Rails Business
Brendan Buckingham and Ryan Frisch talk about developing with Ruby on Rails and how to leverage it to build a business.
Rails Business
Chris Oliver
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
In this episode, the hosts welcome Chris Oliver, a prominent figure in the Rails community, known for his work on numerous projects including GoRails, Hatchbox, and Jumpstart Pro to name a few. Chris shares his programming origin story, his academic journey and numerous early projects, explaining how he transitioned from Python to Rails. He recounts his experiences starting GoRails, managing multiple projects, and the challenges faced with Hatchbox. Chris elaborates on the impact of AI on the programming landscape and its potential risks and benefits. He offers advice to aspiring developers and reflects on his journey of self-driven learning and continuous improvement.
LINKS
- GoRails
Send us feedback or questions to railsbusinesspod@gmail.com
All right, today we got Chris Oliver with us, a man who kind of needs no introduction in the Rails community, but if, for those who don't know, he does go Rails. He does Hatch box. I mean, too many for me to even remember, honestly. But, uh, really excited to have you here, Chris. So, uh,
Ryan:Let's say J Jumpstart, pay remote, Ruby, uh,
Brendan:you, Ryan.
Ryan:you know, hatch box. I think, I don't know if you said that and Access tenant noticed. I mean, the list just keeps going on and on and on. But, um, thank you for joining us.
Chris:Yeah. Thanks for having me. I'm excited to be here.
Brendan:Yeah, and as we generally do with our guests, we like to start with, you know, how did you get into programming and especially Rails, you know, what, what was your origin?
Chris:Yeah. I got into programming. Um, dad was, my dad worked for a bank as like a farm manager and like. I don't really know. I need to ask him more about this, but he was like really good with computers back in like the DOS days. Like once it got into Windows 3.1 and stuff, he started to, you know, fade off an interest or whatever. But that's like, I remember as a kid, like we had this Packard Bell desktop that he's like taking apart, putting in a, um, it was like a 14 K modem or something. Like, not even 56 K or whatever. But, he was like, you can't touch any of this. And I'm like, what is this? And I was so excited. And, he was um, going through a bookshelf and he was like, you might wanna read this. And it was his old Atari. Like 800 or 400, Atari basic book and was like, yeah, you might wanna read this. We don't have a Atari. Like my grandma, uh, recycled it or gave it away or something, you know. And, uh, I figured out though that our other older computer that we had was this Magnavox desktop computer that had GW basic on it. You type your own line numbers and everything. So I taught myself programming on that. when I went through college was getting a computer science degree, one of the professors, said Hey, and pulled me aside and was like, Hey, you know, one of the physics teachers has a grant for doing this web development stuff that you might be interested in. She needs more students to work on that. So I had been teaching myself Python and, uh, Django and Flask. And then that grant, uh, what they were using it for was the Zoo Universe. Um, which are these projects that are like, there was a Galaxy Zoo is one that I worked on where you had like photos from the Hubble telescope of Galaxies that there were so many of them, they were crowdsourcing the categorization of it. So you'd be like, look at this photo. How many tails does this galaxy have? One, two, whatever. And, uh, we would go through and, and build out, All, all kinds of things for that. There were several other projects there under that too, but that was, that was all in rails. So I got started on that, like Linux on my computer, couldn't figure out how to get rails to run,
Brendan:Hmm.
Chris:Ruby. And you know, that's still hard today, unfortunately. And, uh. So yeah, that kind of was my first real job as a programmer. I feel like I'd done some little consulting stuff before then, but like that was my first actual stuff that shipped to production and stuff. Back when I was like junior in college or something. That's what got me into Rails and was like, yeah, this is really, I hate it.'cause Rails was so much like, there's, there's magic happening here. Where coming from Python was like, everything was super explicit and
Brendan:Mm-hmm.
Chris:I was like, this sucks. And then I read that meta programming Ruby book that explained, here's the philosophy. Like you don't need to mess with all this, like these a hundred imports at the top of your file.'cause like we know how to find that. Don't worry about that. Don't put any thought into like, how does this class get required? We'll take care of it. You focus on your code and it's like, oh, this is cool.
Ryan:right, so you're the first person now, uh, in, you know, very limited, uh, episodes for us that didn't go from PHP to Rails,
Brendan:Yep.
Ryan:So that like, I think that's like a, okay, so may, maybe it wasn't a big part of your story, which you're the first one that, uh, has been on that trajectory.
Chris:I mean, yeah, my, my trajectory was like GW basic. Then I found QBasic on our computer as well. QBasic was great 'cause I could write my own functions and not line numbers. And then I like got c plus plus for dummies and then that was too hard to understand. Then I got like my TI 84 plus calculator for high school, you know, and I was like, oh, this has ti basic on it and I can cheat on my homework and stuff. Um, so really that's probably the first time I ever made money for my code. Actually. Uh, I was telling somebody this at RailsConf, but like, I. One of our exams in high school was coming up, so I had spent, uh, three entire days going through all of the homework we had ever been given that year. And I wrote a program on my calculator that, did basically fill in the blank for every problem set. So, you know, they just changed the numbers, but the format of the problem's exactly the same.
Brendan:Mm-hmm.
Chris:show your work to get full credit. So I
Brendan:Yep.
Chris:fill in the numbers and you hit enter and it shows you the first step and then the next one and the next one. And, so I wrote this and had like every type of problem I was even like. Crafty with it. So it starts out, you open, you launch the program, and it's like a guess a number game. So if you put in a special number, it'll take you into the secret menu with all the homework, cheating stuff. and I sold that program to some of my friends and I made like 11 bucks on that or whatever. I, I also built my first DRM for it too. So
Brendan:Oh man.
Chris:easily type like the variable theta. In the calculator. it was there, but I don't know that there was like a button for it or something easy to access
Ryan:a key combination, right? Or something crazy.
