Context engineering with Dex Horthy

What happens when you let AI agent ship code for months, and no developer reads a single line. Today's guest tried exactly that. He built the lights off of Software Factory, and four months later he had no choice but to shut it down as things just stop working. Dexworth is a founder of Fumilier, and the person of coin-determined context engineering days before Andrew Carperthi and Toulby-Loozka made it famous. He spent the last two years talking to hundreds of AI engineers about what actually works when you build with LMs and is testing the most recent ideas with his own team. In today's conversation

we discuss context engineering, what it is and the physics of context windows including what the dump zone is, lube engineering. From the Ralphwogum technique to the slow loop, Dex's team runs every night to wake up to co-clean up PRs. The rise of Software Factories, from a NATO conference in 1968 through DevOps to today's agent factories. Spectral development and why specs always drift from the code itself, and many more. If you want to understand increasingly important concepts like concept engineering and harness engineering or want to know how far you can push the let agents build everything idea from someone who pushed it further than almost anyone, then this episode is for you. This episode was presented by Intecisys. If you work with agents,

your job is no longer just writing code, it's specifying and testing it. Intecisys is the most effective method of verifying agent code today. Today's episode was brought to you by Buildkite, the CI organization platform trusted by OpenAI and Tropic, cursor and video, Uber, Canva and more. Today we're talking about pushing the right context into models so that they write better code. Right after that starts working, your agents will write more code, a lot more. Trusting that code avalanche is where many teams face a challenge today. Every

change that an agent makes still has to be built, tested, and proven safe before it ships. Work on my machine is not enough, so you obviously need CI. But what agents are pushing five, ten or fifty times a commit value into your pipelines faster CI or nurse won't save you. Shaving three seconds off a single build is meaningless when a QS-100 plus job steep. What you really want is a CI system that gets faster as the volume grows, and CI that offers instant parallelization to give you unlimited concurrency and to intelligently route changes at runtime.

This is what Buildkite does and why Global Software leaders continue to rely on it. The same architecture that observed a scale of Shopify an Uber a decade ago, now runs about 1.4 billion job minutes a week across cursor, meta-redded, and snowflake. While the rest of the CI world are cracking under the weight of re-architecting their platform, Buildkite continues to reliably grow. Agents running on your infrastructure are Buildkite, any cloud, any chip, your secrets, your scale. Every artifact and log is captured, so when something fails, either you or your agents have immediate insight for why.

As you're entering the context you'll get to your agents, think about how you'll verify what they hand back. If your system is buckling under the increased volume, head to buildkite.com slash pragmatic. Three-day, all-access trial, no credit card, and an actual human engineer on standby. His name is Ola, and he's very helpful. So, Dex, welcome to the podcast.

Super stoked to be here, dude. Before we get into some of the context engineering and some of the more spicy stuff as well, how did you get into tech? How did you follow up with computers? Oh, man. So, I was doing undergrad as a physics major, and I realized that I didn't like academia, and there's basically two or three paths out of physics is basically you go get a PhD, or you go into finance, or you go do programming. At that time, this was 2012, 2011, when it was in

the middle of undergrad deciding what to do. I had done an internship when I was in high school. I was working with NASA researchers to eject propulsion lab in California. They had just gotten this really high fidelity, the most fine-grained dataset of altitudes, the height, top graphical map of the South Pole of the Moon. The South Pole of the Moon is really interesting because some of the craters there are so deep because of the angle it has. It got hit by meteor storms like no other

part of the Moon. There's very deep craters that have never seen sunlight. So, there's frozen liquid water in there from the formation of the Moon. So, science, we're really interested in getting down there and exploring. So, we have this really fine-grained map. It's like, okay, cool, let's build software so

that I have point A to point B. I know the limitations of my rover. Maxine climbed up is this. Maxine climbed down is that. I was 17. I never cracked a CS textbook. So, I wrote a really naive, bad version

of Dijkstra's algorithm for pathfinding. I was in college. I don't know if I want to do the academics thing, but I really enjoyed programming back in the day. So, I decided to go like half of a CS minor and then started working on an API platform team at a software company in Chicago and in the social rush. Basically, never went back. Where did you pick up the parts of the trade?

