do you get back as fast as possible to the thing you want to do? So if it's not run your test locally all the time, it's like this, push it. It'll run in the PR. You'll have automated testing and all that stuff just works. Uh, and then with copilot, it's, it's find your answer faster.
So like if I am in an office and I ask. The senior engineer or the staff engineer of like, Hey, I don't know what's going on here. Like, why did we make the decision where did this come from? If I can like not take him away from or her away from whatever they're working on to then just ask the question and search it myself.
Now I can get back to fast. I can get back to writing code. They can get back to writing code and like we're all better for it. And I think that's the thing that a lot of people just over index of like, am I going to lose my job? Because now like. I can't spend an hour talking to somebody about something historical record that I could just search for or just find in the documentation.
And that's like the, the other thing is like what the, let me Google that for you is like the, the response, uh, for a lot of folks, like you can Google enough, but now if you now have context in your search based on historical record of like some code base, uh, from like what GitHub has been around since 2007 was first commit.
First thing I did when I was in my onboarding was go back to the first commits and read. Like all the commits, because like I'm already a GitHub is, Oh yeah, here's how to use GitHub. Here's how we interact. Like I already know all this stuff. Like they're teaching marketers how to use GitHub. So I was just searching through commits and like learning about historical record and like big swings in the company just by going to the commit history.
Now imagine you could do that. At GitHub as an employee, be like, what was the last commit that the last CEO did, uh, or does the CEO even write code like that? Like to have that, that question and answer that question pretty quickly. And with that, with commits and have that knowledge is like extremely powerful.
Andrew: Yeah, it'll be exciting to see when it can consume like a large repo and tell me things like that. Cause like the, the code base here at Descript is enormous and just like trying to get that content like that to fit into context is basically impossible. So very excited to see where it all goes.
[00:16:46] Ad
Andrew: Once again, we'd like to thank our sponsors without our sponsors. This podcast wouldn't be possible. This week's episode is sponsored by Ray cast. Ray cast is an app for Mac that's like spotlight. But way. Way better. And you do all the things spotlight does, all the things that Alfred does with a whole lot more built in and a whole lot more you can install what their extensions. They've uh, Very large extension store with everything that you could ever want. And if it's not there, you can build it yourself.
This week, I'd like to cover my own. Raycast wrapped just like Spotify and every other app out there. Ray cast, as I yearn summary for how you use the app. What's cool about Ray cast though, is it's basically a summary of how you use your computer just in general. To nobody's surprise my top opened app and recast is Ray cast aI. Ray cast aI comes with Ray cast pro and it lets you access an LLM with just a keyboard shortcut away. My second, most used app is the emoji app. And I opened that one up a hundred times. I opened up Ray cast AI 600 times. So it's really changed the way I've used my computer.
One scary stat raycast tracks for you is how much time you spent in meetings.
This year, I spent a total of nine full days trapped in zoom calls.
A coworker of mine shared his and he had been in zoom calls for 24 days of the year. Stats are so fun and kind of scary sometimes.
If you want to learn more about Raycast, head over to Ray cast. cast.com. Or if you want to hear more about the product, their methodology behind building it, why they built it you can head over to episode 38, where we interviewed the CEO, Thomas.
Do you want to sponsor the podcast? Head over to dev tools.fm/sponsor to apply.
Do you want to not hear any of these ads anymore? Well, become a member on one of the various platforms we offer memberships. You'll get the episodes. Ad-free. And you get a nifty badge in our discord.
And of course, if you want all of the week's latest and greatest dev tools news, make sure to subscribe to our newsletter, mail.dev tools.fm, and get all the updates about what happened this week. And a little reminder of what happened in the episode and some of our post show thoughts. With that, let's get back to the episode.
[00:18:51] OpenSauced: A Tool for Developers and Businesses
Andrew: Uh, but switching gears here a little bit, let's talk more about open sourced.
So, uh, you mentioned you kind of had like the inklings of the idea before you joined GitHub. Uh, you and me met up at a conf or like a little meetup, and you were saying that another part of the idea was kind of incubated there, GitHub, and was like, oh, a thing we don't wanna build. Uh, so could you tell us what open source is and like, uh, the history of it that led you up to this point?
Brian: Yeah. It's the open source and it's inside the open source. It's really org level insights. So like, again, that context of like descriptive, like, Who's who? What's what's happening within the code base? Like that's what we're working on actively. Like we're supporting, we're looking to support more enterprises, uh, with the product.
But the original idea was like I was an engineer at Netlify. Uh, I was managing, uh, opening up PRs to random projects to make contributions. And then if there's a docs change, I'd always make docs change if like there was anything changed and like I was doing like gulp contributions and a few other JavaScript libraries and uh, I would link Netlify.
Uh, so then if I was like, Hey, I'm doing open source at work during my day job, let me at least justify it, uh, by having a link to like, Hey, I mentioned Netlify in passing. Uh, but then it got to like, I'd have like 20 PRs open. So, like, how do you manage 20 PRs that are open in random projects? Uh, so I built a dashboard in React, uh, to just look at all my PRs and find out, like, what's in flight, who do I need to, like, bump to get a review, uh, and come back in a couple months or a couple weeks to see what's still in flight.
Um, so that was OpenSauce, the sort of GitHub companion. I built that, showed it at GitHub Universe, and showed how, like, basically GitHub GraphQL API worked. Uh, joined, and immediately found out GitHub was working on something like this. Um, so I was like, ah, this is not going to work on this. That project never went, it never saw the light of day.
And, like, GitHub had a lot of, a lot of features that kind of were skunkworks or bottom up or somebody had a sort of fever dream. And worked on it, kind of like GitHub pages being like one of those. Um, so, that reason, I was like, ah, I'm not going to work on this. And then, that project got killed and never actually saw the light of day.
