Christopher WeisUSOmarchy revives old hardwareAI removed the implementation wallAgentic coding, not vibe codingLocal vs frontier modelsOwn your dataContribution over glorySelf taughtbuilder/builders/christopher/avatar.jpg
// builder #12 · interviewed Oct 17, 2026
Christopher Weis
Automotive pro who never bought a new computer, revived old boxes with linux, then hit Omarchy and AI and turned a year of stalled ideas into shipped software: an internal dispatch app in real daily use, 2 Omarchy plugins, and small model research where he reports a 3B model jumping from 58% to 90% on bug fixes.
first computerAn old Dell with a 2 GB hard drive. He was told he would never fill it up. He said watch me.
linux since2014
omarchy since2026
rigHand me down hardware only, on principle. Small model work on a gaming laptop with an 8 GB card.
found omarchyMade an X account to share his AI built apps, kept scrolling into the Omarchy crowd, and has been obsessed since
modelschatgpt, claude, smollm3
ai workflowPlans an idea in 5 minutes, walks away for a coffee, comes back to a mapped out working prototype
// life timeline
An old Dell with a 2 GB drive. Watch me.
Out of high school and into the automotive industry
2014Revives an old computer with Ubuntu and a TV tuner
2025Spends over a year on an app idea with no way to build it
2026Opens an X account to share the apps he builds with AI
2026Finds Omarchy and goes completely obsessive
2026Ships an internal dispatch app, 2 months in daily use at 2 locations
2026Publishes Omarchy Vault and Omarchy Beam plugins
2026Reports a frozen SmolLM3 3B going from 58% to 90% on 100 MBPP+ bug fix tasks with deterministic repair tools
2026Builder #12 on omarchy.builders
// 00
The rundown
Christopher Weis is a rock star, and he got there without ever buying a new computer. He came up in the automotive industry, not a developer by trade, carrying ideas around for years with no way to build them. Then Omarchy and AI showed up, and the ideas started shipping. An internal dispatch app in real daily use. 2 Omarchy plugins, one already in the plugin store. Small model research that makes a 3B model punch way above its weight. His advice to everyone else fits on a shop rag:
"it's really hard to diagnose a car with a flathead screwdriver"
Get the right tools. Then go fix the thing.
“it's really hard to diagnose a car with a flathead screwdriver”Christopher Weis · L174
// 01
Watch me
His first computer was an old Dell with a 2 GB hard drive. Somebody told him he would never fill it up. His answer: "Watch me." He started with games, the way every good story starts, and then "it got really fascinating."
Out of high school he fell into the automotive industry. Cars took over the day job, but the itch never left. He just had to scratch it with whatever hardware came his way.
// 02
No new computers
That hardware was always a hand me down, on purpose. "I've never bought a brand-new computer because I don't believe in e-waste."
So when an old box landed on his desk, he asked the right question: how can I use this today? The answer was linux. Ubuntu, around 2014 or 2015. He tuned the machine up, figured out it could do a lot more, and bolted on a TV tuner. "I thought that was fantastic, to be able to record TV on a computer."
That is the whole Christopher philosophy in one move. Old machine, free software, new life. Long before Omarchy, he was already a linux guy rescuing computers from the landfill.
// 03
Completely obsessive
A few months ago he opened an X account for one reason: to share the apps he was building with AI. He looked around, saw how crowded that corner already was, and kept scrolling. He found the Omarchy builders shipping in the open, sent his first ever cold message, and that was that.
"from then on out, it's been completely obsessive with Omarchy and AI."
He showed up to the Omarchy Sunday Spaces. He found people who got it. And for someone who had spent years with ideas nobody around him could see, that mattered: "it's reassurance that what I am doing and my ideas actually have a value to other people."
// 04
Uber for the crew
Here is the part that should make every developer sit up. Christopher built an app that real people use every day. For 2 months now it has run at 2 locations, built on Omarchy with a little ChatGPT and Claude in the mix.
It is Uber for an internal crew: valets, shuttle drivers, mobile service runs. The old way was a sprawling group text or a Teams thread. Now everyone gets a text, answers yes or no, and the ride gets covered. Simple. Useful. Live.
He had chased an idea like this for over a year with no way to build it. The tools caught up with the ideas, and the ideas won.
// 05
5 minutes and a coffee
Before AI, a vision board meant a long fight with Microsoft Paint, ending with one screenshot of an idea already living in his head. Now he does it differently.