Because early on, your first job, that's not really common. You were doing platform engineering back in more than a decade ago. From that point, it took me about two or three months to notice that the most valuable work that was being done in the company was being done by the first couple engineers who know everything and understand where everything was. You spent a day on a support ticket from a customer and they solve it in five minutes, but you have to solve it so you learn and whatever. I realized the most valuable people in the company were the people that were building the developer platform. CICD, sandbox environments, preview stuff. So, that was my first step into

the journey and I've basically been obsessed with software factories since that three or six months into my first job. We talk about software factories now, but you're talking about software factories back then. So, you're starting to already think that this is how we can produce better software inside. Is it pre-AI world, right? Well, and I'm always surprised. There's

a huge class of developers. I don't want to work on CICD. I hate CICD. I'm like, really? Because building the thing that builds the thing and building the thing that builds the thing is like, as software engineers, we're lazy. We want to do the most high leverage thing that makes our job easier. So,

how do we build a thing that helps us build a thing that helps us move faster than that's the best use of my time as a lazy engineer? And then you went to another startup as Aspernation? Asperation? Yeah, Asperation also platform engineering. Yeah, I was brought in and then like three months into the job, the VP of engineering who hired me quit or got fired. I don't know. There's

some drama about it. I probably shouldn't talk about it. And then I was there for about a year, and was kind of like acting CTO for a while, like hired a couple people, helped hire the new VP of engineering. But I was out of there. I don't think I'll ever do consumer again. I think I'm actually

a B2B guy. Good to know. And then we went to replicate it. We spent like a good like solid like four years and went from engineer to product manager. Yeah, I was doing core engineering for like two years that we were building a container orchestrator. Like before Kubernetes, before Docker Swarm was really

a thing, we built our own orchestrator. The founders had this vision that like, oh, Docker is going to make it much easier to ship on-prem software. I want to say on-prem, I don't mean literally like a rack in a colo. It's more like, hey, look, bring the app to where the data is rather than sending the data up to some cloud vendor. And Docker makes it much easier to package up apps and move them around. And so they had this thesis that like basically you could build a platform that the experience

that you get when you use GitHub enterprise, which is like you install it and has this admin panel, but then you just get GitHub running in your data center and your code never has to leave your data center. Suddenly you could build a generic SaaS where everybody could have that. But in two years of engineering there, and then our head of sales, we point our ways with our head of sales. And honestly, I was having a lot of arguments about the software factory with our CTO and it's kind of like almost like a too many cooks in the kitchen kind of thing. I'm sure many listeners have had this experience of like, well, yeah, I know I have these tickets to build, but like CI sucks. I got to fix

CI because it's too slow, or it's like there's too many different builds. There's always breaking. I'm going to fix that. And then I'm going to do the end is just like, decks, and you just stop fixing the build pipeline and like do the tickets I gave you. I'm sure you've had this experience perhaps.

Yeah. And then was this what led you to either forward-employed engineering? Yeah. So I really loved our customers. Our customers are HashiCorp, Datastax, Poppin, all these really cool engineering brands, Travis CI, Circle CI, I was like, yeah, I actually love working with our customers. Our

customers are awesome. And it was a great way to get in the trenches. A lot of really good engineers. We're solving the hardest problem at the company, which is like, how do we take this three to five year old SaaS platform and package it all up so that someone who knows nothing about our architecture can run it reliably in their own AWS VPC in their own on-prem data center, whatever it was. And so I spent I was our first kind of customer facing engineer, and it was about three months.

We closed with like every company customer that was kind of in the pipeline, but wasn't moving sales-wise. And we closed like 12 deals in three months and the CEO was like, holy crap, Dax, like the investors are taking my calls again. Like, I don't, I know you want to get back to coding, but like, I need you to go hire three people and like build this team out because I think you might have been like born for this. Wow. Yeah. So I did that for about four years,

built that over to like 25 people and ends are happened and got a lot smaller and we kind of realized like, Hey, we have a product that's like pretty good. And we've been solving what lots of early startups do is like, okay, there's some usability issues. We'll throw a bunch of smart people thrown in the trenches with our customers, great for sales, great for retention, all this stuff. And I was like, Oh, we actually like the margins on that aren't good enough. And so we basically were like, cool, we actually just need to make the product way more usable, do a more PLG-shaped thing, make a little more self-service. You don't need an expert to teach you how to use it. And I was

like, well, that's the most important thing that I want to go be a product manager because I have tons of opinions. I've now spent four years in the trenches with our customers. I have a laundry list of roadmap things that I think would make the product way easier to use and adopt and implement and deploy. And now you went to the full argument, where the dark side? Exactly. I did. I was like,

this is going to kill my street credit, isn't it? But I was really glad, you know, I think a lot of engineers are afraid that if they go do a customer facing thing, they lose all their credibility. And like, yes, I wasn't coding for 10 hours a day. I was coding for like three or four hours on a Saturday for fun, not, but I mean, we were helping people build YAML. We were building CLIs. We owned

a lot of the tooling that customers use, but it was like the last mile delivery side of it, not the core platform. And like a more personal note, I had spent the last like most of my 20s feeling like, okay, a little bit introverted, a little bit like socially awkward, what a lot of engineers I'm sure experience. And I had talked to my uncle's a music producer. So he used to work at like Randy Newman and a bunch of like really famous musicians. Oh, wow. Yeah. Sky Mitchell from

and he, I was sitting with dinner with him at some point when I was, I think it was when I was still in undergrad, but he gave me this lecture. He's basically like, if you want to be really good at something, you have to make it the only thing you do. The guy playing guitar nights and weekends trying to get his band off the ground will probably never achieve greatness. The people who become great are the people who basically make it like, if I don't play guitar, I don't eat and you go and you sit on the street all day and you play for 14 hours a day or whatever it is, that's the only way to become great. So I said, okay, instead of trying to like read self-hob books about how to be less introverted and less socially awkward, like, what have I just made it my freaking job to just talk to people and make friends and like help people and solve their problems. And I think it worked out.