So, I was like, oh, let me pick this up. Uh, so, 2020, uh, I started live streaming on Twitch. Uh, because, actually, it was like right before the pandemic. I was like, I'm just going to start writing code more consistently because my day job. Was it technically day to day writing code? It was like, do examples, work alongside engineering teams and product managers.
Um, so live stream three days a week, building open source, the new version, uh, or the last version. We're currently on the third iteration. Uh, and it was just a polished version. It actually had sign up, log in, uh, some sort of user settings and stuff like that. Uh, fun little project I built in like about a year.
Uh, and then I was like. Talking to a bunch of maintainers and they kept saying there was like this missing feature, uh, which is the GitHub Insights tab is pretty underwhelming today. And, uh, what if you can get more insights into you, not just your one repo, but your actual organization? But then what if you can benchmark that against, like, what's also happening open source?
So, like, what's success? And I've been doing a talk about success in open source, like, If a PR stays open for 30 days, is that good or bad? Who knows? Well, let's look at all the PRs. Like what's the average across the entire repo and like, what's the benchmark for successful projects. And, um, yeah, this had a lot of like really good serendipital, um, conversations with maintainers and then found there were some key metrics that I wanted to build into a platform, uh, that we're building today.
So, uh, yeah, fast forwarding. I, I, I ended up taking a sabbatical last summer for three months. So I spent five years at GitHub and was like, ah, I. I was doing a lot. Like I was, I was managing a dev rel team as a director of dev rel, but I was also pulled into a bunch of other conversations like with co pilot.
I was like the, the person who reached out to a bunch of developers and maintainers to get their feedback before we went live. Um, and then I was, I built the first co pilot demo, um, as well, the public one. Um, so I was just doing a bunch and then we, every, every GitHub universe, I'd always, I was part of the keynote team, uh, would be in the keynotes and built demos for the CEO and stuff like that.
So, I just needed a break, so I'm like, I'm just gonna take off, I'm gonna build, write code for three months and see what happens, and what happened was we got a customer, uh, we got users, and I started hiring a team. Uh, we also raised money as well, um, in that like, literally that, that summer, uh, so I decided not to go back to GitHub and pursue this full time.
Justin: That's really awesome. So, um, I mean, I love the story first off. And, you know, it's, it's interesting as you talk about this is like, I've had similar, sort of similar thoughts and ideas. Uh, it's, there are so many interesting metrics that can be had, especially from the GitHub platform, you know, just like as you're developing, uh, both on, uh, your own like performance of like how, like how quickly you respond to PRs or like, you know, but also, yeah, organizationally, like how, like, I, in a past job, we've like wanted to institute sort of a health check for like things like, is CI pushing too long?
Is there a delay in the, like, when we start a PR to like, when we merge it, like, how can we sort of improve this? And it was the same sort of thing. We had to rebuild it. So, um, I'm curious to just like, know a little bit more specifics around like, what do you think organizations could really leverage this for?
Like, what, what is your superpower? What can you really help with? Uh, yeah,
let start there
Brian: Yeah, I would say like the, the things that you were mentioning, Justin, as far as like, uh, like PRs, time to close and stuff like that. There's a framework called the Dora metric, uh, which has got a lot of backing, a lot of research around of like, so like, like how often when you cut a release, do you have bugs in it?
So like, like that one release, how often does and like the speed of reviews and their speed of the actual getting stuff merge? Does that affect that? So like there's some established metrics around that, but we were focused like before is like more like the structure of your team. Um, so the one thing that we I discovered and the team discovered at GitHub was the correlation to like success and open source.
And like we're going to open source first because there's like a lot of good structure outside of enterprise. Um, new contributors. So like, if you have an influx of new contributors, there's actually a good indication that there's adoption somewhere downstream. Uh, and then if you look at a new contributor, it's like Cooper Daddy's being an example.
I talked to Brendan Burns on my podcast as well, and they had a correlation between issues. Like, they never look at stars, because stars is like a good indication of like awareness, but not an indication of actually usage, downloads, or Contribution or weather stickiness in the sort of industry. Um, so they looked at issues.
Issues are interest. So if you have a correlation of issues up and to the right, it's not a bad thing. It means that there's some interest in some sort of future of this project happening. Um, if you have zero issues, that's great. Pat yourself on the back. But as an open source public. Project like if folks are not constantly giving you feedback in the open, I'm providing conversations like you lose the benefit of even being open source.
You might as well just close the repo and like keep the issues to zero. Um, but PRS is adoption. So when you start looking at the adoption of like contributions and releases that are being cut now, you think some sort of upward trend. And that adoption can correlate to things like stars and like that awareness, uh, but like there's nothing moving and there's no contribution, um, or there's like the last commit like six months ago.
It's not a bad thing. It's just like looking the like the broader ecosystem. So like, for example, javascript typescript, those projects and that tend to release much faster. But if you look at a go project, those releases are much slower. Like there's a little more fine tuning and more consistence or sorry, consideration when it comes to review.
So like there's no like, Okay. Excellent. Like, we're actually putting out some opinions, uh, before the end of the year around, like, what is good, um, but, like, things like having two approvals on a PR, uh, usually gets merged within two weeks, uh, so if you have approvals on a PR, uh, but it still sits, there's probably some conversation or question about, like, why is this thing not getting merged, why was it approved, and then you sort of reverse engineer from there, um, and then that, that correlation of, like, new contributors, uh, that the, so, like, um, So there's like no right answer, so I, I don't want to be like spicy for the sake of being spicy, but like about seven, eight weeks ago, BUN announced their 1.
0, uh, BUN got about 10, 000 stars within the one weekend. Um, about 11 months prior or 13 months prior, they got about 40, 000 stars from their funding announcement. Um, so. The star correlation for BUN, if you start really breaking it down of like issues open, to PRs open, to forks open, um, Downloads is hard because that's something that BUN sees on their end.
