// life timeline
1980Asks a school library computer how fast an airplane can fly
1982TI 99 and the Bojangles listing
1997Brings Red Hat into a room full of SCO Unix boxes
Five years of quality systems for food and beverage giants
2026Moves his agent fleet from Claude Code to Hermes
2026Omarchy becomes the daily driver
2026Ships Omaudible
// 00
The rundown
Chris Gray is a rock star. He has been in tech for more than 30 years, he brought linux into a room full of SCO Unix boxes back when that took nerve, and today he runs a fleet of 14 Hermes agents that do real jobs for real clients. Not demos. Jobs. His rule for all of it fits on a sticker:
"it's only vibe coding if you don't have a plan."
“it's only vibe coding if you don't have a plan.”Chris Gray · L192
// 01
How fast can an airplane fly
Around 1980, a school library got a computer and young Chris sat down at a bare command prompt. He typed the obvious question: "how fast can an airplane fly?" The librarian explained that is not how computers work. Chris tells that story because, as he puts it, "but now we can."
A few years later his family got a TI 99. He typed in a BASIC listing called Bojangles for two hours straight, and at the end a little ASCII figure danced across the screen. That was it. Hooked. "And that's why the minute I saw AI, I was like, well, I've, you know, this is exactly what we've waited for our entire lives."
// 02
Red Hat in a room of SCO
He went to college planning to teach high school English. The computers had other plans. Around 1997 or 1998, Chris says, he became the first in house IT hire at a regional oil company that distributes for ExxonMobil. He walked into a room full of SCO Unix boxes, introduced linux, and built file and mail servers on Red Hat.
He has perspective on the internet's favorite hobby. People who think heads explode over Omarchy, he says, "should have seen the split between Red Hat and Fedora."
Next came five years at a consulting company building quality systems, LIMS and statistical process control, for Kraft, Hershey, Miller Brewing and Campbell Soup. Since then he has lived on the client side of the table: project plans, requirements, teams, rollouts, integration. The stuff that decides whether software actually lands.
// 03
The third act
That is exactly why AI clicked for him. "AI is gonna give guys like me a third act," he says. Decades of knowing what good looks like turn out to be the scarce part now. "I can smell a project going bad, and it doesn't matter if it's a room full of coders or if it's a room full of AI."
His advice is to start every AI conversation with people and business goals, not plumbing, and to treat the agent like a coworker. "if you're treating it like a, a screwdriver, uh, you're, you're missing about 70% of the value." The bottleneck, in his view, is not the tooling. It is imagination.
// 04
Fourteen agents with jobs
Chris builds agent fleets for clients in healthcare, legal and manufacturing. They watch security systems and manufacturing systems, and he ships practical things on top: a refund system for a healthcare company, HR onboarding, and client onboarding for law firms. Chris says the refund system alone would once have been a real project with real time attached.
The fleet started in Claude Code, one agent per directory, with cron spinning agents up on the fly. Then, Chris says, Anthropic changed its billing so that pattern was not allowed anymore. He moved the whole thing to Hermes and likes it a lot. He runs it all with Herder across a cloud server, a laptop and a desktop. "I was clicking tabs three months ago. Click, click, click." Not anymore.
On the side he builds for Omarchy itself. Omaudible puts Audible in the Omarchy bar, and he has been working on omamail, a mail plugin, plus a plugin for Hermes. He writes it all up at latentoperator.ai.
// 05
Exactly what we need
Chris always kept a linux machine around, Ubuntu or Debian, but it was never the daily driver. Then he heard a podcast tearing Omarchy apart for being AI first. His reaction: "And I was like, no, that's exactly what we need." He started right before the next big Omarchy release landed. When you are talking to 14 agents at once, an OS built to juggle a ton of windows fast is the whole point. His Windows laptop is still in the bag. He says he has not taken it out in a couple of months.
You are a rock star, Chris.