I recommend it. I think everyone should spend a year or two at least doing something really like customer-facing. Did you do this because you felt that I was holding you back being introverted or like, what what what and I know you got the motivation from the whole musician motivation? I get it on one part, but what was it that you said like, is a customer-facing thing that I'm gonna be doing? Because really, you were pretty great at like writing code by that point. You could argue you were doing it

night and day. So where did you find that like, I actually, I think like customer-facing or like getting this introvert off of me, did you feel that I was holding you back or you just wanted to be good at it? It was just kind of a thing that was like interfering with my like general life satisfaction. And it's also like, I'm not a very type A person. I'm very disorganized. I don't know people call it like,

okay, I'm like ADHD now. That's why I can run 30 collads in parallel or whatever it is. But it's like, I was really bad at email and calendars and spreadsheets. It's like, didn't care about these, didn't understand them. And so like, another side effect of this was like, it just forced me to be organized and keep a lot of things going. And so like, I don't know, there's like weird benefits you get from like stepping

outside your comfort zone and learning like industrial disciplines that are separate from what you've been doing. And so the opportunity presented itself, I was like, oh, I like, I'll try this for a little bit. Started going really well. I'm like, cool, let's see, let's see how far this thread goes. And then afterwards, you're now in your second startup, you become a founder and you also got involved in AI pretty early as it was, even before it was so obvious that it would change how it would change how we developed software, right? Well, I would say I was later than I could have been.

Because we started the company, me and Abunya in Chicago started a company in the data engineering space in about 2020, November 20th, we decided in like August of 2021, metal edicts, metal edicts, technically still the same company as human layer, we just like pivoted the mission. But yeah, the advice I got from every angel investor that, you know, people who just knew CTOs had worked for before and stuff, they were just like, look, hitting a lot of headswinds. I don't know if you know, like the whole DBT data engineering, five train that whole arc where it was like this huge party and tons of investor money going into all these different companies. And then within by like 2021, 2022, there was kind of a zerp thing and just this general realization that the TAM for those sorts of tools is not as a general market. Yes, the total addressable market for those sort of tools was not as quite as big as we all thought it was. So it was a hard place to raise money,

it was a hard place to get customers. And then I met you while you were at Fumen layer in a CFS event, we actually talked and we chatted afterwards. But by that, this was about a year ago, you were already you start to have some really strong opinions on using AI. And one of them was this now famous 12 factor agents manifesto. Are we calling it manifesto now? I'm calling it a

manifesto. It's a manifesto. I'm calling it. Let's talk about this. This was 12 engineering principles to build reliable production ready apps. How did you come up with this? And maybe we can also talk

about some of them. Yeah, so I'll kind of like go to like around August, the co-founder was working with the kind of burned out and left and it was very, very beautiful. And I decided to start messing with AI stuff. And I was building AI agents and what was really info great them was like the Lang chain, the crew AI, these like agent frameworks. And it seemed like there was a ton of you going to Korea, I just scored this 10,000 people. It's like, okay, this feels like the right shape. And

there's clearly this eco, you go in every single one of those projects, they have a chroma DB plugin, they have like a composio plugin. There's like, clearly, like this is the shared interface that everybody is building for. And so, okay, what's missing from all of this, the agents can call tools. But it's really hard to like control which tools they call. And if it's a chatbot, obviously you can show approved and I in the UI of your application. But I kind of was obsessed

with what I would call like outer loop agents or proactive agents that would run in the background and get triggered by advanced, I mean, open clause, basically like the biggest manifestation of this, if like you have a heartbeat, it wakes up, it sees if there's any work to do it, it tries to do stuff. And I thought was like, I'm not going to trust that agent to do anything meaningful. If I can't get like a Slack message or an iMessage or something when it wants to do something and kind of guarantee deterministically that I can approve or deny that or deny it with feedbacks, actually know do it like this. So we played in that space for a while and talked to a lot of founders and founding engineers and builders. We came into YC in the fall of 2024 with this idea. We're building out

this API platform and it was sort of like pager duty, but like it wasn't who's on call to fix the servers. It was like who's on call to this like routing mechanism for like who needs to approve this agent and can they like escalate it or delegate it or defer it all this stuff. And we built it for the ecosystem. Great. I like saying, fi, there's so many grip tape. There was so many in that in that

time. And then I talked to tons of AI engineers who were actually building really interesting things and like actually making money doing six figure contracts, shipping AI to the enterprise. And all of them had tried that stuff for like a month or two. And then they had thrown it out and they were just writing all the API calls by hand. And they were building more things that look more like pipelines and workflows than these sort of like hands off call tools and a loop kind of thing.