Uh, you can kind of see it on the NPM side, which is public. Uh, but it's hard to correlate that as an outsider. Um, but of the 10, 000 stars I have early this year, they had 131 forks. Uh, of the 131 forks, very few of those made into a release. So like, when you think about, if I'm choosing, and this is like, again, Bund's gonna be successful, it's gonna have a longevity in the market, they've got funding.
But when you look at adoption today, like, would I use Bund today, in my project that I'm dependent on, based on those numbers? Probably not. Uh, would I contribute to Bund upstream? Maybe, but there's probably some sort of disconnect in like documentation, awareness, and then the majority of the contributions are happening from the, from Jared and then one or two other engineers.
Um, so then like there's a barrier of entry. So like, maybe Bud's not the best place to go contribute. Maybe it's the best place to watch, uh, and like, wait and see. Um, but I imagine that story might be different next year, uh, based on some other announcements they'll have. Uh, but like, that's just like the, I see a lot of these celebrations, like our success in open source or a lot of startups are starting open source first, uh, where they're like hit their 50, 000 stars and that's great, but like, what does that mean?
Uh, so like picking another, uh, picking another project, I'm not really trying to pick on folks, but like super based another one's 57, 000 stars. They're actually topped in for contributions, um, topped in of all GitHub repos. And the reason for that is they've got a decent sized team, like 50 to 60. Um, I'm, I'm probably way off on that, but, um, they have folks who are consistently contributing in the open as a public repo who are paid by Superbase.
So, like, the correlation of that being the top project with contributions, like, okay, there's, like, a bit more story there. But, like, for showing up and trending, like, they were trending almost every week last year. Um, Like, that's, that's great marketing, that's great DevRel, like, that's great awareness. So they're doing something right, and their adoption, like, has been pretty great as well with, like, PgVector, and they're, they're, they're sort of taking over the Go, GoTrue library and doing authentication.
Like, that's an amazing story, but, like, the, I guess the clarity of what I'm trying to bring, and, like, what we're trying to bring in OpenSauce is, like, what is the story? So I can't just stop at stars, and, like, we're trying to figure out what stars the PRs look like.
Andrew: Are you planning on like adding that sort of data in there too for super base like companies where it's like, well, every single employee of this company commits to open source. So like, they really don't count.
Brian: Yeah, well, I, I wouldn't want to go that far and be like, they don't count. I think that's, I think we should, we should encourage SuperBase to stay open and make contributions and like, SuperBase does a lot of up downstream contributions to other projects as well. And they also, I know SuperBase well, uh, for disclosure, I did Angel invest in their project.
Um. But they also have a, like a, a super squad where they contract contributions and the projects that matter to them, uh, and that helps their support their road map. Um, so I would, I would love more super bases like they, they just kind of really know how to handle, uh, open source and engagement and take on contributions, but also they recruit through that same pipeline as well.
Um, so the idea there is like for super base or for like a Vercel like. We technically need extra permissions from them to help identify for them, like, what's the breakdown of repeat contributors, uh, for Vercel, for Next. js, like, and then that breakdown is who's an employee and who's not an employee, would actually be super useful, uh, but what we're doing right now is we're, we're trying to get partnerships with some of these projects and companies to start sharing that data to showcase more of the story, uh, within the open source ecosystem.
Andrew: So it seems like there's like kind of like two angles to the company right now. There's like the, I'm a developer and I just come to the platform and I can use it in some way. And what we're talking about right now is more of like the, I'm a business and I come to the platform and I use it in some way.
Yeah. So like what are the tools that you're providing and like what value do you think that provides to the business?
Brian: Yeah, and this is like, if you go into the website today, it's very much geared towards the individual developer. And the reason for that is like, I always, my goal for open source is always to grow the pie, like literally. And like more people contribute to open source is better for open source. So if I can convince senior engineer at Netflix to like make a contribution during the week on work time, like.
We're, we're providing more success and more contributions downstream and like we need more senior engineers contributing consistently and giving up some of their free time for that can always just be hacked to profess an early new contributors and good first issues. Um, so like what I found and discovered on the, so I mentioned I had a version of open sauce and I was live streaming.
That was the version of new contributor make contributions. But I find like that the space is crowded for that and no one really goes the next step of like, okay, It's great. Yeah, I get first contribution. What's the second contribution? Uh, so the distinction is like we open sauce does target enterprise and customers, um, our companies because it needs a place for folks to land.
Uh, but also encouraging growing the pie. So like there's a ton of engineers already writing code. Could you help out fix some bugs and then celebrate it within the platform? Um, so the company side is like, you're an open, you're a company, you've already got open source, like a super base or a Vercel, um, let us share more of your story, let's share more of like, what's, what's interacting.
And like the biggest thing that most people fall on their face on when it makes contributions, like I want to contribute to react. Well, what if we could like, take that energy and not put it to react, but put it to something up and coming like a solid or up and coming to like something like react table instead.
And I think a lot of folks just don't know that there's an awareness that you can contribute to an ecosystem, another library with an ecosystem and then help that sustain itself and make contributions. Like the best contribution to make is like. Brand new project that showed up yesterday. That's what the HotOpenSauce.
pizza project is, is like up and coming projects that are getting traction, like new contributors, new stars. Like, go talk to those folks and see if there's like opportunity to build a community from the ground up. And um, so. That, that's the focus, but like the, the tooling is really like we connect to your GitHub organization, uh, you import your team, you can see contributions across your entire organization, but also see where contributions are going out your organization.
So like if you, for example, are a Stripe and you got a contributor onto another project, like now you have a, you have somewhere you like Stripe has a, they're public about this. I think they, they. Well, Sentry is a better example, but you have a budget of just donating to open source projects, and I think donating to just your dependency graph, like a things.
dev. It's great. But like, what if you could donate to contributions for projects that are represented within your employee graph? And what we're trying to do is like build more of that social graph and awareness of I've got a core contributor of some random Python project. Um, we should learn more about this and talk about it and put it on our engineering blog because then we can attract more talent and interest to our company.