Chris:knew how to do it, so I figured out how to do that. So you had to set theta to like 42 or whatever, and if you didn't, then when you, the program, uh, it would print out like this has been pirated or whatever. And some of my friends did that. And so what after prints that out, I also found out that the operating system has a sort function. So I did the largest array I possibly could with the largest numbers for each one of those. So it would just sit and sort for like 30 seconds or something and you couldn't interrupt it. most people thought, and I did that in an infinite loop, so it'd print out like, this has been pirated and your calculator would just start crunching numbers, trying to sort this giant
Ryan:That's amazing.
Chris:And it
Brendan:brilliant.
Chris:doing that. And you could kind of cancel if you got lucky in between, uh, when it loops and goes to the, the next sort. But most people just had to pull the batteries out of their calculator, which means it erases all the
Brendan:Everything.
Chris:Um, so that was my first ERM for that. Uh, back in the day. It was fun.
Ryan:That's amazing. Yeah. So the, the second thing I wanted to to point out is, uh, you. Got introduced to Rails and you, your experience was you hated it because it hit a lot of stuff, which I think is, uh, different than, uh, a lot of people, uh, that we've spoken with that is like, they had trouble with all the detail when they found Rails. It was an unlock. That's how I felt. Um, so,
Chris:Mm-hmm.
Ryan:you know, I think that's, that's interesting that you were the other way. Of course, you were also a Compsci major, so you know, that helps.
Chris:Yeah. Yeah, it was, I don't know, like I had built enough stuff in, in Python, like so. I had done this really big open source project in Python, at least it was big for me at the time. Like I was learning Linux for fun and just installed that on my computer.'cause Windows, windows Millennium Edition wasn't very great. Uh, and so I, I had been learning Linux and on their, the Ubuntu forums, like there was some, there were some college kids that were building tools to make installing like the Nvidia driver easier and configuring that.'cause I, I literally like downloaded, uh, Ubuntu, formatted my laptop, tried to install the Nvidia driver and screwed it up. I'm stuck in the terminal. I don't know any of these commands to run. So I would reformat my laptop every single day for like a week until
Brendan:Oh geez.
Chris:you know, working. And that kind of inspired me. It was like, um. I am on dial up at my parents' house, so I wanna do some open source stuff to give back to the community. uh, high school has high speed internet flash drives are just becoming a thing. At the time I had like a 16 or 32 megabyte one, you know, not much, uh, at the time. And, you know, it was like I could probably write some code that could run on windows at school, download software for me to take home and install on, Ubuntu offline. So that became my first like, real big project that I ever did. I, um, would run like AP, install Firefox and do it in verbose mode and write down every single file it downloaded. And I would go manually download those myself and open 'em up and read 'em and like, oh, this is, all of the packages in their dependencies so I can write some Python code to like. Parse that and create, you know, a big hash of all the available packages and versions and their dependencies. And, that became this project called Keryx, that like I open sourced, uh, got interviewed for Linux Journal magazine. I have like at my parents' house, like physical copy of that.'cause when it came out we ran to Barnes and Noble and I was like, oh my God. Like I'm in this magazine.
Ryan:amazing.
Chris:Um, and then that, that like was open sourced and people started to use it.'cause I posted on the forums and like there were a couple other pretty clunky tools. I was trying to build like a UI and everything, and the reason like I ended up on Python was like Ruby and Python were recommended the most, but like Python came pre-installed and it was easier for me to run on Windows. Still, still
Ryan:Yeah.
Brendan:Right.
Chris:you know, that was kind of what was like, these languages look very familiar or similar, but like, they're very, very different. Um, syntactically, they're pretty close, but like functionally they're very different. So I ended up in Python land for a long time, but yeah, that, that became another thing that was wild.'cause while I was in. college, like I was finishing that, I think I rewrote it like 14 times, like from scratch.'cause each time I'd learn a bunch of stuff. Like I knew nothing of object oriented programming,
Brendan:Hmm.
Chris:I was teaching myself what the hell a class was. Uh, I, I like start doing the gooey for it. And didn't understand, like I'm, I click the button to go download and like all the dependencies for Firefox. I click that and the whole application freezes.'cause I'm doing the logic in the main thread.
Brendan:Oh yeah.
Chris:I've never used a thread before. I don't know how this works, but it freezes. Then you're like googling, like, why is this freezing? And then, you know, there's a stack overflow post. It's like you need to spin up a new thread, do the work over there and talk to each other to go update the UI for progress and whatever. so that was why it took me like 14 rewrites to, to get to a decent spot. But like seriously got emails from the two. I remember the best, somebody in South Africa was setting up a computer lab with Ubuntu and they didn't have reliable internet or even reliable power, and so they were like downloading stuff. using that to set that up. Um, and then somebody in the US government emailed me and was like, Hey, just wanted to let you know, like I'm working on this air gapped, super secure server, uh, and so I'm using your tool to go download packages to install on there. And I was like, I'm just some, you know, random college kid. And like, this is crazy. Like, I had no idea anything. I figured nobody would ever use it. You know, and
Brendan:Yeah.
Chris:you get to that point. And that kind of led me into like, I got to from now on the bar's kind of high. Like I want to make sure that what I'm working on people use.
Brendan:Mm-hmm.
Chris:that became the, the thing that was most fulfilling about writing software is like making sure people use what I'm building. So.
Ryan:Well, you certainly, uh, succeeded on that front, with all the stuff that you, uh, you know, provide to the Rails community. can you tell us a little bit about, you know, how you think of your sort of product suite, uh, for lack of a better term, from a business perspective? Um, you know, is is there a, a strategy there, uh, or is it just that you know, you, you see opportunities and you go after 'em?