And so I talked to 100 people and I spent a lot of time, a lot of time hanging out with one of my best friends, Vibov from Boundary. So they built a programming, they built like just like proto buffs for AI thing. And they're about to launch their like full fat like programming language Turing complete thing. But he had this way of thinking about agents and building with models and building with inference where it was a lot more about understanding what structured output really is under the hood. And every single step in your AI workflow is just tokens in tokens out.

And your job as an engineer is figure out, okay, what tokens do I need to put in to maximize the chance that the tokens out are going to be good and kind of distilled all these ideas and do about 12 principles and wrote about it on GitHub posted just like this like 12 page GitHub repo thrown on Hacker News got like five times on the front page for like two days. And it really resonated with a lot of people. Yeah, so I'll just quickly read the 12 principles. And then let's talk about like one or two that resonates in 12 are natural language, a tool calls, own your prompts, own your context window tools are just structured outputs, unified execution state and business state, launch, pause resume with simple APIs, contact humans with tool calls, own your control flow, compact errors into context window, small focus agents, trigger from anywhere, meet user where they are, make your agent to stateless reducer. The stateless, yeah, the stateless reducer one was a little actually someone hit me up on Twitter and corrected me. It's actually a transducer because there's

technically multiple steps in the workflow. But there we go. But this one, this was a year ago, so like, which is like forever in and how the tooling is evolving, which ones still stick with you wherever like, right, these were good that that still seem to hold off. Yeah, I think I meant most of March writing it, published this in April, and then Swix have you up from AI.engineer and he said, hey, can you come, you want to come talk about this? So I get this talk to all factor agents in June 6th, I think, and small room, maybe like it was packed, but it was like maybe 100 people.

That was the year day I engineer, where like the lower physically, like on the second basement floor was all the super corporate stuff and you go up a level, it's a little bit more and then like on the top floor is all the like weird cutting edge, like startup stuff that like you probably shouldn't care about yet kind of thing. So we're up there on the top of this like weird way of thinking about agents. And then about a week later, two weeks later, Toby Lucky from Shopify says, I really like this idea of like context engineering. And I'm like, I wrote about this two months ago. This is great. Toby gets it. And then a week later, Andrei Karpati is like,

well, I really like, I think what we should think about is not prompt engineering, but context engineering is like, yes, that's my two anyways. I don't know. If you ask Gemini, it depends what day it is, they will tell you either me or Toby or Andrei came up with context engineering. You can't really own a word like I don't know, and remember to invent the word prompt engineering, but of all the factors factor three of owner context window and basically the only way you can, whether it's agentic or a single step of pipeline, the only way you can impact the quality of your output from AI is by caring a lot about what the inputs and crafting them. Let's talk about context engineering, which I am going to credit you that you coined it. I did some research and I think you were earlier

but a few days. So there you go. You coined it. We're adding to that. We're adding to SEO juice. We'll

have it in transcript, DEX, coin, context engineering. Well, and like a asterisk on that is basically context engineering from talking to these hundred engineers and founders. I just kind of like, what was the same about what they were all doing and I put a name on it? So like, I didn't invent doing it. I was just like, I think there's this thing and like vocabulary and names are really important and having like clean ways to talk about the problem, especially when like a lot of the content about AI right now is so much hype and jargon that is like meaningless. I was like, okay,

I think there's a word here that is useful to builders that explains how they should be thinking about building their software. So what is context engineering? It's kind of like de-abstracting a lot of de-abstracting that have been layered on top. So you have a rag, you have memory, you have agentic history, you have structured output, you have all these things that are like different ideas in the frame of agentic programming. At the end of the day, they're all like different ways to pass tokens into a model and ask it to produce usually some structured output. And

understanding that is a lot more powerful than trying to learn memory and trying to take some agent framework off of a shelf and some memory framework off the shelf. I mean, those are these things are all really good. If you want to get to like 80%, you want to get a really good demo. But when you have to go from 80% to 95% to 99%, you need to go down a level and think about what's everything we're putting into the context window, what order is it going in depending on which model we're doing and all of this stuff matters. You have all of these levers that you can pull. And it just felt

like the right abstraction for thinking about how do I get AI to do the thing I want as accurately as possible. Why is context engineering started to become more talked about? It was about a year ago. Was it a bit of a bit of the context window that we could pass onto LMM, pretty much the district expand or did we just start to realize that we can do a lot more by passing on from the easiest one is of course system prompts. But of course, whenever you build an LMM behind the scenes, you will pass additional context as well, not just a prompt of the user, you will add a bunch of so that's like a dirty secret of any LMM. But why do you think the focus is moving on to like,

right, content is important? I think it always was important. I think what had to happen is a ton of smart people. Again, like all these builders I talked to a ton of smart people had to like focus really hard on producing like, I want to make software that I can sell, I want to make something that's accurate enough that I'm proud of and I can sell to an enterprise and they're going to be happy with it. And there's just like the easiest way to get to really high quality applications is by thinking at that token level, thinking about a string of different LMM calls, like rather than just tools in the loop and it's kind of open ended and very flexible, but not that reliable.