And um, that is a, every open source program office, like that's what they need. And uh, that's what we're trying to provide.
Justin: One of the things that we talk a lot about, uh, and, and you just mentioned it is, is open source funding, um, how, you know, we just make open source more sustainable, honestly, that's, that's a big thing. And, and I love, you know, a lot of the things you're talking about here is like, to me, it seems like one of the big values of the platform that you're building is, is like knowing where to spend your time or like having more signals to know where to spend your time, which is a really important thing for like making impact.
Um, And that's really needed because it's like, you know, one of the things, and you've already mentioned this too, one of the valuable things in open source is just people's time. But like funding, funding is, is a thing. Uh, do you have any specific concrete features or plans around like how to help enable funding besides just like prioritization and maybe, you know, connecting people,
Brian: Yeah, I think so. I was also part of, uh, I don't know if I shared the story publicly, but like I've been part of like to get up sponsors awareness campaign and problem space since I joined GitHub. Like I Nadia Ekbald was the original PM on that problem. Uh, and she wrote the book, um, working in public. Uh, yeah.
I would sit in meetings with Nadia, who would, who would talk to maintainers about this problem and discover this, and then Devon Zuegel had joined GitHub right before I got on stage for the DevEx summit, uh, and I introduced Devon Zuegel to their former, uh, Stripes former director at DevRel, uh, and that, that conversation ended up becoming GitHub Sponsors, um, but what I'm getting at The space of sponsorship or funding, it exists.
And I think it's not crowded. Like there's definitely needs to be more support, but I think we've all raced to like, okay, let's just get money, extract money from companies. Uh, but now the maintainer now has money coming through, get up sponsors or whatever open collective. You can like distribute funds and be transparent about it, but like just giving folks money is not.
It's not the end of the solution, and I think we've got that box checked. The challenge that a lot of maintainers have is like when they get sponsorship, especially early new maintainers is what do I do? Like how do I even like do anything with this? Like who do I distribute this to and then what I do to get more funding and justify.
So like for me, I took a angel investment and I sent updates every month to my angel investors like, Hey, we did this, we did that. Here's what we're shipping. Here's a story that we're sharing. And um, Like, what we're trying to do at open sauce is like, how can you tell the story of like, you cut a release, we only able to do this because of this.
So like being able to say, I have a, I'm working on a project, I'm a maintainer. I want to tell the story of like who contributed in this release, what was the story where the conversations. But that becomes like tab on tab on tab on tab on tab of Chrome tabs of trying to like distill that information. Uh, and there's no way to generate that pretty easily.
And what we want to do is you connect your GitHub org, your repos you care about, you generate reports. Um, I actually chatted with the super, uh, super bases about this, like they, they contract folks based on the work done. So like to build an invoice of like, I made these contributions. That's also a manual effort and like, there's no tooling to like, extract information from GitHub or what, what tooling you, whatever you use to say, this is what I did.
Here's the track record of my work. Here's my resume. Like if you will, like we're trying to make it easier for developers to showcase their work. Like this exists in like time tracking software. This exists in Fiverr. If you do stuff there or, um, up, up work or whatever. But it doesn't exist in GitHub and like we want to do is like build a resume, build an invoice, build a sort of like reference doc of like, this isn't the work I've done and that's what we did in open source with profiles last year or I'm saying last year's November, January, we should profiles like I could showcase my work, my last 30 PRS and summarizing like this is this is where I've touched open source.
Um, please give me more. Don't give me more. Whatever the whatever the conversation needs to be. Okay.
Andrew: Yeah, it's, uh, I feel like GitHub has like, it's one really good feature and that's GitHub repos, obviously, uh, but like all the other, all the other features seem like they, they are just like very surface level and kind of gloss over it. Like, I, uh. I have a home feed re implementation that I used to find the tool tips for this podcast because it just doesn't prevent, present the data in the right way to me.
I think GitHub really has an opportunity to like be, to be that platform. And it's just so odd to me that they aren't like, it's like I get a social network out of combining GitHub and Twitter and then I have like kind of half a social network. It seems like open source could kind of almost fill that void in a little bit of a way.
So do you guys plan on like growing the, the user portion of that into something
like that
[00:39:07] The Future of Open Sauce: Building a Better Social Graph
Brian: so I, I, I wrote a blog post on this called the return of social coding. So the answer, the quick answer is yes. Like we, we want to, we want to build a better social graph of like, these are the folks they interact with. This is my squad on open sauce on GitHub or whatever it is, but like sometimes your squad, like I, if you play games, like you might be playing League of Legends, but then you also might play Fortnite.
Uh, so yeah, like different squads and different arenas. Like, how do you, how do you showcase like the, the level of your interaction? Uh, but what also is like even more valuable is like, I, I might make a contribution and spend a lot of time with like some maintainers. I have a connection, but I don't have like a representation of like, Oh yeah, I know this core team or like I participate in these projects.
Uh, so like we're actually currently, we, we, we had a meeting this morning about like how we mapped this out for next year on like what the squad. Experience looks like, um, but we also have like the ability to highlight your contributions because I think that was always, it was, um, so I worked at GitHub for a while.
Uh, the home feed was like a kind of a sad story where the PM left right after it shipped. So there's like no new iterations. Uh, there's a few other features that that happened to as well. Um, we could talk, talk about badges if you want. Um, but what I'm getting at is like, GitHub has, it's still a collaboration platform, um, and it, it started as social coding and social, like seeing what, what's going on in the feed and your contribution graph.
Eventually, even have the get up jobs for a while was another 1, 000, 000 feature that, um, that got sunsetted. Um, there's a lot of opportunity to like connect that network. But I think what we've seen is like, I think the best community have done this so far has been Web three like they have. Yeah. Cause like this is so, it's so opaque on who's doing what, so then there's a lot of these Web3 tools that show what's happening on, on chain and off chain and stuff like that.