- X: @LatentOperator
- GitHub: latentoperator
- Site: latentoperator.ai
// 06
In his words
"it's only vibe coding if you don't have a plan."
"AI is gonna give guys like me a third act"
"I can smell a project going bad, and it doesn't matter if it's a room full of coders or if it's a room full of AI."
// their setup
Setup
hardware
Cloud server where most of his dev work livesdaily · server
Local laptop on Omarchydaily · laptop
Local desktopdaily · desktop
tools & models
hermes-agentherderclaude-codecronclaudepythonqmljavascriptagent workflow
A fleet of 14 Hermes agents→each with a specific job for a specific client
// full transcript (cleaned)
Show the cleaned transcript (185 lines) · open full page
L1[redacted: private]
L2Nix: What's up?
L3Chris: Hello sir, how are you?
L4Nix: I'm doing just dandy. More importantly, how about you?
L5[redacted: private, lines 5 to 17]
L18Chris: Uh, well, I, uh, really appreciate that. And, uh, I feel like, uh, so much of the stuff that all of us do these days is kind of in anonymity. Uh, we're, you know, sitting behind screens.
L19Nix: Yeah.
L20Chris: Um, I've never been a big social media guy.
L21Nix: Right.
L22Chris: I've never been a build in public guy.
L23Nix: Right.
L24Chris: Uh, but I've started doing some of that because you don't know how many of us are probably reinventing the wheel every day and not needing to. Uh.
L25Nix: Yeah. And our, our enemy is the AI bros that are like, just load Qwen one time on your 8 gigabyte GPU and get frontier models. And they literally don't run businesses on AI.
L26[redacted: private, lines 26 to 43]
L44Chris: I have. And in fact, I've, uh, watched some, uh, in preparation for today.
L45Nix: Awesome.
L46Chris: Uh, I don't, I don't know that I'm gonna be a world different than a lot of the guys you've already talked to.
L47Nix: Uh, you're already way ahead of most of them. I have a way of bringing it out of people, my friend. So can you just verbally consent that you're okay to put, uh, our conversation on the website?
L48Nix: And at the end.
L49Chris: Absolutely.
L50Nix: And at the end, you'll have a chance to speak up and say, I don't want this in there. I don't want that in there.
L51Chris: Okay.
L52[redacted: private, lines 52 to 118]
L119Nix: So let's get this thing started 'cause I feel like I could talk to you all day long and we won't have your content. Okay.
L120Nix: So here's how it works. Um, what I wanna talk to you about is, is I'm gonna ask you pretty open-ended questions and I'm gonna shut up.
L121Nix: I'm gonna continue.
L122Chris: Okay.
L123Nix: Continue to ask you pretty open-ended questions. I want you to just, just work and just like, talk through it and I'll just keep.
L124Chris: Okay.
L125Nix: The ball spinning, keep you talking. Um, and what I want you to do is, is start, I'm gonna give you a framework and then you can go.
L126Chris: Right.
L127Nix: Wherever you want. The framework was the first time that you looked at a PC and you went, what is that? And the fir, the second thing is like, when did you get to play with it?
L128Nix: 'Cause what I'm hearing from a lot of people is they see somebody else using them and they don't get a chance to access it for a little while. That's been a common theme.
L129Nix: And then.
L130Chris: Yep.
L131Nix: Go like the middle ground. Like, are you a guy? Did you grow up in the Windows world?
L132Nix: When did you get to linux? And then if you like Omarchy, Omarchy, go into that and then we'll go from there.
L133Nix: So like, what was the first.
L134Chris: Yeah.
L135Nix: Computer that you ever saw and, and what, when did you know you were gonna be a nerd?
L136Chris: Uh, I think I was a nerd from day one. The first computer that I ever recall seeing. Um, so the school library had a computer.
L137Chris: I'm guessing it was probably an Apple, uh, or Commodore. This would've probably been around 1980.
L138Chris: Um, uh, just based on what I can remember. Um, I tell this story sometimes.