He spends 5 minutes planning the idea. He walks away, grabs a coffee or a bottle of water, and comes back. "There's a vision board, everything's mapped out, there's a working prototype."
Months of work, done between sips. He does not memorize what every byte of code means. He directs. "Hey, Mr. Agents, I need you to move this button from here to here." Weeks of meetings and back and forth over whose calendar is free, all of it gone. His summary: "Instantaneous. Boom, done."
He calls it "slingshotting yourself through that entire process." The rest of us call it being a product manager with a team that never sleeps.
// 06
A big brain on a small card
Then there is the research, and this is where he gets loud in the best way. His question: why not have a model of around 280 billion parameters running on an 8 GB card? Slower than the frontier, sure. But on hardware a normal person owns.
His approach uses a big model as a database that a smaller model can call on. And he has receipts. On a gaming laptop with 8 GB of VRAM, he reports that deterministic local repair tools took a frozen SmolLM3 3B model from 58% to 90% on 100 MBPP+ bug fix tasks, without adding a single neural token. His post has the details. Same small model. Better tools. Way better results. Sound familiar? Flathead screwdriver, meet the full toolbox.
He is also studying LoRA for a bigger idea: a community marketplace of skill packs. Not Markdown skill files, but trained packs. Want Rust? TypeScript? Automotive smarts? Grab the packs, and "each model is custom-tailored per individual."
// 07
One thing, really well
Building is easy now. Finishing is the new hard part. Christopher has a sharp read on the trap: feature creep and over analysis paralysis, where "you end up building an entire operating system when your goal was to only do one specific thing really well every single time."
His fix is pure product sense. If a new user has to learn a whole new language to use your app, you lost them. Give them one job and one step, and "it will speak for itself and then launch like crazy."
He ships what he preaches. Omarchy Vault turns a spare drive into a private cloud on your home Wi-Fi, with QR code transfers to and from your phone. Omarchy Beam beams your clipboard to any device. Both MIT, Vault already in the Omarchy plugin store, both doing one thing well. He built them for himself first, which is the best reason there is.
And the big one, per his own site and bio: TouchWorkstation, a self hosted, touch first way to drive your linux workstation from your phone. Terminal, repos, live previews, agents. Open source.
// 08
Mentally life changing
Ask Christopher about the Omarchy community and he lights up. "I've never seen anything like start from nothing and explode to what we have now as fast as it has." He reads every post from the builders he met on X and looks forward to what they share. His verdict: "it's mentally life-changing."
His message to the rest of us is the one he lives. "Follow your heart when it comes to what you want to do, and use the tools available to you." You can build an entire application with the tools already out there.
So here is Christopher Weis, builder #12. Rescuer of old computers. linux guy since the Ubuntu days. Automotive pro turned shipper of real software. Small model scientist who reports a 90% result. Christopher, you are a rock star.
L28Christopher: I do not, but I honestly love to find out some more.
L29Nix: All right, cool. So you can't really go to the website right now, but the Omarchy Builders website. I've interviewed 11 people now, and I just talk about Omarchy, how they found it, what they do with it, and then I do a full spotlight of them, uh, based on what they talk about.
L30Christopher: Awesome. I think that's fantastic. I went to the website briefly, um, trying to find the link to the Zoom, and, um, yeah, it looks amazing, by the way.
L31Nix: Yeah, thank you, brother. Thank you, man. Well, it's cool to finally put a name on the face, because I've—we've kind of orbited.
L32Nix: We've kind of orbited a little bit, and here we are, man. That's really cool.
L33Nix: Um, so you consent to having this recorded and me putting your, uh, you know, I don't put personal information, you know, just basics of, uh, a lot of data where, like, the things that you say that are similar to the other folks and make correlations to how people are doing the same type stuff, and then, uh, maybe you've got some stuff that's unique and we'll spotlight that, and that's where we go.
L34Christopher: Yeah, absolutely. I mean, the more [redacted], the more people you have in, I mean, it's problem of building a community, right?
L35Nix: Brother, you got it, man. That's exactly—like, it—I'm so glad you get it, because believe it or not, some folks I talk to don't quite understand it.
L36Christopher: Oh, no. It's all about being open.
L37Nix: Yeah, exactly. Um, so you're building a lot of stuff, man. How did you—like, let's start with the first computer that you ever had and what you did with it and how you came across it, and then go from that part, build an arc, a little bit of a story, a little stuff in between.