Uh, but they're also niche, but no one's doing like what's happening in open source and the open source is hoping to, um, to fill in that void.
Andrew: Yeah. Like people might go, Oh, well, that's not a thing I need. I've never gone to a profile and wanted to check that out. But literally just today, me researching this podcast, I was like, okay, what's so what's Brian been up to? And I went to your GitHub profile and I basically can't find anything because it's all grouped and like most not shown.
And so, yeah, definitely a need there.
Brian: Yeah, I go to my open source profile. You'll see my, my, my last few contributions and that tell a story. Uh, and, and that's a, that's a thing is like we, I was also part of the squad that launched the readme profile. Uh, sorry, I didn't launch a readme profile, but I did the, um, the code newbie conference where we, so it launched two weeks before that.
Then we had like a go to market. Of like this set up, you develop a resume and that was the engagement. That's kind of how people use to read me as well, but the read me, it's gotten to like a place where there's a bunch of widgets that you throw in there and like your a plus javascript or a B minus and rust and that doesn't really tell you anything.
And this tells you there's a letter associated to a language. Um, and what I'd love to see is like, if you're doing rust, like what rust did you do? And to find that on GitHub, it's not possible to find that open sauce is like what we're working on.
Andrew: Yeah, I do find it funny when I go to a LinkedIn profile and there's someone saying, Oh yeah, this guy's real good at JavaScript. And it's like, well, what bearing does that hold?
Brian: Yeah, I actually, my first job in San Francisco, we endorsed each other for random stuff on LinkedIn. So I've been endorsed by like 10 people for pancake making. Um, it's just hand waving that as, as, as it can be,
Justin: well, these are the, like, Signal to noise ratio on a lot of this stuff is, is hard to get right sometimes, then, you know, I feel like this is always the advantage of a startup over an organization like GitHub is like a massive organization is hard to get focus and it does sound like GitHub has some really novel ways of generating this focus because you're talking about, like, we have sort of like this vision for this thing that we want to do, which Had echoed to me a lot of like how a startup might organize itself, but, you know, they can't deliver the same like clear vision over a long time.
It's just like, this is the problem that we're going to solve, you know?
Brian: Did they do have? Well, they used to have a team. They call them special projects and some of the coolest features have come out of GitHub came out of a six week cycle. Between hypothesis to prototype within six weeks, and then they put it up against leadership to get investment and like a lot of projects failed.
So I mentioned like this, like, live streaming, like, GitHub had live streaming built in the platform, and I remember actually talking to loom about this looms now at, at, at, uh, What, uh, and I was going to say Adresin, Atlassian, um, the, um, but like these like little special projects actually birth things like Copilot.
Um, that's like the GitHub Next team, which is like not six weeks, but it's a little bit longer cycle. Um, but that's where a lot of these cool little prototypes like get like Copilot chat just came out of that. There's like search plus, uh, a open AI, basically, I think it's called Copilot chat is still what it's called.
Uh, but now it's in browser. Um, That's what we get like these skunk works projects, but unfortunately these are only few teams that get to do this stuff So if there's no buy in or investment, there's no like actual sort of Transition in the profitability for github because they like that they're in Microsoft now So like that is the goal is like profitability and moving the stock price based on like different features Like unfortunately, we won't see an update for badges for a while
Justin: Yeah.
Andrew: So, uh, the platform helps people define projects and data about those projects. Can it help me find like projects that are unmaintained in my stack?
Like, uh, you have all this data, like there's probably floundering projects somewhere, does it help me with that?
Brian: Yeah, yeah, this is something that we, we haven't shipped it yet. Like you can kind of see it when like, if you, if you track a bunch of it, so with open source, we ask you your interest. Uh, but then we also, I'll be honest with today, we don't have a good, we don't do a good job of getting you to the inside pages to then tell you what to do next.
But the idea there is like sync your org or sync your repos and then we'll tell you activity and velocity and stuff like that. Uh, but like, there's probably. So like, well, we haven't shipped this yet, but like we're working on this, uh, this like concept, which is bus factor, uh, and bus factor is basically if you get on a bus and you never come back, like how detrimental is that to the project?
So anything over 60 percent between two contributors, uh, is, is bus factor at that point. So like, there's not a distribution of work, um, between contributors, uh, is questionable. But then if the, the contribution velocity is also not going up into the right, then that's also questionable and, and bus factor.
So, uh, this is stuff that, uh, only I can see internally right now. We're actually trying to surface these in charts and graphs, uh, eventually. Uh, there's another thing that I was, we're actually playing around with, um, in partnership with another company is the, the, the cross between contribution and downloads.
It's like, if your downloads continue to go up, but your contributors and contributions go down, like there's a bit of a tech debt legacy project thing happening there. And this is something that enterprise suffer from consistently.
Justin: Something that I also, this is just a random thought, but something, I don't, I don't even know if this data is a visibility, but when like projects change hands, you know, it's like someone like is like, Hey, I can't maintain this anymore. Well, someone take it over and then they change hands. It's like, you know, we've seen that a few times where that's actually been at a point where like something malicious gets shipped in a project.
Hopefully that doesn't happen often. And you need a story for like, when a maintainer wants to step away from a project, having someone step in. But like. Those inflection points are often something that would be really nice to know about. It's like, Oh, you know, this changed. Uh, and I, it's, it's hard too, because yeah, I don't know,
Brian: yeah. No, it's hard to see that too, as well. If you, if you just hit download or NPM install, like you don't see upstream of like. Oh yeah, there's a new core contributor. Where'd they come from? What's their story? What are they contributing to? Um, this is something that's actually been top of mind, at least in the security space, like, um, supply chain security.