Chris:Um, yeah, probably a bit of both. I mean, it started out simply as after I graduated just working on, this little agency and, and like we did stuff for design agencies in town, building these apps that would get thrown away in a month or something, and like, I really would rather go on my own consulting. So I started doing that, but it was the same problem of like, I do work and kind of gets wasted or not really use too much. And I never found anything that worked and. It was just kind of like, I miss, you know, rails Cast hadn't been around for a year and a half. I miss those videos. They were all so helpful for me. and I was like, well, you know, if Ryan's not coming back, maybe I could teach other people who are just newer than me. I'll never be as good as Ryan. Um, you know, and I was like, I, the only thing I can do is try to teach people who are more beginner than me. Like I could teach myself something that I learned, uh, in the past week. You know, I wish that was more content out there. So that became GoRails and like nobody knew who I was. So the, when I launched GoRails after like six months, I like January, 2014, quit consulting cold Turkey. then. I hated listening to my voice. And I'm also, you know, up until then I've been programming in my head the entire time. So I couldn't even articulate anything I was thinking while I was coding. So I forced myself in like March 'cause I made no progress. Then in March I was like, I have to force myself to record 15 minutes a day and just delete it afterwards. I just have to record something and if I do enough of that, eventually, hopefully I'll become, you
Brendan:Hmm.
Chris:somewhat accustomed to it and stuff. So that became. Enough to put together like a course and like a free version of a course too, to hopefully, like, here's a trial, here's the paid course. But if you're buying a course, you have to already have trust with that person. Um, so that's when I like switched to, um, to doing GoRails is like the weekly screencast. And I was like, well, if Ryan was doing that, then maybe that was a good reason. Doing a free video every week, every other week builds trust and maybe that will help with you know, my business.'cause at the time I was like, you know, a developer and going through cs, everybody there was like, yeah, salespeople and marketers are the worst and whatever, you know? And so yeah, that was, that was fun. But then to know and talk to so many people through GoRails. Ezekiel Roses, uh, had DMed me one day and he was like, have you seen Laravel Forge? and I was like, no, I've never even looked at Laravel before. And he was like, look at this. It's basically your deployment guide, but automated. And he was like, you should do that for Rails. And I was like, I should do that for Rails.
Brendan:Hmm.
Chris:became the first version of Hatch Box. Um, ' that guide got linked on Stack Overflow and stuff and got a ton of traffic for basically nothing. Like I didn't,
Brendan:Nice.
Chris:just posted the, my own private notes and was like, I'll just share them. And so that still got a ton of traffic and was the reason why I was like, maybe I could convince people to go to homepage and like. Maybe I could sell them Screencasts, but that was the main place people were going. That became Hatch Box. Hatch Box was terrible. I rewrote like, you know, the payments differently. I never even built teams for GoRails or for Hatch Box in the beginning. So that was a nightmare to like bolt on later on. Which eventually was like, I'm gonna do this and have several products, I need to then build Jumpstart where they're all standardized and I can focus on the actual product instead of payments and accounts and teams and APIs and notifications and all that. So literally all the stuff is just evolving out of like seeing my struggles and assuming that all other people probably also struggle with this too. that's also like where several of our courses came from. It's like we have this advanced Ruby course, which is pretty much going through like all the stuff I wish I knew 10 years ago and put together that into like, just a super in-depth course. And same with like the Hotwire course we just launched this year. Same thing. It's like now that I've done Hotwire for long enough. Let's go compile all this knowledge into one course. Um, we've seen it in production for many years now, so we've seen some of the warts and issues literally yesterday, somebody reported a bug in Jumpstart and was like, yeah, if you override the layout, you can accidentally lose the turbo frame, um, layout. And so I was like, well, let's fix that, but let's also make that a lesson for the Learn Hotwire course. And it might end up being on GoRails too, or whatever, but like, because all of the products are so, you know, intertwined, I kind of do work once and get paid for it three times,
Brendan:That's awesome.
Chris:you know, what has kind of worked out. Wildly. It's also the hardest part of the business.'cause now I'm context switching between stuff times a day. Um, so on one hand it makes it possible to get to the level of success. Uh, but also it's the thing that probably holds us back from going further.'cause it's like, how are you gonna maintain all this crap now?
Brendan:Mm-hmm.
Ryan:that that was gonna be my next question. From the other side of the room, it feels like you have your hand in a ton of stuff. You know, personally, I have a hard time keeping up with the one code base that I have. Um, you know, for, for, for others that are in a similar situation, like, do you have any advice on how to, do multiple things well, 'cause I mean, you do, you do do them well.
Chris:It's, it's hard. It's so hard. I mean, standardizing everything on Jumpstart
Ryan:Yep.
Chris:one of the biggest things that was a nightmare to do, but like, has made it a lot more sustainable.'cause that bug that we fixed yesterday and Jumpstart merged into Hatch Box, merged into GoRails, merged into the learn Hotwire.com code base, like did work once, fixed it in four apps, you know, and that helps. But at the beginning when you're starting your own business, you're like, I just need to make a hundred bucks now. A thousand bucks now 10,000 bucks. you don't really put much thought into what you're doing. You're just like, I just need to hit a hundred bucks. Well, can I do anything? Period. people will pay me a hundred bucks for. over time you like learn the pros and cons of each product. Like GoRails has is an education, so it's naturally high churn. it's not very sticky, like need help with one problem. So people sign up, then they cancel. Uh, 'cause you solve their problem. They don't stick around necessarily. But, um, you go to the other end of the spectrum and like, hatch box is the opposite. It's very sticky, but it's also urgent. So if anything goes wrong, 11.30 PM on a Sunday night, you gotta get outta bed and go answer that ticket. because this is critical to somebody who's, actively working on their app. they probably live in a very different time zone like Australia or something. And so to them it's like not a big deal, but. It's also painful because that is Um, and like commoditized to hell because the VCs have invested in digital ocean and render and Heroku and yada yada. And so the, the prices are going down in nothing. And here I am, like we gotta have similar prices. Unfortunately, we can't be that expense that much more expensive, especially when we don't offer as much anyways. Like it's not all, uh, you know, Kubernetes and fully managed everything. It's like we're giving a. unique product that none of they, they also, none of them choose to serve this market and they don't want to do that. So we're in a good spot there.'cause like, we're unique, but it's also like one of the most painful things I've ever done. But you're just naive when you start and you're like, well, I don't know. I'm trying to find some way to stop the churn on the education business. This seems like a good opposite of that. And then you realize like, oh, it's got its own problems too, you know? So it's, it's really tough to manage all of them. Luckily they're all around rails, you know, and Ruby, so they conceptually are all very similar. But you know, I'm constantly watching my email for a ticket to come through and from whatever I'm doing to go do that. Then I've committed to. At grinding out a screencast every week for the rest of my life with GoRails. So I've got to make videos. Sometimes I'll do like five in a week. This is like probably gonna be one of those weeks, but then it'll be a month later and I'm like, we're running out of videos in the queue. I haven't recorded anything for a month, so it's really tough to like get back into the swing of things. So there's lots of little optimizations you find along the way. Like I found an interesting thing in some code I was working on yesterday, I will try to stop and record a video on that the next day or the same day while it's fresh in my mind, while it's, easy to record. And kind of optimizes things a bit, but it's, it's not easy. It's exhausting.