L139Chris: It illustrates how long some of us waited for AI. Uh, but the first time I was presented with a computer, it just had a command prompt.
L140Chris: I just had a prompt on the screen, a cursor on the screen, and I typed in, how fast can an airplane fly?
L141Nix: Right.
L142Chris: I hit enter. It dropped down to the next line. And the librarian said, that's not how these things work.
L143Chris: You know, you can't.
L144Nix: Right.
L145Chris: Just ask it a question. Um, but now we can.
L146Nix: Yeah.
L147Chris: You can now. And so, uh, that distance in time from, let's say, you know, 1979, 1980, wanting your first impulse being to ask a computer a question and that not being how they worked. Um.
L148Nix: You're a Timmy Ty.
L149Chris: Yeah. But that desire was always there. And that's why the minute I saw AI, I was like, well, I've, you know, this is exactly what we've waited for our entire lives.
L150[redacted: private, lines 150 to 153]
L154Nix: Yep.
L155[redacted: private, lines 155 to 156]
L157Chris: So I'm guessing by '82 or '83, uh, we had a TI 99. And that's the first time I ever touched code. It came with a book, uh, just, uh, how to write basic or, uh, basically just, uh, code samples.
L158Chris: But there was one in it called Bojangles and you sat down and you typed for two hours. And, uh, at the end of it, you got a little dancing figure on the screen, uh, ASCII figure on the screen.
L159Chris: Um, and that's the first time I ever, uh, attempted anything with code or anything like that. But I was completely hooked.
L160Chris: Um, uh, and, you know, I've never really honestly considered doing much else. I went to college thinking I wanted to be a high school English teacher and wound up doing, you know, 30 plus years in, in tech.
L161Chris: Um, and, you know, it's, uh, I just don't know that I ever really considered doing much else.
L162Nix: Right, right. That's cool.
L163Chris: So my first IT job, um, and let me know if I'm talking too much, but.
L164Nix: No, that's fine.
L165Chris: My first IT job.
L166Nix: I'm talking too. Please do get it out.
L167Chris: Yeah. So my first IT job was around 1997 or '8, uh, was the first in-house IT hire at a regional oil company. And they were, uh, distributor for ExxonMobil.
L168Chris: Um, and, uh, they had had IT people they'd worked with from the outside, um, but they hired me in-house, uh, just on a recommendation. Uh, completely a lucky break.
L169Chris: I've had several lucky breaks in my career. That was the first one.
L170Chris: Um, but I walked into a room full of SCO Unix boxes.
L171Nix: Nice.
L172Chris: Uh, and had no idea where to start. But, um, I was there for several years, supported, uh, Unix and Windows desktops. Uh, I introduced linux to that company.
L173Chris: We built, um, uh, file servers and a mail server on Red Hat.
L174Nix: Mm-hmm.
L175Chris: Um, which is one of the reasons I find the whole Omarchy controversy stuff funny is if people think, um, heads explode over Omarchy, they should have seen the split between Red Hat and Fedora.
L176Nix: Right.
L177Chris: Um, where.
L178Nix: Right.
L179Chris: Yeah, it was, it was ugly.
L180Nix: Right.
L181Chris: Um, and so this is nothing. Um, a after I'd been at that job a while, I went, uh, to work for a consulting, uh, company. Uh, and this was the second lucky break of my career, but they needed an IT guy.
L182Chris: They needed somebody with kind of a tech generalist. And they literally needed, this was on a Thursday, and they needed somebody that could get on a plane on a Sunday.
L183Chris: Um, and so, uh, they were desperate and I was available. But, um, but, um, I did five years, um, with some of the largest food, man food and beverage manufacturers in the world.
L184Chris: So Kraft, Hershey, Miller Brewing, uh, Campbell Soup, uh, all of these companies. And we were doing, uh, quality systems.