Where, like, we had a couple different, like, weird attacks where, like, colors. js is being one of them. Like, unstable maintainer, hijacking the community, and like, demanding based on funding needs, ransoming based on, to make contributions. Core. js being another one. Actually, I think January, February, had an announcement of like, hey, I don't get funded to be the Core.
js library that everyone depends on. I could do something malicious, like, Hey, I don't know, maybe I just push this button like there's literally the post that they made and then they had a whole article about this, but like security wise, core just is like 100 percent 95 percent like secure keeping the updates going.
I was talking to for us actually from socket security. Um, and he was talking about this thing called the pervasive incentive. So like, if things are chugging along, Um, there's no incentive for me to fund it. There's no incentive for me to even pay attention. But as soon as their vulnerability is like, Oh, let's all throw money at this thing.
Um, so like when the, when you talk about the changing of hands, like A lot of times maintainers are burnt out and they just don't have not just funding. It's like contribution. Like I'm the only one working on this thing, but now I've got like this random, uh, I don't want to call out specific like this. I hear, I talked a lot of maintainers.
So like large enterprises will basically dictate what maintainers should work on without even providing funding. And like, they'll just like engineers at different companies will just be like, Hey, this is blocking my release, or this is blocking something upstream, fix it. Um, and that's like true in GitHub issues.
And like the sentiment is just like. I can't believe this company is like, you're representing this company. You tell me to ship this for your company to make more millions of dollars. Uh, but I'm over here like living in a one bedroom apartment with my two dogs. Like, okay. Uh, and that becomes super challenging.
So if, if that's actually, so not on the roadmap, but I think I'll put it on there, like sort of signaling when there's a change of guard or when there's a new person in the core contributor, um, cohort, like that should be something that folks should be alerted of.
Andrew: Yeah, definitely. Uh, yeah, I've been, I've been in that same position too, where it's like, well, this is a library I kind of care about, but now this company is asking me to do security updates and I don't care. It's like, uh, hard, hard falling in the chain. Uh, but let's, let's start looking to the future a little bit more.
So we've talked about it in, uh, light terms, but like, what are your plans for the platform? What do you guys want to do next year? What are you building towards? That's
Brian: Yeah, yeah. So we we've been working on this tool. Um, uh, so our infrastructure is basically we have this tool called pizza oven where we bake repos and we turn out. Insights, basically. So we want to we've been working with folks not on GitHub as well to kind of figure out how do you make an agnostic platform so you can see the full scope of open source.
Um, so we're working on that. We're also working on this new new service to basically capture event data as well. I think there's a tool that folks everyone should look at. Um, you consider them a competitor. Maybe not. Yes. OS. It's insights. They're doing a really good job in time series database. So I think that's what we want to move into next is like Getting event data capture.
If you, if you want to look at 2015 from March to April, um, like be able to go back in time to see where are some inflection points, uh, that's something we'll be shipping before the end of the year. Uh, at least shipping on our beta, uh, branch. Um, but everything we're doing is open source, so I definitely pick a peek in there, uh, and then we've been, we've been working on some really cool things that we were hoping to ship by February.
around like the vector database space. Uh, so like being able to understand the, the actual social graph of different contributions. Um, we'll be calling this code name Star Search. Um, so Star Search being like where Justin Timberlake and Beyonce were discovered with Agment Man on TV. Think about that for developers.
Like, how do you discover up and coming developers, not just up and coming projects? And, um, that's something that we, we've sort of mapped out. I've, I mapped this out, and when we talk with the team internally, we've already started a process on this, and we're hoping to have, like, the first version, uh, Q1 of next year.
Andrew: interesting. So that sounds very AI powered if you're doing vector database thingies.
Brian: Yeah, it's like, it's definitely AI powered. It's definitely way more ML. It's like, it's funny how AI, like how they're interchangeable. Um, but yeah, definitely AI powered PG vector. Uh, big fans were using super base. So looking forward to really starting to dig into that stuff.
Andrew: so just like another broad, like future facing question. I mean, obviously your, your, your organization, your product is all oriented towards open source and thinking about open source. So. You know, if you look ahead. The next five years or whatever, how do you think open source is going to change?
[00:51:33] The Role of AI in Open Source and the Future of the Industry
Justin: What are, what are some of the trends that you think might continue or like, I don't know, what are you looking forward to? And then in the next bit.
Brian: Yeah, I mean, one of our directives and goals is really just increase the pie. So when you look at like 370 million repos on GitHub. Only 0. 01 percent have more than five contributors. So like our goal is to increase minimum contributors contribution to five. Um, so if we go from now, which is, uh, 900, 000 developers working in open source to 9 million, like that changes the game.
Like if more people participate, like there's a lot more stuff to watch. Uh, it's like the same thing with the NFL and the NBA, like more money in the NFL, more money in the NBA. Now gives us more excitement to watch on the weekends. Um, and my hope is that folks participate in a way that actually is inclusive in their work as well.
Um, so like a lot of times you think nights and weekends and open source, like my entire career has been like, I do open source at work. Um, so like I've been, I've been lucky enough to be in the Bay Area and have work at companies that I'm allowed to do that. But the idea there is like if more companies are adopting open source for security reasons, like all fintechs.
They all participate in open source for regulations reasons, and like, this is a showcase that they're above board. They're using like top of the line software and their ops and DevOps. So, if that's the trend, then there's probably more adoption in open source. And now we're looking at this sort of like AI question.
Um, I don't know if we actually shut down more stuff because of AI. I think we license things differently in the AI future. But I think more people will participate, uh, because of the advent of AI. So like, if I could just like, Tell me about this project and how I would implement this with my, I don't know, random TypeScript library.
Like now, we're making things a little more approachable. Uh, it will have more spam and more of that stuff, but, uh, we didn't talk about spam, but, uh, we hope to, uh, also ship some spam tooling, uh, for open source as well.
Andrew: Nice. Um, so, uh, this is a question we like to ask the over the past few episodes, but what is your spiciest dev take?