Ryan:Yeah, I, I can imagine.
Brendan:You have some help right on. GoRails. I know you got I think a couple employees.
Chris:It we're back to just Colin and I and
Brendan:Okay.
Chris:pretty much the same things as me. A bit of everything. Um, yeah, we had Kent on the team as well until recently. And like the, the support tickets we've gotten are a lot less than we used to. Like when Hatch Box was newer. I mean, the sheer amount of nuances to Linux stuff for Hatch Box is like ridiculous. Like literally, uh, this one config file has to have an empty new line at the end of it. It's like, well, when I edited it in Vim, everything worked, but when I automated it with Ruby, uh, it didn't work. And that's 'cause like Vim automatically put a new
Brendan:Puts there. Yep.
Chris:for me, but like my upload content over SSH thing did not do that. And
Brendan:Geez.
Chris:forever to track down. And
Brendan:I was just gonna say debugging. That would be a nightmare, I'd think.
Chris:and you gotta write a comment there 'cause you're like, this new line is really important. Do not
Brendan:Mm-hmm.
Chris:because how are you gonna, like, how are you gonna bake that into your code? Uh, otherwise it's like, that's some weird knowledge you need to know. You need to have it somewhere,
Brendan:Mm-hmm.
Chris:uh, you can't lose that otherwise things go down. other big learnings were like, you know, uh, relying on other people's code. with the hatch boxes is like just part of everything. Like it's the Ubuntu server code base, it's all these repositories. And you know, originally we used passenger and then Ubuntu came out with a new LTS version, people upgraded immediately. Passenger did not. And so then we were stuck waiting like a month trying to tell people, Hey, uh, I'm sorry that you upgraded your server. You can't do that. Um, we haven't said that it's ready and compatible yet. And so like they have to go spin up another server on the old version of Ubuntu and like redeploy their app. But it's like just bad for us, reputation wise.'cause
Brendan:Yep.
Chris:know, we technically should be in control of all that, but like, we're not, I don't, I'm not able to, you know, build a distro for
Brendan:Mm-hmm.
Chris:distro for Linux, just for Hatch box and whatnot. So it's been a tricky, tricky thing to do all this time. But like, originally it was such a, a struggle. I had considered, like selling Hatch box, uh, just shutting it down. I, I didn't know what to do with it.'cause it was like, yeah, you're making 2000 a month and you're spending about 80 hours a week or whatever on the one thing and you still gotta find time to put into making screencasts for GoRails. And it was like, if this is so low of ROI, is it really worth it? And so we had so much time with people asking questions. There were configuration things that weren't possible. Um, so over time I had to like re-architect everything four or five times. And now it's like really good. Like a lot of the stuff is all easily doable. We have minimal dependencies. Things are like way, way simpler and more stable and everything now. That it means a lot less support tickets and stuff, but for the longest time it was like, is not worth it at all for the money that I'm putting in or I'm getting out of it. But it was only, it only felt that way because like Go Rails is making decent money. And I was like, comparatively, this is a
Brendan:Yep.
Chris:thing. But at the beginning when I didn't know any better, I would've been like, I can't believe I'm making $2,000 a month
Brendan:Right.
Chris:I built.
Ryan:Yeah.
Chris:So it's funny to look back 'cause I, I hated it and was just like stressed out. All my gray hairs come from Hatch Box. I would, I would probably say, and. Yeah, looking back now, I'm like, I'm glad I stuck it out.'cause nobody else cares about that like, niche. There were people that, you know, needed to run some genomic software or whatever, or they got a server at home that they want to run, but it's a pain in the butt to keep up to date and stuff. Um, so we can go help them, whereas everybody else is like, nah, you use our cloud, use our servers. You like, use our configuration and play within our boundaries. And we're like, just if we can access your server, we'll help set it up. And, you know, that has caused a lot of problems and somebody goes and installs weird stuff and
Brendan:Sure.
Chris:or whatever, and I'm like, you know, you gotta be aware of this. That's limitations here. But it's, it's, um, it's not been that big of a problem. I, I don't think. And those people are already sort of familiar with Linux anyways. so that's been cool.
Ryan:So as, as you're talking about Hatch Box, I am getting the impression that, uh, I don't fully understand. I thought I did, but I, I don't think I fully understand what it does. Is it more of a deployment orchestration?
