Section Insights
Evolution of Programming Setup
How has the programming setup changed with the advent of agentic engineering?
The transition from traditional programming to agentic engineering necessitates a different toolset and approach, moving from single-threaded to parallel processing. The speaker reflects on their long-term use of TextMate and the shift to using multiple agents, which requires a more complex setup.
- Agentic engineering requires a different mindset and tools compared to traditional programming.
- The speaker emphasizes the importance of parallel processing with agents.
- Traditional immersion in a single problem is replaced by managing multiple tasks with agents.
Adapting to Multiple Agents
What changes in setup are necessary for managing multiple agents?
To effectively manage multiple agents, the speaker transitioned to using tmux for terminal management and later adopted Herder for notifications and tracking agent status. This setup allows for a more fluid workflow, enabling constant decision-making and task management.
- Using tmux and Herder enhances productivity when working with multiple agents.
- The speaker appreciates the terminal environment for its efficiency and aesthetics.
- Effective agent management requires tools that facilitate communication and task tracking.
Networking and Remote Access
How has networking technology impacted the speaker's programming capabilities?
The speaker discusses the advantages of using WireGuard networks and Tailscale, which simplify remote access to multiple computers. This technology allows for seamless integration of additional computing resources, enhancing the ability to run multiple agents simultaneously.
- WireGuard networks and Tailscale reduce friction in accessing remote computing resources.
- The speaker maximized their processing power by connecting multiple mini PCs.
- Remote access technology has transformed the speaker's ability to manage and run agents.
Productivity Metrics in Programming
How does the speaker view productivity metrics in programming?
The speaker critiques the use of lines of code as a productivity metric, arguing that it does not accurately reflect the quality or impact of work. Despite this, they note a significant increase in output due to the new setup, emphasizing the importance of context in evaluating programming work.
- Lines of code is an inadequate measure of programming productivity.
- The speaker has experienced a dramatic increase in output with their new setup.
- Contextual understanding is crucial when reviewing code changes and outputs.
Challenges with Mac Setup
What challenges does the speaker face when using a Mac for programming?
The speaker highlights the limitations of Mac systems, particularly in automating setup and configuration processes. They express frustration with the GUI-centric nature of Mac applications, contrasting it with the more flexible and scriptable environment of Linux.
- Mac systems present challenges in automation and configuration compared to Linux.
- The speaker values the flexibility and ease of setup that Linux provides.
- Frustrations with Mac's GUI limitations highlight the speaker's preference for command-line interfaces.
Transcript
0:03 I got to ask you about how is your setup programming setup changed? So, keyboard, voice, what's the IDE? >> It's crazy to think about now, but yeah, I used TextMate for almost 20 years. I used TextMate starting in 2005, I think. I helped get the first version out, and then I just wasn't interested. I wasn't in the market for an alternative. And it wasn't until the switch to Linux that I was forced out of my habitat. And now, with the switch to What are we calling it? Agentic engineering? I hate that term.
0:44 We got to come up with something that sounds as plain as programming, but encapsulates the fact that it's with agents. But, >> I still think it should be called programming. >> let's just call it programming. >> Yeah. >> The programming with agents requires a different tool set. It really does. And the main change here is that you're going from single-thread programming in your head to parallel processing. When I was writing code, chiseling it by hand in TextMate or even Neovim not that long ago, I would just focus on one problem at the time. And I would methodically work my way through it, and that was actually the portal to flow.
1:23 The portal to flow was deep immersion into a single problem, see it through to the end. >> Mhm. >> That's not how it works with agents. In part, because the agents are at once both too fast and too slow. They don't give you an immediate reply on something that you asked them to do that's the same as typing on a keyboard. So, you have to let the agent cook >> Mhm. >> for a bit. And therefore, you realize, well, if I just sit around waiting for them, first of all, that doesn't feel productive. Even if the agent, just one of them can be highly productive, it does not feel productive, it doesn't feel good. It feels actually like you're a little bit useless.
2:06 >> Mhm. >> And maybe I had a moment when the first agentic moment was there and we I was running mostly one agent at a time where I felt like I don't know about this. But you can solve a lot of hard problems by just throwing more resources at it. This is the whole scaling law of AI itself, right? That if you paralyze these things and you're not running one agent, but you're running a handful, you can feel like you're in a flow state because you're constantly doing programming work in the sense that you're making decisions and you're helping either unblock an agent because it has a question about which direction to take or you're ready for a new task. And to do that, you need a different setup. I started first doing it in tmux and just having separate panes and having separate splits. Basically, a terminal with tabs is a good way to think about you open a bunch of tabs. I think most humans know exactly how that works. They don't work in just one tab.