L185Chris: Um, and so it was, uh, uh, software systems put around LIMS and, uh, uh, statistical process control and things like that. Um, I have typically beyond that point in my career been more client-facing.
L186Chris: Uh, so project, project plans, requirements gathering.
L187Nix: Okay.
L188Chris: Uh, putting teams together, uh, doing rollouts, doing integration work. Um, and I point that out because the, the only, I was never the guy that was gonna sit in a room and code for six months or, uh, write SQL queries for six months. I was way more the front-end guy, way more, uh, the project plan guy.
L189Chris: AI is gonna give guys like me a third act, uh, where the, the, the, the last mile piece of actually shipping is the part that I could not have done on my own, and now I can. Uh, and it gets better every day.
L190Chris: Uh, and so I think that's a really interesting thing with AI, um, that, uh, that having a project manager's mindset.
L191Nix: Yes.
L192Chris: Uh, I told somebody recently, it's only, it's only vibe coding if you don't have a plan. Like, if you don't.
L193[redacted: private, lines 193 to 211]
L212Nix: Yeah.
L213Chris: It's, what's the business use case for it? What do you want your user experience to be? Um, and because the, the, the plumbing, if it's not already completely possible to do that with AI, it cer-certainly will be within a year.
L214Nix: Yeah.
L215Chris: You know, some of the best coders in the world are not reviewing code anymore.
L216Nix: Right.
L217Chris: Um, and so a-at that point, the, the real bottleneck is imagination. It's knowing what a client wants.
L218Nix: Right.
L219Chris: And I don't hear enough of that. And that's not me blowing my own horn, but I just don't hear enough conversations online in the AI community or in the linux community anywhere about, well, I had a meeting today with a client and here's what we came up with for them. Um.
L220Nix: Right.
L221Chris: So, um, I think it's gonna be huge just for you, you talked about 20-year-olds and, and not to talk down about 20-year-olds, but, uh, rather than them, you know, cursing the fact that they're not, you know, handwriting code, they need to be focused on what does good look like? You know, what does a good project look like? I can smell a project going bad, and it doesn't matter if it's a room full of coders or if it's a room full of AI.
L222Chris: I can smell a project that has gotten out of scope. I can smell a project that's, uh, um, you know, i-is misaligned.
L223Chris: Um, that's gonna be a skill set that people are gonna, that some people are gonna have to develop pretty quickly.
L224Nix: Yeah, I agree with you. Um, two, two comments on those. When, I think it's two weeks ago, Linus Torvald said that he's using, uh, Claude code for the kernel work and he only reviews.
L225Nix: That was enough for me. And I'm.
L226Chris: Yeah.
L227Nix: And I'll take you one level deeper. When I talk to my agents a-initially about a project, I have found that when I go straight into, here's the goal, that it just goes, I'm gonna do this, and I'm gonna do this, and I'm gonna do this, and I'm gonna do that. What I, le-let's say that we wanna make a customer their own custom CRM.
L228Chris: Yep.
L229Nix: I don't, I don't start out by saying, hey, we're gonna build a customer CRM. I start out having a conversation about, and literally would go like this. What do you think the good things and the bad things about a CRM are?
L230Nix: What are the things that people and like get it talking about the concept and the ideas. And then after.
L231Chris: Absolutely.
L232Nix: An hour of doing that and having it like nailed in and drive into like the agreement that I have, then I go.
L233Chris: Yep.
L234Nix: Then I take all that and go, well, let's build one of those for this company and here's blah, blah, blah. Here's the goal. I've found that giving it the goal of the first prompt, it gets too eager and, and doesn't do its homework.
L235Chris: And I've, I've had people look at me like I was insane, but not because I'm, I am not one of these people that believe it's gonna be alive and, and things like that.
L236Nix: Right, yeah.
L237Chris: But it's built to behave that way. And so I treat it the same way that I would a coworker. I don't, I sit down and say, I had a meeting yesterday with the client.
L238Chris: Here is what their problems are.