Chris:yeah, so like, you know, Kamal will and set up, um, the reverse proxy and it will go and set up your Docker containers and stuff, where you have that and you have sort of the ability to, I think now Kamal two can do like multiple apps in the same cluster. have all of that as well. So it's not using Docker, but, um, we'll I'm sure have, uh, better support for Docker right now. But we're, we'll go into your servers, configure them, set up caddy, for SSL. Set up your databases, uh, your CR jobs, your processes. So we'll set up, you know, this app will be assigned Port 9,000. So then caddy requests on that domain will be proxied over to your Puma service or whatever. Frankly, it doesn't really matter. Um, I wrote the first version with Passenger and it was pretty tied to, uh, either, Ruby or Python or Node. And then now it's just like, well, it's a system D process. Anything you wanna run in there, you can grab an app off the shelf and deploy it. Um, so, you know, a lot of the open source apps have environment variables that you can configure everything with. So now, we can kind of deploy everything for you. So, you know, one of the differences still is like, Kamal requires you to go maintain those servers and stuff. You've gotta do that. It doesn't handle that. We will go in and update things, um, and like configure security updates to be automatically installed. We've got an easy thing to go and update your server for all the bug fix packages. Um, you can ss you can have it SSH in and run scripts for you, like if you need to compile lib VIPs with some additional, you know, um, format dependencies or something. So it's, you know, similar in, in that sense. But we're trying to be that, like with Kamal, you have nobody but AI to go ask for help with. And it's like if something goes wrong with Hatch Box, you send us a message, we'll take a look and check it out and help you out. Um, so it's, you know, worth it to a lot of people just to have a little bit of support and stuff, but. Yeah, we kind of SSH in configure everything. Um, we pre compile your Ruby version, so when you're, when you're on a $5 server or raspberry pie or something, it's like, try and pre compile it for as mu as many common runtime so that it doesn't spend 25 minutes on the first deploy, compiling Ruby or whatever. Um, those little things of, just making it easy. So it's almost like the Heroku experience, but yet you get to control where this runs. Heroku's only got their, you know, couple regions and like, we've got customers on Australia, in Australia where they couldn't use Heroku because they don't have a, a region over there and it would be high latency and stuff. So we get, you know, more flexibility that way, um, and give you some support. So that's kind of the, the sales pitch, I guess.
Ryan:How is um, AI affecting. All of your, your businesses, right? Is it making it easier, harder, uh,
Chris:Yeah. It, seems a lot riskier now because they, uh, somebody recently was saying like, you know, used to be Google made the pact with you. It's like, you make good content, we'll send people your way and we'll give you AdSense or whatever, a way to, to monetize it. And now every AI tool and the LLM is like, you make good content, that's ours. Now. We'll forget about you. Uh, and there's no like, support system for that anymore. So we, I was talked to a bunch of people, like we've seen. Google Search console show, like yeah, you get about a third, the traffic of 18 months ago because they're pushing everybody to Gemini, um, or the AI results and stuff. And like, frankly, that's way better most of the time.'cause if you search for something specific, like Postgres internals. Like there's this n two pulls, thing that I saw in a Rails pull request that, sped up Active Record in Postgres, um, by like two x 'cause it's, it's N two pulls is basically the count of records that it's returning. Um, but Ru uh, Active Record was actually using Ruby Enumerable, uh, and counting them all also in Ruby. And
Brendan:geez.
Chris:way slower. And I was like, okay, what is n two pools? Like? What, what? I want to go learn more about this. You search on Google and it's like, here's some Stack Overflow posts. Here's a really old Postgres version seven doc, about two pulls that is not in the Postgres, eight or above docs. So it's like, well, where did that go? And it also didn't explain really what these were. uh, the Gemini results were word for word taken from like the top 10 or 15 results that I found and went and like manually looked through and it was like, here's a summary of all of that. And it's just snippets from everyone and it explained it way better. You're, you're gonna have a much harder time as a content creator making a business going forward.' cause we're just seeing a decline in people arriving at our site or our videos or whatever.'cause you're gonna get pretty darn specific results from ai when you ask it and you give it the context of like, here's my application. It's Rails eight, it's Ruby three, four, uh, whatever. It's using Postgres. a lot of context. So when you are watching a tutorial from GoRails, you gotta translate it to your problem manually. But when you're using ai, you can kind of say, okay, here's what you've got. Here's. What we think we can try and apply to your code base and stuff. So in theory, it'll make us all a lot better in the long run. But in the, in the meantime, it's like no longer incentive to make content anymore 'cause they're just gonna consume it for everybody else. CloudFlare tried doing the,
Brendan:I was just gonna ask if you'd seen that CloudFlare thing.
Chris:What was it called? Um,
Brendan:I forget the name, but they were trying to monetize it right? Somehow.
Chris:it's like, okay, you know, if, if I turn that on for GoRails
Brendan:Mm-hmm.
Chris:they bought from OpenAI or where whatever it is, pay once and then they're gonna cash the results and it's not like they're
Brendan:Yeah. Right.
Chris:anything. So it's like, what do you, do you charge an AI a thousand bucks to access one of your videos or something? So I think it's gonna be weird. It's not exciting, uh, for us as a business. It's really exciting to see where all this goes. Um, but yeah, it's gonna be weird 'cause it's like if we don't have incentive to make more tutorials and content, like what's the LLM gonna train itself on? Like at some
Brendan:Yeah.
Chris:Rails 10 may come out and if nobody's bothering to put in, uh, the time to make Rails 10 tutorials and stuff, all you're gonna have is the API docs and whatever the Rails Foundation invests in. and so it's gonna be weird, like the AI's like gonna cannibalize itself
Brendan:Mm-hmm.
Chris:they haven't given creators a way to like run their business anymore. So I think that's gonna be as strange times going ahead. I'm also kind of like, I spent 11 years making a video every single week, like a break in that wouldn't be so bad. So. Looking at like, pretty much everything we have is kind of B2C in a sense. Like Hatch Box is probably the closest B2B product we have. It's the best, probably long term, you know, sticky and everything. I'm like, yeah, I could, I would really like to sink my teeth into one product, you know, and do something B2B in the long term. And, so we've been playing with some ideas like what are all the tools that we could build for ourselves to run GoRails better? And you look at support software or marketing software or whatever it is, there's really not a ton of competitors coming out. Like there's a lot of people building new things, but. For all the hype of ai. It's like I haven't seen people launching new email marketing software or support software or whatever. Like you would expect that if I'm 10 times more productive then I should be like shipping competitors to ConvertKit and Intercom and whatever, like crazy. But you know, those things all take so much nuance to get to. is really good at prototyping stuff right now, but at a certain point it needs like very specific problems to work on rather than being able to keep the whole thing in its head. And um, so we'll still be really valuable as software engineers in the long run. I think it's gonna be real hard to get into the market. I think as a junior, 'cause you're
Brendan:I, I agree.