Thinking of agents as as workflows, as pipelines, as some mix between maybe a couple tools in a loop versus just, Hey, I have my tools and I have my model and I have my system prompt. And these are the only levers I have. And it's actually no, you have way more levers going to take more work and you're going to have to like understand the LMM with a deeper intuition. But it was a thing that we always needed and it just took time for people to build with this technology to figure out that like, this is the layer of abstraction that allows you to break through the quality ceiling. And how are costs and context engineering connected? Yeah, I was talking about this with someone this

morning about like when you're working on LMs, one of the things I like to say is kind of like, make it run, make it right, make it fast. See if the world's best LMM at the time, I think we did a podcast episode at the time, it was like, Oh, three, see of all three can solve your problem. And then give it to people and see if they want that. And then if people want it and use it a lot, then go do a bunch of context engineering, because your engineering time is always the bottleneck. Like humans trying to figure out and solve problems and build e-vowls and improve and try different dimensions or set up JAPA or whatever it is, is always going to be more expensive than just using a smarter model until you have millions of requests a day. And then it's like, okay, we're going to do a bunch

of context engineering, break this up into three calls and get it to work on GPT for O. And then we're going to take two of those and make those two work on GPT for O and using old model names. So the point is like, for a certain task in your workflow, can you get GPT OSS 120B, which is like one one thousandth of the cost of opus, can you get it to solve parts of the problem so that the are just the things that you really need that level of intelligence? But you shouldn't go build all of that over engineering until you prove that you need it, that it's valuable, that it's like, okay, this is now, I'm going to be like, gold rat. And like, what is the, he had this book, the goal, right? It was about how to model your factory. I'm sure we'll get to that. We talk about

software factories. It was like, what is the bottleneck in your system? And one day, it will be latency and cost, but it's probably not that when you first start out. And context engineering is how you move from the, you add human effort to the equation to improve the efficiency, the speed, the cost efficiency of your system. Interesting. And then one thing that came up more recently and a lot

later, recently, it's hardness engineering. What is hardness engineering? So I made a post in like October, I think about it, maybe November of like, hey, there's this new thing that I see is like, I'm calling it harness engineering. My definition that I had at the time is not what actually this guy, Viv, who's at Langchain now does a lot of really good writing on agents and how to think about harnessing. He had written something called harness engineering like a couple weeks before me, but I hadn't read it at that point. And my take was basically like, okay,

when you build an agent, you use our use context engineering. When you use an agent, because we get this talk in August of 2025 about like, how to apply context engineering to how you use coding agents. And that kind of evolved into this idea of like, how do you take a harness like cloud code, how do you engineer against the integration points of that harness? So commands, MCPs, skills, how you organize your code base, how do you kind of optimize the environment that the coding agent runs in to like get the best results, the same way with context, how do you optimize the inputs to every single problem? Well, harness engineering is like, how I raise the floor so that every single turn of this thing, the results are as good as possible. And the term got super blurry,

and some people think harness engineering means building a harness and some people think harness engineering means building around a harness. I actually like what Martin Fowler came up with, as usual, he's very good at naming things. And he kind of defined the, you have the LM, and then you have the inner harness, which is like the thing that the tool definitions and the integration points that like say like a cloud code or a codecs or a amp actually exposes that's your inner harness, and then you have the outer harness, which is the stuff that you the human do to customize that for your specific needs, your code base, your languages, etc. That's the best definition I think we have for harness engineering. It's interesting how naming is still so important, isn't it? Well, it's like

as soon as you name anything, people are most people are, I'm actually surprised that context engineering still means the same thing to most people that it did a year ago, and that it's even still relevant. Like, that's honestly the craziest thing to me is like, you wrote, how many things are written about AI 15 months ago still matter or are still interesting, or are still like, have good advice baked into them stuff changes a lot. I think context engineering has been so long lived because it's it's grounded in the fundamentals of how transformer attention works. And until we have post transformer models or linear attention or whatever it is, whatever it is, when that's going to happen, context engineering will be interesting and important to anyone built in AI. Can we talk about the physics of context? You had a tweet. This one, the context reality check, this is a graph of