Brian: Uh, DevRel is probably not needed.
Andrew: Big words from, uh, the big dev rel man himself.
Brian: Yeah, no, I, I think, and I, I had a tweet not too long ago, which was like, You probably don't need DevRel, you probably just need an engineer that writes a blog post. And the response, actually, Prime responded, it was like, No, I'm not writing blog posts, like, that's, you need to pay me for that.
And that's like, not exactly what I was saying. I'm just saying, if you want awareness around something you're building, just get the engineers to talk about what you're building. Uh, and that's like, better DevRel than like, standing at a conference and handing out socks. And, um, and I think we, we, we've seen an influx in dev rel of people hiring first 10 employees like someone's dev rel.
And I think at the end of the day, it's like if an engineer can talk about what they're building and tell that story, it's going to be way more valuable than like figuring out how to like win SEO and pay for Twitter ads.
Andrew: Yeah. It's a, it's a weird advent of the like celebrity dev rel. There seems to be a lot of those lately.
Brian: Yeah, it's not sustainable. Like, we saw, like, so when I joined DevRel forcibly, so Netlify was like, hey, you're doing a great job. Do DevRel. I was like, no, I don't think so. So then they called it DevX, and I was like, okay, I'll do that. Um, so I did that for about a year before going to GitHub. But at that time, I remember the team at Microsoft had the biggest heavy hitting DevRel team on the Azure Dev Advocates, and it lasted for 12 months because, like, when you get a bunch of LeBron James on one team, like, it's hard to even know what you're doing, and it's hard to even know, like, how do you, how do you manage that?
So are they self managed? And then now they're all managers at other places, which is great, but like, it's just not sustainable. Like you won't, you can't have the super squad of a celebrity DevRel's on a team. You need an engineer who cares about your product enough to show it. Um, and it, there's not a, there's no DevRel that's going to learn your product well enough than someone who actually is building it.
Um, so my, my rep, Recommendations like find someone who's excited about your product early days, uh, and then get them to like, help build it with you, uh, and help talk about it. And sometimes you could justify hiring a junior engineer that way because then they could just like, get out of the way and like, talk about it, but then learn at the same time and share a story and write internet engineering blogs.
But, uh, I think we'd be better off if we, if we just focused on like more developer advocacy and less on like celebrity.
Justin: Yeah. I think when I worked at artsy, artsy had a big internal writing culture and, you know, the arts, the artsy engineering had its own brand outside of the artsy company just because of how much writing they did. And, you know, it is interesting because there, there are like things that are extremely valuable to you as a person.
Uh, so one of the things that Andrew and I have talked to a lot of folks on the podcast about is like how valuable open source is as like. Just to you as an individual, it's like you write open source, that's yours, you leave that company that, you know, was paying you to write some open source or whatever, that, that's, that's still you, you know, that's you, you can point to that work and say, look, here's what I've done, and it's the same thing for all this public writing, uh, you know, making a case, uh, Uh, you know, really working to describe features and functionality, or be it culture, coding practices, how you build products, you know, whatever it may be.
This is all extremely valuable for your public profile as a resume item too. So it's like, it's worth the effort from just from that perspective, I think.
Brian: Yeah, it is too as well. Like, you don't have to like, you don't have to do content to be a successful engineer. Like you don't need a fancy camera and like a twitch stream or YouTube channel, but it does help. Like if you could explain concepts in a way that it's convincing, like the only words that I do DevRel.
Uh, it's because I had to convince other engineers because I was a career switcher of like what I did was actually knowledgeable and like valued. So like, I'd go get a bunch of people to like, like and subscribe my Twitter and like my blog. Uh, and then I'd validate that by putting it in my PR and be like, Oh yeah, so and so responded to my PR and I would get like some heavy hitters in the industry and open source maintainers.
They're like, Oh yeah, this is the way to do it. Uh, and then I validated that within doing it again and like learning and then doing a conference talk and then. Like it I'm better off because of it and like I was able to accelerate my career because I was able to like stand in front of an engineering sink and say hey This is why I made this design choice in this decision And the challenge like I've worked with a lot of junior engineers like a lot of times you don't feel adequate enough to speak up And then one of the best things you do is just like start a blog.
It's like what did you learn yesterday? You don't have to share it with anybody. Just like write something down and then use that as like the fuel to like, Oh yeah, I've seen this before. Let me go back to the stuff I've done before. And that usually becomes the best content. And the best way to like showcase your product, uh, is like having a junior use it and give you feedback.
Andrew: Good words to live by.
[00:58:13] Tooltips
Andrew: Okay, with that, let's, uh, move on to tooltips. My first, uh, tooltip is a library called Melt. Uh, all Melt is, is Radix, but for Svelte. So, it has a bunch of headless UI components that you can use to create your own, uh, Accessible UI.
Uh, I don't use spelt myself. I've never used this library, but any, I love the trend of just creating these headless component libraries throughout all the frameworks so that you can just like basically have the same API jumping between everything. Uh, it's just a, it's a boon for accessibility in our industry and devs, uh, making good UI from the start
Justin: Yeah, it's pretty cool. I've played around with it before. It's awesome.
Andrew: and it'll look very high
quality.
Justin: Yeah, definitely.
Andrew: Next up, uh, auto automa.
Justin: Yeah, this was one that was, I think it was trending on Hacker News today, so a cheap rip for me. But, uh, I love node based editing tools just because I think they're neat. So, uh, this, uh, Atama is a node based editor. Based so visual node based, uh, browser automation tool. Uh, so it's got a, uh, browser extension that you can install and then you can build automation workflows and just these like little nodes that you snap together.
Um, so, you know, again, sort of bringing this relatively tactile interface for folks to do what is, could be a rather sort of advanced operation. Um, and I think this is awesome. Um, I've seen some other really cool tools that are just like record your workflows. You can just like do the thing and record it and replay it.