Chris:by AI pretty easily, but
Brendan:Mm-hmm.
Chris:in the long run, like, where are we headed? No idea. It's gonna be pretty interesting I think no matter how it but. I don't know, there's a long way to go. And it's like indexed the entire internet supposedly. now AI's gotta actually get smart. It's like, it can't just be guessing what the next best word is to get better or whatever. Like, it's gotta be able to understand more, to really like, think better through things and whatnot, like humans do for it to make the next leap. We've been seeing a pretty amazing leap forward and it's like, that's cool, but that will run out of steam pretty soon, know, because it's gonna require actually to get past kind of this level that it's at, which is like pretty darn good
Brendan:Mm-hmm.
Chris:So
Brendan:Yep.
Chris:But, know, I'm sure that there will be some sort of, uh, upper bound where it's like, okay, now we're at the hard part. Like, nobody's ever indexed the entire internet, like. These LLMs have, which is amazing. So we've seen some pretty crazy things happen all of a sudden. Um, where yeah, you go analyze all of GitHub and uh, yeah, you should be able to generate some pretty good code. You know,
Brendan:Right.
Chris:lot of code on there, right? Uh, and all that code in theory is functional in a way too. Like that's hopefully people are checking in functional code. Maybe, maybe I'm wrong on that one.
Ryan:I think that, I think that depends.
Brendan:Yeah, was also thinking too just about,
Chris:don't analyze the draft pull
Ryan:Yeah, exactly.
Brendan:right, right.
Chris:merged ones,
Brendan:Yep. Well, you know, just even around
Ryan:I.
Brendan:I was kind of curious 'cause. I've messed around with, you know, vibe coating rails, apps
Chris:right?
Brendan:for prototypes. Like there's no way, I don't think the three of us, we would never throw one of those in production and think we're gonna make a million dollars because there's too many holes and all the issues. But my head always goes to, this is great to figure it out, but like, I would buy Jumpstart, because I know that you've curated it
Chris:Yeah,
Brendan:that there's maybe not enough people, or there's not as many people that are gonna think that way, but
Chris:yeah. That one's, that one's been really interesting 'cause like there was, feel like for a while that hype of like, well you don't need jump star. You can just vibe code it. And I think people are seeing like, yeah, the, it doesn't know how to do performance and think about that
Brendan:Yep.
Chris:it's writing this code. It's really just trying to make anything functional, you know? And it stops about there. But then. We are selling the, like, look, we've already thought through these things like having your payments, uh, billing stuff separated out by accounts and yada yada. We've got a whole bunch of stuff like that already done, including like the Android and iOS apps that you may eventually want to add here. So that is all still extremely useful. I think it's harder of a cell these days to say like, just go use that. But I think people are seeing like, actually yeah, there's still value in that. but for a while it felt like, know, maybe there's no need for that anymore, but I, I, that's not really how it's
Brendan:My take is, you know, is my experience is I would rather start there when I'm actually building something I wanna build.
Chris:and one of our customers, Karen, uh, Klassen is using Jumpstart for, uh, Quora, which is like this AI email platform. Um, and he's given us a whole bunch of awesome stuff, like the claw cursor rules and then a clawed
Brendan:Oh, cool. Yeah,
Chris:that. So he's, he's like vibe coding that entire product and then giving us like some of his notes and improvements and stuff along the way. and that's been really fun to see too. it's really good because like you can then not only start with billing and all the notifications and other things set up, right, but then you just give it to Claude and you're like, Hey, we have this stuff. I need to add this feature. And it's like, I can play within these boundaries.
Brendan:exactly that. That's a better place to start. Yeah.'cause otherwise it's thinking of every, everything. And you're like, wait, what is this? And then you're not even sure sometimes like, is this right or not? And you,
Chris:Yeah.
Brendan:yeah. Yeah.
Chris:long term too, like having better context windows of like, look, I want to tell you to ignore all of the stuff. Like, like this is how I tend to work lately a lot more is like, I know that I'm using Rails eight and I want to write my code in a future compatible way. So don't, use any methods that might be marked as deprecated or whatever. But like, don't even really look at Mitch much of the docs of Rails seven two or older, like just Rails eight. Where are we, where are we going? Like look at Rails main, because if I write a feature now and there's something that's going to replace it in Rails, Maine here in a couple months, let's just like backport that ourselves temporarily or whatever. I think it'll be really cool to be able to tell your LLM like, look. Ignore everything except for like the rails, uh, code base and version, whatever's in main. Like use that as your context along with all my code here. And it's gonna be tougher, but it'll be hopefully able to prioritize that stuff over the other knowledge.'cause it's gonna need to understand Ruby, but it's like if you're reading a, a blog post about Ruby two seven, probably not near as relevant as one in Ruby three or four
Brendan:Yep.
Chris:whatever, you know? And so like being able to get that context and prioritize it properly and whatever in the LLMs probably will be some of the stuff that makes those bigger leaps for us, you know, in the near term. But that's, you're, you're gonna want it to mimic that. Kind of process that you do as a, you know, senior developer where you're like, I just care about making this future proof if don't predict that. Um, like the Access Tenant gym has so many different options 'cause people use it in all these different ways, but it's like, you want the LLM to be able to make that decision as well of, well, like you would do is like, do I need all these features? Can I build a very simple version of Access tenant just with current attributes and, you know, built in rails features. If so, then let's just do that. And then I have one less dependency. and that's the better way to go. But if it is pretty complicated, like noticed there are so many different ways you wanna send notifications and lots of, uh, complexities with it. It's like, well maybe we should use that gem here. Whereas right now, like the LMS gonna be like, what's the most recommended thing? Like, that's typically what the result you're gonna get
Brendan:Yep.