as you get to 1 million context, just the quality just drops, it goes down. What do we need to know about like the context? Again, we now have models that do have a 1 million context window, maybe we'll have even longer ones, but when you start to put in more stuff into the context, it starts to become a less efficient. Like, what do we know so far in terms of from the practical perspective of like someone who is using the context window to add a bunch of stuff, made that the MCP, made that the tools, made that the scales, made that be all these things? Yeah, I mean, so the longer context windows are good. You can talk to it for longer, like they're

doing a good job. But at the end of the day, like, especially when you had like, opus, it was like opus 4.5 and then opus 4.5, one mill or 4.6 and 4.6, one mill, you're not actually getting a like smarter model. Like the intelligence of the model is what drives its ability to attend to all of the tokens in the context window to figure out on the next turn which parts of this 100k or 200k context window are the most relevant to making the decision of like, what is the next tool we call and doing that over and over again in a loop. So there was some study that came out in 2025, which found that and again, these are old models that like inflate your numbers, but it was like frontier LMs can follow about 150 to 250 instructions before it starts to drop off their ability to follow all the instructions just like drops off pretty quickly. And I think Lori Voss that arise, I haven't actually looked at the data that they did to study with like the next generation models a year later, and it looks like it's like much better than number of instructions you can get in. But in any case,

you have like, I split context engineering into like two categories, you have like the most people think about like the information budget of like, okay, I can do rag and I can pull out chunks of this document rather than putting the entire book into my context window, I can just go grab the pages that matter. But it's also your instruction budget is like, if you give the model too many instructions and especially too many conflicting instructions and that's in your initial prompt and also like, if you have a conversation, you start going down a path and you change your mind and you start going down a different, you actually don't want to do any of that, I want to do this. It's like, it's a lot of computation the model has to do to notice that it has to ignore that whole thing. And when both of those things are kind of far back enough in the context window that they're only half getting attended to, your likelihood that it's like actually going to like, remember the exact instructions you gave it 100,000 tokens ago is like, it goes down quite significantly.

This is all very interesting because as engineers, we are expected when we're AI engineers, which now a lot of software, meaning you just like use LMs to build software like underneath there's an LMLA or somewhere, you're an AI engineer, congratulations. But it sounds like the expectation is to be a good software engineer, pre AI, you need to understand how to write a code and it helps when you understand a little bit of the underlying, we didn't need to do that that much over time, but it never hurts. But it sounds like right now we're in this phase that to be an engineer who can write an efficient AI system that use LMs, you need to understand the dynamics of the context, you need to understand why stopping your context one way or the other can be compute, can introduce a latency. And all of these, it sounds like it's kind of more of an intuition. And of course, they're some understanding, but from talking to you're like, well, it does this computation like I know, you know, because you've tried it out, right? Like, I'm not, I'm not a PhD in machine learning, like I couldn't actually go

like draw a mathematical proof of how this works, but we know attention is quadratic. And the more stuff you put in, the more it has to spread this attention out over everything. This just feels like an absolute new area. I'm like a little bit very different to like what were you like software engineering, which is like pretty kind of like black and white, right? The compiler doesn't compile. That's true. I mean, there's a different kind of intuition. I was talking

about this earlier as well. It's like there's a different kind of intuition that you develop over years as a software engineer. And there's many categories of it. But the one I'll call attention to that is like a thing that you cannot teach, you cannot do, you cannot learn in a textbook. The only way to learn it is like, I know bad patterns and software because I have debugged them at three in the morning. This is my buddy Jake from Netflix at this in his talk at

engineer code is just like, there's no better way to learn what is good and what is bad and what works and what doesn't, than suffering through the thing that doesn't work. Well, speaking of suffering through the things that doesn't work, a new paradigm that is starting up as a loop, loop engineering, the idea that instead of writing prompts, just write loops, set up your loops and this all started with the wrong wiggum technique where it will just, well, I guess it's an early version of loops that we're just loops around. And now we're hearing some of the biggest laps talking about that. They're actually just doing loop venturing. What is your take on? Have you done some loop engineering yourself? Have you set up some loops?

And what do you think is good about it? And what do you think is bad about it? Yeah. So I think of loops as, I mean, this could I could ramble on this for 10 minutes. This is an entire talk. But I'll try to, I'll try to lay out some high level stuff and then we can dig

in where we think is most interesting. We had a Ralph Wiggum. It's actually a year and four days ago was the first time I saw the Ralph Wiggum demo. Like, Jeff Huntley was just like visiting SF and he just like came through and like dropped everybody's jaws with his like, yeah, I just ran it on and around the clock and spent six grand in six weeks. And like, I built an entire gen Z programming, like, look at compiles and it has a stage two compiler where the compiler for the language is written in the language itself and all that insane. And the core lesson from all of that, I think, was the idea

of back pressure, which is basically, and I think a lot of people were doing this for very, have been doing this for a long time, which is how do I let the model check its own work? How do I automate the process of getting feedback into the model? And there's lots and lots of different flavors of this. You can have deterministic linters, you can have unit tests, like part of what made the programming language easy to build with Ralph is a programming language can be infinitely verified. You write the code in the language, you compile it, if the compiler fails, you go fix the compiler, you run the program, if the program fails, you go fix the compiler, like it's like, it's very, very verifiable.