L239Nix: Mm-hmm.
L240Chris: Here is what we talked about. Here is what they are thinking. I, I start with the people portion of it or the.
L241Nix: Right.
L242Chris: You know, the, the business portion of it. Never the technical part. I have seen people working with AI get themselves in trouble because they will sit down and say, build a CRM, make it look like this.
L243Nix: Right.
L244Chris: That's not the best use of this technology, not by a long shot. We need to start with the personalities. I the, this is my client.
L245Chris: This is what they're trying to accomplish.
L246Nix: Right.
L247Chris: You know, this is the market they're in. This is what they do for a living. This is what they're doing now.
L248Chris: We gotta back way up and treat this thing, uh, with the respect it deserves or other, you know, if you're treating it like a, a screwdriver, uh, you're, you're missing about 70% of the value.
L249Nix: That's right. Let's see here. What, um, what are the things that you're working on that, that have your attention that you feel like are making a difference?
L250Nix: And, you know, all of us have 53 projects going on. What are the things?
L251Nix: And you've sent me a ton of them, and I'm doing everything I can to stay afloat and stay on top of all of it. Uh, but I, I, I, I have been attracted to what you've been building more because there's something in the quality of what you build that's a little bit more than just you.
L252Nix: And this, I say this in a positive way. You, you seem like there's a team of people working on your stuff and not one guy in the, in a positive way.
L253Nix: Um, like.
L254Chris: Mm-hmm.
L255Nix: What's the, what are the things that, like, that you have built that you're just like, you wanna make sure everybody knows about that you're the most proud of? It's time to brag for reals.
L256Chris: No, I don't know. And I'm not building a ton of stuff publicly. I support, um, my clients are in the healthcare.
L257Nix: Uh-huh.
L258Chris: Uh, field, legal, and in manufacturing. Um, and so the agents that, and agent fleets that I build out are monitoring these systems. They're monitoring, um, you know, uh, security systems.
L259Chris: They're monitoring manufacturing systems. They're monitoring, um, systems.
L260Chris: The things that I have built and shipped that I really, uh, am excited about are things that would be very boring to the outside world, but like a refund system for a healthcare company.
L261Nix: Sure.
L262Chris: You know, a patient overpays and gets refunded. Uh, HR systems, um, uh, where companies can onboard new, uh, clients.
L263Chris: Um, a law firm that, and, and, and that's a very touchy, you know, you have to be extra careful.
L264Nix: Sure.
L265Chris: Uh, but, but onboarding clients to a law firm. Those kinds of things are the kinds of things I'm shipping every day. Uh, but not just building it and, and delivering it and walking away from it, but kind of doing this iterative process with clients where, um, after they get a taste of, you know, this is a software project that we would've never done on our own, and we would've probably never gone to market and paid for it.
L266Chris: You know, the little refund system that I described for the healthcare company, I'm guessing that would've been a real project and taken a couple of months. They're a very successful smaller company, re, you know, local company.
L267Chris: Um, they would've never spent that money. There would've always been something that, that they could have justified more.
L268Nix: Right.
L269Chris: I think that's what AI can do and what, uh, guys like me and you can do is give companies that normally, you know, they're not gonna draw the attention of a Palantir. They're not gonna draw the attention of, uh, companies that are giving away Ford deployed engineers, and, and they're not gonna spend a, a million dollars in tokens a year.
L270Nix: Right.
L271Chris: Uh, but you can give them these low-hanging fruit projects that, that were just never in the budget before, and you can, you can deliver those.
L272Nix: Right, right. That's amazing. And are you on the Omarchy train?
L273Nix: When did you discover it? And like, what, what's it done to, you know, let's talk before you knew Omarchy was, was 1X.
L274Nix: Now, don't consider all the, all the 12X stuff from AI, just. What you were doing three months ago with AI, and then how did you find Omarchy?
L275Nix: I don't know how to say it. Omarchy, whatever it is.