L39Nix: And then jump to, uh, um, jump to where you found Omarchy, your Linux experience previous to that, if you're a developer, if you're not, how do you use it, what does it make your life better. It looks like you're in a Tesla, are you?
L40Christopher: Yeah, this is an old Cadillac SRX.
L48Christopher: So basically, this all started in high school when I got my first computer. It was an old Dell. Um, it had a 2-gigabyte hard drive, and I was told, "You'll never fill it up."
L52Christopher: And I was told, "You'll never fill it up." And I said, "Watch me." And then it all started with, you know, games and stuff like that, and then it got really fascinating.
L53Christopher: And then I fell into the automotive industry out of high school. Um, I lost touch with the cutting-edge technology.
L63Christopher: I kind of fell into Linux. Um, and then, yeah, I kind of, you know, modified the computer, got it going, figured out it can do a lot more. I bought a TV tuner for it.
L64Christopher: I thought that was fantastic, to be able to record TV on a computer.
L70Nix: Yeah. So keep going, keep going. Sorry to interrupt.
L71Christopher: No, you're perfectly fine. Um, so yeah, in the automotive industry. And just recently, um, I created an X account, I'd say probably 2, 3 months ago, specifically for releasing all the applications I built using AI for [redacted].
L73Christopher: And then realized how saturated that environment is already. So I just kind of threw that out. And then I was scrolling one day, and then I saw this guy on X, his name was Nix.
L74Christopher: I don't know if you know him or not. That's you.
L75Christopher: And I was like, "You know what? I'm just going to do my first reach out and just message this guy, because he looks really cool and he's building a bunch of stuff."
L77Christopher: And then—and then from then on out, it's been completely obsessive with Omarchy and AI, and right now I'm trying to get a large model to run on a small computer by dissecting a large model, using it as a database, and then using a smaller model to then use the bigger model as a database, because I want—it's going to be slower than your frontier models, but at the same time, why not have, you know, a 20, I don't know, 280 billion parameter model running on an
L78Christopher: 8-gig card? I mean, that sounds fantastic to me.
L84Christopher: So basically, having a marketplace—like, my idea was to have a marketplace, not like a plugins, but more of like a skill set, but not like a Markdown file, but like an actual skill packet that the community can build and train. So then ultimately, you can download a small model and then build it to what you want it to do.
L86Christopher: Meaning, here's the conversation part: I want to be able to do coding. Just as an example, I want to do coding in Rust. I want TypeScript.
L87Christopher: I want some artificial intelligence for automotive. Um, I want to be better at speaking this language.
L88Christopher: So then you download these packs, and then each model is custom-tailored per individual.
L90Christopher: Yeah. So that's what's going on in my head right now. Um, other than that, I built a couple plugins for Omarchy, specifically for myself, and I released them on the plugin store.
L91Christopher: Um, I haven't checked on them in a couple weeks.
L92[redacted: private, lines 92 to 94]
L95Christopher: So, yeah. But no, other than that, I mean, just within the last month or so, just the Omarchy community, like I said in one of the Sunday meetings, is it's reassurance that what I am doing and my ideas actually have a value to other people, because I'm strictly from the automotive industry.
L97Christopher: You know, you're kind of left in the dark. You know, I kind of compare it to mushrooms, and excuse my language, you know, we're left in the dark, fed s***, expected to grow.
L100Nix: And I got a feeling like the automotive industry, they may be technical, but they're not great, I bet. You probably shine.
L101Christopher: Yes. And [redacted]. And then, oh, yeah, I forgot.
L102Christopher: Um, I've been using one of my applications in two [redacted] now, two months, that I built using Omarchy and, um, a little bit of ChatGPT and Claude in there.
L106Christopher: It's internal. So it's your own drivers, like having an organizational system for valets and shuttles and mobile services [redacted], and it's Uber for that.
L108Christopher: So depending who's available for a shuttle, you know, like the old method was text, a text message thread or a Teams message thread, but now it's everyone gets a text message. They say, "Yes or no, I'm available. I'm not available."
L109Christopher: Um, and then from then on out, it's Uber, but for [redacted] internally.
L110Nix: Right. Now, you said that you were, or you are not, like a developer in by, like, normally?
L116Nix: So six months ago, did you have these kind of ideas that you wanted to do these kind of applications for [redacted]?