And I think the lesson in loops engineering is like, if you can make a problem very verifiable, you can kind of like treat it like a black box. And then have it loop because it will keep improving itself because of the verification loop is already there. Exactly. And so like you can do this with CI CD is like, I do this every time I do it really like I'm tired to see I CD is slow. Cool. Go research the code base, make a change, make a pull

request, run the test, see if it's faster. Try again, run the pull run the test, push to the branch, check again, see if it's faster. And so it's like, if it can verify its own work in a loop instead of design, instead of saying, let's try this approach, or let's try that approach or success and being really back and forth, you just say, my goal is to make CI faster. And you tell the model, here's the steps, here's the five, here's the five steps, you're going to write some code, you're going to commit it, you're going to push it, you're going to launch a sub agent to watch the job until it's finished, it's going to tell you what happened, then you're going to decide what to do next. And so that's like the very simplest example I have of like designing loops. And you just set the goal, which is cloth

code and I think code except both ship flash goal, which is you just set the goal and it iterates until it reaches it or as long as it makes progress towards it. Exactly. And so it's like, if it's verifiable, if you can measure this is auto research, auto research is like, Hey, go make this model twice as fast. And like, it's just a prompt that tells the model like, go to it over and over again and try things until it actually has good results. So that's what I think of loops engineering. I don't

know, we do a very interesting kind of loops engineering where like the challenge is like, I think it's very easy to get very excited about building the thing that builds the thing or building the thing that builds the thing we talked about. And so people say, Oh, we need to like redo everything as this big like agentic first factory, maybe even a dark factory and they're like redesigning their entire thing to be their infrastructure for the next five years. And I'm sure one thing we know of in engineering, especially a pragmatic engineering is how can you make this more incremental, how can you make it more continuous? And a lot of people don't have the option to just, Hey, I ran a route loop for three days and it fixed every lint or error in our code base. Here's a 60,000 line PR, who wants to review it and who wants to sign off on merging and deploying it and that there's not going to be any bugs, nobody. So I think the thing I'm most excited is actually like what we call

iterated loops or like slow loops where we basically have a cron job. We have the structure of the loop is really easy. Like run this linter, fix one thing, commit and push. And then we run that every night in our GitHub actions and we wake up every morning to one PR that makes the code base a little bit better. I like the stow loops. Yeah, and it has two dimensions. So you can add now we have a blueprint

for it and actually Kyle just shipped a scale so that you can build these yourself. You can add more like feedback mechanisms. So we have React doctor for the front end, we have another anti pattern that has no deterministic tooling. Kyle's like, here's what good looks like. Here's what bad looks like go fix one thing and bring up as like prop narrowing base. We have a bunch of optional props

and most of them don't need to be optional. It's like here's how to make the prop not optional so that you know that the code just is like clinging or easier to reason about. And so you can add more conditions, more things of like fix one thing. I want to wake up to a PR. So now we wake up to like four PRs because there's four separate things. And then the other dimension you can do here is as

you gain confidence, you can increase the scope instead of fixing one thing, fix four things. And so these are like other ways to think about loops where it's like something that's not a human triggers it to start, whether it's, you know, an alert from century, whether it's a user feedback, like support ticket, whether it's PM writes a ticket, whether it's a test is failing, any of or it's a chronic runs on a schedule. But it's like the trigger should be something that you don't have to like press a button on. And there's a defined workflow and it makes everything a little bit better. That just describes letting agents fix things without a human pressing a button. But what if a bug is too difficult,

not just for an agent, but also for humans to reproduce little fix? This is representing sponsor and this is comes in. I was recently pairing with the Anticists team where we did a walkthrough of how they help fix a nasty bug in that CD, the open source key value store used by Kubernetes. This is a bug that actually happened in FCD. The team noticed that the linearization validation assertion failed during the regular synthesis runs. This is not good because the linearization guarantees strong consistency,

so this needs to be fixed. So what the SCD team did was run a casualty analysis inside Anticists. This generates this graph, which is a bug probability graph. Here, the x-axis is virtual time and the y-axis is probability. Now we see that something happened just before virtual time 24 that caused a huge jump in a probability that the bug would occur. Going deeper, we can look at the entire set of timelines.