L276Nix: Uh.
L277Chris: Yeah.
L278Nix: How did you find it? What was the first machine you put it on? And then where are you now with it?
L279Chris: So I've always had a linux machine around, um, but the goal has always been to for it to be the, the sole driver, the daily driver, but that's just never worked out for me, uh, until this year. Uh, and so, uh, I think as recently as, um, you know, the last, you know, six months, I probably had an Ubuntu machine or a.
L280Nix: Sure.
L281Chris: Debian machine sitting around. What's changed for me is, and I think for a lot of other people, is if the AI workflow, I don't think people realize how hard it is to keep those plates spinning. I mean, you could work 24 hours a day.
L282Nix: Yes.
L283Chris: Uh, especially if you have clients to support. Um, the, the first time I heard about Omarchy was on a podcast, and it was somebody tearing it apart, uh, about how that it was AI first and agents were in the mix. And I was like, no, that's exactly what we need.
L284Nix: Right.
L285Chris: Um, and I think I started playing with it right before, uh, Omarchy 4 came out. And the reason it's such a good fit, uh, is because, you know, I'm talking to a fleet of 14 agents, um, and those are not, um, frivolously built agents. They are agents that have jobs, specific jobs for specific clients.
L286Chris: Uh, keeping them working even with automation and even with autonomy and things like that is a lot. And so having a, an operating system like Omarchy or Omarchy, however we're calling it, uh, but using tools like Herder, um, and, and things like that.
L287Chris: So you asked me what I was doing three months ago. Uh, I do a lot with Hermes agents.
L288Chris: Um, I built my fleet originally in Claude code, uh, so each agent just lived in a directory. Um, and then at some point, uh, Anthropic changed their billing model.
L289Chris: And so I was doing a lot with Cron and spinning up agents on the fly to respond to things. And they started, uh, they changed their billing model so that that wasn't allowed.
L290Nix: Wow.
L291Chris: Uh, and that's what led me to Hermes, uh, which I really like. Um, but I was clicking tabs three months ago. Click, click, click.
L292Chris: So moving back and forth between conversations. What Omarchy does well and, and, and applications like Herder, uh, is just give you a way to efficiently work with a ton of agents, give you a way to efficiently work, uh, along a ton of surfaces.
L293Chris: Um, you know, Herder especially, uh, being able to 'cause I have a, I have a server, uh, in the cloud, uh, where most of my dev work, uh, lives. I have a laptop, desktop locally, being able to work across all of those surfaces.
L294Chris: Uh, Omarchy has just made that, uh, so much easier and so much more polished.
L295Nix: Mm-hmm.
L296Chris: Uh, I still have a Windows, uh, laptop in the bag. I haven't taken it out in a couple of months probably.
L297Nix: That's okay.
L298Chris: 'Cause I haven't needed to.
L299Nix: Right. And when, when, when I get back on my Mac, I end up hitting the super button or the super key and like, ah. And then I, I'll admit, I use, I still use my Mac for Zoom calls because it's Zoom on Omarchy is a little weird.
L300Nix: I don't know if you tried it.
L301[redacted: private, lines 301 to 304]
L305Nix: What's your, um, what's your X handle so I make sure I got the right Chris Gray?
L306Chris: Uh, it's, uh, latent operator, L-A-T-E-N-T operator. And that's another example of me not building in public.
L307Chris: So, um.
L308Nix: Well, awesome. Um, so anything you'd wanna, anything I didn't ask you that you wanna get on the record? Anything that, um.
L309Chris: No.
L310Nix: Um, it's been really.
L311Chris: You're gonna do a lot.
L312Nix: Pretty, this has been amazing, man. And, uh, so the process will go from here is, uh, you know, it'll, it'll be, it'll go up there. Is there anything on here that you said that you don't want to be published?
L313Chris: No, I didn't speak about any current clients by name. That would be something that I take very seriously.
L314[redacted: private, lines 314 to 326]