L117Christopher: Yes. I've been working on one of them for over a year, and then fell into AI. But just as, like, it would be kind of cool if this happened idea.
L122Nix: Omarchy, your ideas were stagnant because of, no disrespect, a little bit of skill, lacking in skill. Not lacking, but you know what I'm saying. You didn't have them.
L125Nix: You enter Omarchy and AI, and that jump starts or propels your ideas and your ability to drive it forward. And it sounds like you said you were working on that app for a year, and then Omarchy plus AI, and you've already shipped. And you got customers.
L127Nix: So talk about how the mixture of your ideas that were a little bit groundlocked, and then enter AI, Omarchy, and what you're capable of now.
L128Christopher: Okay. So basically, from idea, you know, I would, this is very rudimentary, but I didn't know anything, so I'd be using, like, Microsoft Paint.
L132Christopher: And then, you know, at that point, you're exhausted. Like, what is this one screenshot going to do for me, other than solidify the idea that's already in my head?
L134Christopher: Um, but then you get these tools come out, Omarchy and AI, and now it's, I have an idea, I can spend 5 minutes planning the idea, walk away, grab some coffee, grab a bottle of water, and come back. There's a vision board, everything's mapped out, there's a working prototype.
L136Christopher: Very, very, very, very minimal MVP that at least gives you the idea that would usually take, I don't know, I guess I don't know how long it would usually take to do that, because I don't have the developer background.
L142Nix: So instead of having skills, you have. Yep. Could you repeat that?
L143Christopher: It kind of broke up there for a second.
L144Nix: Yeah, it did for me too. So you, you know, most of the time, folks that are product marketing and developers, or not developers, creators, they start out as 10 years of development, and then they say, "Well, I think I can do this," and then they develop it on their own. And it sounds like you had the ideas, and you skipped that whole 10 years and went straight to being a product manager, managing the tools, which are essentially developer human, you know, could have been humans.
L147Nix: If you think of, if you think of Omarchy and AI agentic coding, you know, those could be considered like human-type things, if you look at it that way. And now you're managing 10, 15, 20 different agents, and you're the boss or the manager, and things just happen for you.
L149Nix: I'm not great on that. I don't want to put words in your mouth, but I'm trying to prompt you to, like, I'm going to make shorts out of this and make you.
L150Christopher: Okay. Okay. So essentially, instead of learning the entire skill sets of what each single byte of code and what each language means, you can end up slingshotting yourself through that entire process by essentially saying, "Hey, Mr.
L151Christopher: Agents, I need you to move this button from here to here, or I need this workflow to work like this." And at that point, it just happens based off of the tools that we have, which used to be a meeting.
L153Christopher: I'm assuming it would have been like a meeting, conference call, a play-by-play of going through the application with the development team. Can it be easily translated as 5 or 6 words quickly typed on your phone or talked through the microphone?
L154Nix: Right. And again, that would be weeks of back and forth. "When can we meet?
L155Nix: I can't meet until next Thursday." "Oh, crap, I can't meet that day.
L163Nix: What do you see in terms of, like, the downside of being able to try anything? And we end up doing everything, and, you know, there's like this, we went from being limited to being able to do everything, and now it's kind of out of control that we're starting 100 projects and finishing 3, versus having 100 ideas and only doing 3.
L164Christopher: Yes. And I could probably speak to many people when you get to what's called over-analysis paralysis or feature-focused, where you end up snowballing everything into it, and you end up building an entire operating system when your goal was to only do one specific thing really well every single time.
L166Christopher: So when you're trying to get someone who doesn't know about your product to use your product, you're essentially teaching them an entire different language, and at that point, you lost them. To where if it's just one specific thing, "Hey, this will do this for you, and all you need to do is this," it will speak for itself and then launch like crazy.
L174Christopher: I'd say, follow your heart. Okay. Follow your heart when it comes to what you want to do, and use the tools available to you, because it's really hard to diagnose a car with a flathead screwdriver.
L178Christopher: As far as the Omarchy community, I mean, I've never seen anything like start from nothing and explode to what we have now as fast as it has. Like, I'm having a conversation with someone I met on Twitter or X, and I look forward to all of his posts. I look forward to everything he has to say, and everything that this community has brought to me.
L179Christopher: I mean, I don't know, it's mentally life-changing. I think you can put it that way.
L184Nix: Be looking for your spotlight and your tweet out and your, so I'll take all this, summarize it. Will you verbally say, "You authorize me to use this"?