Chris:a lot of things, when you ask it to do something and it's like, well, this is what everybody else is doing, so we'll do that too. that's fine for people who are just developers at their day job and don't really enjoy this for the
Brendan:Yep.
Chris:or they're not building their own products and care about, you know, maybe that's job security for,
Brendan:Mm-hmm.
Chris:Like, well, we did, we did what everybody else is doing. And that's sometimes a good argument to your boss or management, but it's not necessarily the right thing for your product or your business.
Ryan:feel like the role now is like senior architect slash project manager, right? Like that that is gonna be like the primary role of the senior dev. I mean, you gotta know code, you gotta understand the whole ecosystem. But it's how it all comes together that the LLMs usually fall apart right at that level.
Chris:Yep. Yeah, it'll be interesting to see what, what comes out of all this. It's, it's not been particularly useful for me to have it generate code, but like to replace Google search. Yeah.
Ryan:Yeah,
Brendan:Yeah. Mm-hmm.
Chris:good at that. that I, I don't know that I could go back,
Ryan:no,
Chris:it's
Ryan:me neither.
Chris:to read through 20.'cause the, the other thing is too, you search for something in, in LLM and it's like, well, than going back to Google and typing in 20 variations of like, what is an nup L in Postgres, l
Ryan:Yeah.
Chris:Postgres nup, l blah, blah, blah. Like, that used to be our workflow of like typing in 20 searches till you kind of had an idea. And now you just asked an LLM and it's like, boom, we took all the results, we found the variations and gave you the like, best summary. And you're like, oh, okay. This is great.
Ryan:If it never gets better than that, we still are like vastly improved. Right? Like, I mean, I can't, I, I can't remember the last time I went to Google or Stack Overflow, which used to be like a every minute sort of occurrence before.
Chris:I think that's, that's a huge win on its own,
Ryan:Well, I, I, we've been talking for a while and we probably need to wrap up. Um, but, uh, my, my business partner wanted me to ask you a question, which was if you had any advice for someone who's just getting started with Rails, where would you tell them? To start, and for context, he's asking for his 17-year-old son who has shown some interest in programming, but with the shifting landscape, you know, I, I've already told him to, to go to Go Rails, so you can, you can check that off because I, I think it's a great resource. So, and, uh, you know, I've already said it a couple times, but thank you for, for doing it. but what would you, what advice would you give to that?
Chris:Well first thanks. I appreciate that a lot. It's, uh, I, I remember so fondly being in those early days, like for one, go build stuff from scratch. Like, you will understand things way better if you do that. So like when I was in school, I wrote, I was using IRC all the time and I was like, what is this protocol? So I opened Ruby, I, uh, required the socket, um, library librarian was like connect to, um, oh, what's the, the popular IRC channel? I connected to that and it's like rec, I'm just printing out messages from the server. So I connect, it says a ping, and I don't have any code written. So it prints out the ping and then a few seconds later it's like disconnected.'cause you never sent a pong back. And I was like, okay, well uh, let's go build that. And so eventually, like I built an IRC client, you know, even just if it was just basic to print out stuff and it, you couldn't interact with it, it was still like incredibly valuable to just go build that. So for me, that's the things that I've been able to do to learn the most. Building your own web server, understanding how HGDP works. Um, you know, talk to an IRC channel that's super fun. Uh, go learn how to compress an image. That was something that I did in assembly in school. It was like we had this assembly class. My teacher was like. some binary data. Pretend that you're, uh, working on code for the Moon Rover and you've got a camera and you need to send an image back to Earth. So obviously you probably wanna compress it. So build a little color palette, replace each pixel with the, uh, index of the color and the palette and so on. Then the next project was like, here's the same image. Go find the biggest blob in there because we are looking front of us and don't want to run over any rocks and get stuck on the moon because obviously we can't go, uh, reset that, uh, or pick it up. So there were lots of little things like that that just really helped a lot. But probably the most important thing was like, go build stuff that you're gonna use every single day. Um, so like the, you know, tool to download software offline on a flash drive. Like I used that every day, and then I would get frustrated with the like rough edges and the warts and I'd go fix 'em, and then I'd be like, man, I want to build this plugin. Or something, but I've never, I don't know how plugins work. How do you do that? Like how do you dynamically re like search this folder if there's any files in here, then uh, go require them dynamically or whatever. Never done that. Go build that and you're like, oh, that's how plugins work. And uh, it just had to go build stuff.
Ryan:Yeah.
Chris:like my senior project was fun. We did a two, two things around when I was graduating. We did a project for the school. Um, your senior project was very waterfall. You had a semester of planning your project and a semester of building your project before you showed it. And. The admissions department had a form that was a dropdown of times and dates to go get a tour of campus. So we pitched them. We would build a Rails app that would, uh, replace that.'cause this form would submit to a Pearl Script that would send them an email. They would copy the contents of the email into a spreadsheet and they would share the spreadsheet with everybody in the office. But if anybody modified it, or two people modified it at the same, same time, it would create a duplicate that they had to manually merge back together. It was a nightmare. I was like, we could do that in Rails. So I built a calendar in Rails that became my first gem.'cause I'd never. Um, built a gem before. Um, then I found a bug in Rails doing a group by Query and I asked, the team was like, Hey, if this is probably my fault, uh, 'cause I'm brand new to Rails, but it feels like a bug in Rails. If it is, will you giving me like a direction on how to make a test and a, a fix for this? Uh, 'cause I'd love to fix it. So I contributed to Rails while I was in school. Um, and that project, I actually convinced the school after we graduated them to pay me 5,000 bucks a year to run it and maintain it for them, which is fun.'cause then during job interviews I was like, yeah, so, uh, I got the university to pay me to go to school there, which is a
Ryan:Yeah.