And, you know, I think that's extremely valid too, but again, love node based UIs and I thought this like was well presented and looks like a cool tool.
Andrew: Yeah, pretty cool. I like that they have a marketplace and I can just like install other people's things. Uh, I, I'm excited by the GPT announcements cause it feels like that's where it's moving. Like I was admittedly building my own like thing called bot garden, which would be like a GitHub for your GitHub for your prompts.
And then you can like fork and remix them. But it seems like that's where open AI is going themselves. So I stopped coding that.
Brian: You gotta power through until they shut you down.
Andrew: I'm good. I, I value my free time a little bit.
Justin: they could buy you, never know.
Andrew: Next up, we have post hog.
Cool.
Brian: PostHog is something that I love to use every day. It's got a funny name. I only was aware of it because I think they went to YC and like, they were one of the few folks who were open source in their batch. So I had them on a podcast and ended up using, like right before I started open source, the newest version.
Just dropped it into the, um, uh, into the header, uh, and then never thought about it for a bit until we finished the product. Uh, and then the cool thing about post hoc, it's like kind of like Google analytics. If you ever use that, that platform, which is, uh, I don't know if they're like GDPR compliant still, but like post hoc is, um, and it just gives you a bunch of insights in like, who's using your product.
It actually, um, can record events. You can do surveys. Uh, you can like, it's like if you're not a product manager. Because like PMs would probably like swim in this stuff. Um, I'm not a pm so it's like kind of like a, uh, an analytics tool for developers. Uh, so you just like put it into like your next app or your React app or spelt or whatever.
Um, and then you start capturing events and then because you just capture a bunch of events, um, you can then filter it down. So the, the cool thing about, it's like all anonymous data, so it's all GDPR compliant there. Uh, and we were to find out things like weekly active users. So like we were like, no one uses using open sauce in the beginning to now we're getting over 2000, uh, who are weekly active.
And that was because I added it 18 months ago to the app and never touched it since. And now I have all this data.
Justin: That's awesome. Speaking of companies with like good writing culture, PostHog like puts out some excellent blog posts. They're, they're great.
Brian: Yeah. I'll mention that we do have a case study from open sauce and their, their post hog company case studies as well. So definitely check it out if you're interested in finding out about product market fit.
Andrew: so my next tool tip is something from fleet AI. I think it's a new AI company making cool things. Uh, what they made is, uh, a, they embedded all the docs of all the top Python repos into a fine tune GPT that you can use at your command line. So you can use this to like. Bootstrap new AI applications that you're making, because everything in AI right now is Python.
So, uh, if you're looking for something to like, really help you start out, I think a tool like this could really get you going. Because of all things, I think the best area to use this for is in Python right now. Because like, that's, it's just like, why are they the best tutorials for web development? Well, it is the web.
Why is the best things for AI in Python? Well, because Python is AI. So if you're looking for something like this, definitely check out fleets context, except we have Svelte flow, another flow library
Justin: Yeah, you know, keeping on the theme of both libraries that are like really solid libraries that are ported to different ecosystems and, you know, visual, uh, node based interfaces. Uh, so there is a library that I love called React Flow, which is a node. That's a. React library for building node based editors like, uh, Ultima.
Uh, this is just the small version by the same, uh, organization. Uh, high, high quality. Uh, this, this version is still an alpha. I should caveat, but, um, I love the interface. It looks cool. I'm excited to play around with it.
Andrew: Yeah. Great, great demo on the page. I love when I can move things
Justin: Yeah. Yeah. It's excellent.
Brian: Oh, I didn't realize you were moving that around.
That's amazing.
Andrew: Yeah. It's all live.
Brian: Yeah, that's really cool. I love these little tools. Now it's like I try to shoehorn stuff like this into the product I'm working on. It's like for no reason I'm just going to have some
sort of whizzy wig.
Justin: You gotta have fun.
You, gotta have fun.
Andrew: you can tell the whole, uh, developer journey through a flow based node
editor
Brian: Yeah. We're kind of working on something, but we use something different for our visualizations. Um, but yeah, we're going to be networking this social graph with the thing we're shipping in February.
Andrew: and then last up, I did not click on it. Another GPT.
like thing,
rush GPT.
Brian: ';;;Yeah, it's it's almost just like what you just mentioned as well. Um, so, so this is actually a GPT plug in, uh, chat GPT plug in. Uh, so opening, I actually shipped, uh, The, well, they shipped a bunch of different things, but you can now GPT everything. Uh, so just like taking the entire Python library, uh, one of the injures on my team took the rust library or the rust, uh, docs and the rust book, and now has a GPT to ask, how do you write this in rust?
Uh, so he just honestly just tweeted out this morning, which is pretty cool because he's currently the only one on the team writing rust actively. Um, so we're always reviewing his code and asking questions, and now we have a little cool tool called Ferris to answer that, which is basically if you're listening to this chat, GPT, um, but just rest questions.
Andrew: Yeah, I assume this like greatly increases how accurate the output is
Brian: Yes, yeah, and it's, uh, I think there was a joke that, uh, opening eyes killed 1000 startups because of what they shipped and basically what they ship is like, okay, now you get the two natively inside of their interface, which is. Pretty cool and unreal.
Justin: Yeah, that's
pretty awesome.
Andrew: rough time to build right
now,
Brian: Yes. Yeah. You might want to slow down a little bit and like look at the, uh, the ecosystem. But if you're building dev tools, though, always build dev tools. It's you're building the thing underneath.
Justin: We're pretty bullish on it.
Andrew: people need those shovels or better yet wrenches. Um, so that's it for tool tips this week. Thanks for, thanks for coming on Brian. This is a interesting delve into topics that I hadn't really thought about before and I know GitHub hasn't really thought about all that much. So thanks for building.
the platform
you're building.
Brian: My pleasure. Stay saucy.
Justin: Yeah, Brian, it was great to have you on. Thanks for coming on.