3:06 >> So, still sticking to the terminal CLI. >> Absolutely. >> So, you're not using cloud code app or the codex app or the >> I love the fact that this agent revolution was kicked off in the terminal because I was already a huge fan of TUI's, terminal user interfaces, and the terminal in general. That feels like a a really nice place to be. It's a beautiful place to be. The modern terminal is just a good looking place to work. So, >> >> I like that.
3:35 And and then this fact of having multiple agents, especially once it's not just multiple agents running on your own machine, but you start running multiple machines. Now, tmux alone is not enough to keep track of it. And that's why as of late I've switched to this thing called Herder. And Herder's essentially team mugs plus agent notifications. So, whenever your agent is done or needs something for you, it goes ding.
4:06 A little a little bell telling you it's ready for its human Is it Is it master or servant? I'm not quite sure always. >> >> But, it is ready for a decision. And it also keeps track of these Is it is it working or not? So, I have this Herder set up. I have multiple Herder setups actually running on individual machines. I went on this crazy phase just about a month ago realizing that doing this work on a single machine is not fast enough. It's like I've discovered multi-core programming, but I only have two cores. I'm like, what if I had 16 cores? What if I had 32 cores?
4:42 What if I had 64 cores? So, I instantly went out and I bought these amazing KVMs called gli.net comets. >> Mhm. >> And what they do is it's this little box. You plug in HDMI, you plug in USB, and connect it to the computer. It's like a KVM. So, KVM is a a remote way of controlling computer. But, what's special about this is just how easy it was. You connect this thing in, you go to a webpage, log in once at one password. Now, this thing can hop on your Telnet.
5:15 >> Mhm. >> This has been the other revolution of the last year for me. It's discovering these WireGuard networks. Tailscale is essentially turning all the computers you have into a local network wherever you are. >> Mhm. >> Like right now on my phone, I have direct access to all the computers in my Malibu office. I also have access to all my computers in my Copenhagen office. And I can treat them as though I sat right next to them without having to punch holes in a firewall or set up complicated VPNs and what this does is it just decreases the friction it takes to get new compute online. So, as soon as I discovered this, I looked in my closet and I realized I had a bunch of mini PCs from prior experiments. I just said, "What if I just connected all of them?"
6:07 And I just connected four of the computers in a closet. They all had their little comment and suddenly I could run agents on more computers at the same time and I could control them all with Herder. And it did get to a point where I maxed out my own processing power. That I think at about What do I want to put it? About four to five machines running I don't know, three agents. I have about 16 threads. That's what I can run. And the faster the agents run, of course, the fewer threads I can run. But at the current pace I can run about 16 threads.
6:43 >> >> At at full acceleration. And that's part of why I've gone from just being excited about the agent moment to being delirious. Because I think I should say some of it is a little assaulting and again, we're on we're on the dial-up bandwidth wise with our computer. I mean, it's actually hilarious. So, you think of I'm sitting here a year ago, right? I'm chiseling my code. I'm writing my lines. And over the course of like an hour, I will have written one beautiful controller, one beautiful model. Like that's one file.
7:19 And I really worked on it, right? So, maybe there's 60 lines left. So, maybe my bandwidth at this moment is like 30 lines an hour. I think that's even high. Maybe it's 20 lines an hour. Now I'm running 16 threads. I'm producing at sometimes hundreds of lines of hour or or hundreds of lines of code per hour. Now, let me pause myself there for a hot moment. I hate that metric, right? Like lines of code is a stupid metric in general to measure these things, and I think it's right from someone in the programming community to ridicule the AI psychosis when everyone talks about the number of lines of code they're writing, and then you ask them, "What did you build?" And then like, "Well, I didn't And then no good answer comes out, right? It's just a nice shorthand >> It is a nice shorthand. It is a representation and encapsulation of >> how much output that's coming out. Now, whether that output is good or bad is not a referendum on that. But, I was producing and producing so much more now, right? And therefore, I'm able to keep all these threads going. So, that's the setup. It's still Neovim, but at this point we're not writing a lot of code. So, I'm using Neovim as a project browser, and then as a way to kick off lazy get to see the change log for what's there. And even that, I'd say if GitHub was a little faster at showing you your pull request, the web is probably actually a nicer place to do that.
8:45 I don't know. There's also this other tool I've been playing with a bit called hunk, which just produces diffs in a really nice way. So, you could look at that, too. But, I find that when I look at hunk, I only see the change set. And when I'm reviewing output from an agent, I often want to see the surrounding context. Oh, yeah, so it changed this file, but actually, what do we have in this other file that wasn't touched, but maybe should have been touched? So, that's why I still like Neovim as a way of doing it.