Chris:Then when I was also looking for jobs, I was like, you know, this is the weird thing about learning, is like you're always doing these stupid little projects from scratch, but you're not doing anything in production. But what is your job gonna be? It's only working on stuff that's in production. Like you're almost never building stuff from scratch. So. I said, I'm going to build a Rails app that is designed horribly on purpose. I'm going to build a to-do list that the Rails app is just a scaffold of to-dos. Then I'm gonna deploy it to Heroku. I'm going to then, uh, upgrade it and say, now we have multiple to-do lists, but what do you do with the data that was there that didn't have a to-do list? We gotta build a default to-do list. We gotta migrate that copy all those to do, or like, you know, reassign them and associate them with that new to-do list. Um, then I didn't have any security on there. You anybody could go into any to-do list and delete stuff, so then you could, I added the ability to go register and then claim a to-do list as your own. If you did that, then uh, it wasn't publicly accessible or modifiable, then it was like, well, these URLs are just 1, 2, 3, 4, 5, and so on. Let's make them friendly. So I learned how to do that and just like. that as my way of like, what is production life like as a developer?'cause you don't learn that in school. You don't teach yourself that ever. and then I interviewed with this company and they actually sent me a link to a to-do list after I interviewed and they were like, hire Chris Oliver. And it was like, some to-do is like, send him a job, offer check and you know, they used my own to-do list 'cause I showed them that. And it was like,
Ryan:that's very cool.
Chris:anything, it wasn't anything like that. Literally anybody could do that. But people aren't thinking ahead of like, I am going to work for this company. What am I gonna be doing there? They're like a little too focused on they're at today and not where they should be. And so if you just put two and two together, you can stand out pretty easily in some, some job interviews. I mean, junior job interviews probably are way harder now than they were back then, but. I, I just try to do that kind of stuff, uh, and learn the fundamentals, but also like ahead a bit of like, what do they want, what are they looking for
Ryan:I always wanted to take all your go rails, screencasts and try to combine them into some sort of functional app. Now that's probably a, a fool's errand because you're like, the, the breadth of content that you create is sort of all over the map, but. I always see these like different components. I'm like, oh, like I could see how that could fit in with this one. And you know,
Chris:Mm-hmm. Mm-hmm.
Ryan:process of figuring out how it would all go together and then actually figuring out how it goes together, I think would be a interesting exercise. I.
Chris:yeah. Yeah. That's something that's really hard 'cause. People lose attention in 10 minutes or 15 minutes watching a video. But reality of it is like, we're only teaching a tiny part of it in those videos, but the, okay, how do you broadcast a comment but also have like reactions to it so it shows up in real time when somebody else in a different browser created that comment. But we also have to have this dynamic little, you know, interaction. Um, and we wanna highlight that if you, the current user has done that, but then when we broadcast, we don't know who's the current user. So how do we wire all that up? And those are really where it all comes together, but they can be pretty tough
Ryan:Mm-hmm.
Chris:put together. Um, an example that isn't, uh, a six hour video or
Ryan:Right.
Chris:you know?
Ryan:Well, and that's, that's where trying it and learning and, you know, asking questions, I think becomes the, the useful exercise. Right.
Chris:Yeah, I think that's the. Just most valuable thing that I've, I've done and I just do it all the time. Still to this day. Like the, the Ferrum PDF gem, I was kind of like, you know, it'd be cool if we could export PDFs with Chrome but didn't have to have no JS and puppeteer. And I was like, well, I'll just see, you know, I'll spend this afternoon fiddling with that. See what, what would the API look like if I was to do that. and like Grover already does that, but it requires no JS and puppeteer.
Ryan:Mm.
Chris:so I was like, well, um, Cuprite is built on rum, ham's, a ruby implementation of this, you know, Chrome driver protocol thing. Uh, maybe I can just grab that, put a little layer on top and see how it goes. And I published that. Never used it in production and like issues start showing up 'cause people saw it. And then now I have to actually think through this for real. And like the last few days after Rails Comp, that was one of the projects we. We hacked on at RailsConf, didn't get anything done then, but had enough good conversations. It was like, I think I know the direction. So the last couple days I've been refactoring that and implementing those ideas just mostly for my own practice and like how do I build a nice clean API. So we, one of the ones I just did like two hours ago was like, inject a method into all your controllers to render A PDF or render a screenshot, it was kind of generically named render under pdf. And um, it just shows up out of nowhere. And I was like, you know, I think I've seen the way to inject a renderer into rails. So I just refactored this like two hours ago. And now when you want to render A PDF, instead of saying, calling this random method that appears you like just how you can say render JSON and give it a value. Um, with that hash we now have render PDF and you can give it options that will be passed over to Chrome and it will take like the current view and render that out for you and stuff. So it's pretty cool how that works out, although I think there's some edge cases I haven't thought of yet, but it feels super natural now to fit into Rails that way.'cause you're already used to doing render JSON or render text or whatever. And so
Ryan:Yeah. That's awesome.
Chris:a clean fit. But those are like things you can do every day to fiddle with how do I make things a bit more realsy or whatever. And yeah,
Brendan:That's really cool. Well, thanks again for joining us. It's been a great conversation. I feel like we could keep going for a long time.
Chris:probably. Yeah. Cool. Well
Brendan:else besides Yeah, definitely. And is there anywhere else, besides all the places we've already talked about that you wanna point people to, uh, on the internet with you?
Chris:Uh, I don't think so. Um, you can find me on Twitter or pretty much everywhere at EXC, ID three. Um, and yeah, you can find Go rails@gorails.com, but those are kind of the main ones.
Ryan:Yeah. All right. Well thank you so much for everything you do, uh, for the community, and thank you for joining.
People on this episode
Podcasts we love
Check out these other fine podcasts recommended by us, not an algorithm.
Code with Jason
Jason Swett
Standing in the Fire
Very Good Software
Remote Ruby
Chris Oliver, Andrew Mason, David Hill
IndieRails
Jess Brown & Jeremy Smith
Startups For the Rest of Us
Rob Walling
Vibing with AI Code
Brendan Buckingham