9:11 but it's all happening, by the way, of course, in Emacs. So, it's all happening on Linux. And this was the other major breakthrough with agents. Agents love the Unix philosophy. It loves individual tools that it can invoke through the command line. And there is no operating system on Earth of the majors. I'm counting three here: Mac, Windows, Linux, that works as well with that mechanism as Linux. Everything is in Linux it's either a config file or a CLI tool.
9:46 Now, that was its main drawback 5 minutes ago. This was the reason people didn't like Linux. It's like it's all config files and CLI tools. >> Mhm. >> What great irony that the universe has played it upon us that now the drawbacks of Linux 5 minutes ago are now it's major selling points. This is one of the things I thought I was stuck for a weekend with a Mac 4 months ago. I thought I had a computer the place I was going that was a Linux machine so I didn't bring my laptop and I I found out when I arrived I only had a Mac Mini.
10:24 So, I was going to make the best of a bad situation here and set up my Mac in some of the ways I've been thinking about with with Linux. And you can actually do a lot now. Homebrew has gotten really quite good. Homebrew is the missing package manager for the Mac and no one has done more to move the Mac forward in terms of ease of setup. But, still something as simple as Raycast, I don't know if you've used that. That's the >> Yep.
10:53 >> There's no config file that you can just access. You have to go into the GUI, export a file, then take that file, I don't know, in your freaking backpack on a USB key. >> Mhm. >> And then you can import it somewhere else. You cannot automate the entire setup of your machine. You can't automate at all the configuration of Mac's default key bindings. That has to be a manual process where you're clicking with a mouse like a caveman to set up your machine.
11:22 >> I mean, there's ways around it. >> Not good ones. I looked. I tried hard. >> So, here's here's my been my journey. Obviously, I'm a Linux person, but because of Adobe Premiere, Adobe products, I'm also a Windows person. So, often times I would use WSL, Windows Subsystem for Linux. So, Linux inside Windows, which is in the agentic era, is not a good kind of Linux cuz it's just >> It's a sandbox. >> It's a sandbox and you want the Linux to be unleashed to be able to do everything.
11:50 >> Correct. >> And so, I have now woken up to things like Grey Cat and have you know, I'm a keyboard person. Where is the config file? Can I Can I Can I basically do everything where an agent can can set everything up for me and save the configuration so I can replicate across systems and everything is automated. And so, I had to ask a lot of those questions and a lot of them are missing. >> have good answers.
12:14 >> But often times you could actually just build the app yourself. >> But they're a hacks. >> They're hacks. >> They're hacks. You don't have to live like this, Lex.
Summary
- Transitioned from using TextMate for nearly 20 years to a more agent-based programming setup after switching to Linux.
- Emphasizes the shift from single-threaded to parallel processing, requiring different tools and workflows.
- Utilizes tmux for managing multiple terminal sessions and has adopted Herder for agent notifications and management.
- Explores the benefits of using multiple machines and KVMs to enhance processing power and efficiency.
- Advocates for the advantages of Linux in the context of agent-based programming due to its command-line tools and configuration files.
- Critiques the limitations of Mac and Windows in automating setups compared to Linux.
- Highlights the importance of context when reviewing code changes generated by agents, preferring Neovim for project browsing.
- Discusses the irony of Linux's previous drawbacks becoming its strengths in the current programming landscape.
Questions Answered
How has the programming setup changed with the advent of agentic engineering?
The transition from traditional programming to agentic engineering necessitates a different toolset and approach, moving from single-threaded to parallel processing. The speaker reflects on their long-term use of TextMate and the shift to using multiple agents, which requires a more complex setup.
What changes in setup are necessary for managing multiple agents?
To effectively manage multiple agents, the speaker transitioned to using tmux for terminal management and later adopted Herder for notifications and tracking agent status. This setup allows for a more fluid workflow, enabling constant decision-making and task management.
How has networking technology impacted the speaker's programming capabilities?
The speaker discusses the advantages of using WireGuard networks and Tailscale, which simplify remote access to multiple computers. This technology allows for seamless integration of additional computing resources, enhancing the ability to run multiple agents simultaneously.
How does the speaker view productivity metrics in programming?
The speaker critiques the use of lines of code as a productivity metric, arguing that it does not accurately reflect the quality or impact of work. Despite this, they note a significant increase in output due to the new setup, emphasizing the importance of context in evaluating programming work.
What challenges does the speaker face when using a Mac for programming?
The speaker highlights the limitations of Mac systems, particularly in automating setup and configuration processes. They express frustration with the GUI-centric nature of Mac applications, contrasting it with the more flexible and scriptable environment of Linux.