Vertical lines going down represent events branching off from the same state. And the purple dots are where the bug happens. If we look closely enough, we see that all of the failures come from one parent gotcha. This is such a useful debugging tool. Indiana team was able to figure out that process possible causing the bug using all of these in TCC's debugging tools. This non-deterministic bug was

diagnosed in a deterministic way. How cool is that? Oh, and this is an actual bug that then got fixed in that CD. You can see the bug and the fix in that CD is to get a bug. Honestly, the tool that Anticists built for debugging feel pretty darn futuristic, but they are also really powerful. Head over to

Anticists.com slash pragmatic to learn more. I'd also like to talk about our season sponsor Sentry. Sentry is a tool I use for application monitoring on all of my projects, including their pragmatic injury back end. I've used it for 10 years now, starting with one I worked at Uber. And the Sentry V-Trim liking is your Sierra AI agent, which helps investigate production errors.

For example, here's an actual error I had in my application. I can just ask Sierra what might be the root cause and it brings context, and it can also make a plan to fix it right from the web interface. And the nice thing is how Sierra also works right from Slack as well, not just from the web. One place I find even more handy to use Sentry is from CodeX or Cloud Code using Sentry MCP. Also, you can set up neat automations like when I resolve Sentry issue resurfaces, you can get off a cursor agent or GitHub copile agent to investigate the regression, read their relevant code, and open a PR with a suggested fix. I'm not a fan for using AI tool just for the sake of it, but I

really like practical integration where I can fix errors faster and with more context. Check out Sentry at sentry.io slash pragmatic and start monitoring and fixing regressions today. And with this, let's get back to Dax and to agent to Gloopz that triggered themselves. You said we can get more ambitious and we can add more things to it, but I'm going to quote you with one of your tweets, which says, this may surprise you, that does it coming from me, but I think we're in for one to three year period where stuff might break at 3am and you're rolling on loops to fix it and nobody understands what's under the hood and you're looking at it as an existential threat to your company. Yes, yeah, that one was great. That one did a lot of numbers.

It resonated. Here's the other side of it is like, I think that today with today's models, today's programming languages, today's infrastructure, you might get away with not reading the code. Problem with loops is like at a certain point, you're going to generate so much code that you can't read anymore. This is the strong VM dark factory. This is like Ryan LaPobolo's like, harness engineering, just spend as many tokens as possible. We tried this. We built a light software

factory in July of 2025. And by November, we had shut it down. I think it takes about three to six months of you shipping all the time with nobody reading the code before you realize like, wow, this is getting way worse and it's easier to start over than it is to fix it. Like the models have made the code base so bad that it is actually going to be easier to just like rethink this from scratch. I mean, that's okay because we have AI and it's easier to rebuild things from nothing. And like, usually when

engineers say like, oh, we can't fix this, we have to rebuild it. The feedback is like, no, just refactor in place was constantly keep the code base getting better. You mentioned what I said, you'll notice what I said was not use loops to ship the features that users want. We use loops to actually improve the code base quality and we read all the code because we care about how it's architected and we care not just about the system architecture, but what I would call the program design, which I think is something people are going to wear the interfaces, where are the seams, how are we doing dependency injection, all of these things that like make your code base more maintainable over time and keep you from falling into this trap of like, okay, well now if I change something over here, I broke something over here, this is a classic problem software engineering that like software engineering was invented in the 1970s because we realized we needed techniques for avoiding that problem of like this giant ball of spaghetti. And I don't think the models are smart enough. And I don't think we actually have the training and the benchmarking and the eval

techniques to get models to write code that is more maintainable over time versus they're all trained on sweetbench and sweetbench looking things, right? All of the benchmarks are basically like, here's a commit in Django, here's an issue that was filed around that time. See if you can create the fix of the human created and it's Django and it's Apache and it's 100 repos in Go and C++ and Java and all these different languages. But they're all it's like the problem with training models on maintainability is like the cost function of bad architecture and bad program design can't be evaluated by running the unit test because it hits you three to six months later when you're like, holy crap, like no one can make the software becomes so hard to change. Is this not similar to how senior software engineers, why it took years for someone to become a senior because typically, and in some environments you become a senior faster, typically fast moving where there's a bunch of issues and you have to keep fixing it. Sometimes you have some people working the same place for 10

years and they're still not that level. The point was it just takes time for you to understand the small mistake that you make right now that snowballs into like something disastrous later and you get hit by it and you realize like, okay, things like testing matters, architecture matters, tech depth can actually be a killer. We don't talk about it anymore, we just talk about how tech depth kills or slows down companies so badly that pre-AI that their competitors can overtake them or they're just like stuck with a two year refactor and you feature them, the competition, but you have a bunch of other stuff and now they're ahead. And I will say like it is possible that GPT-7 will fix this, but if you are turning the lights off in your software factory and you're saying like, hey, you know what, like we're not going to read the code, it's fine, the models are short enough. If we give it the right feedback and just throw enough tokens at the problem, it will keep getting better. This is what led to this week.

That might work, but if nobody read the code in three months and you replace all of your code review with loops, like, hey, if a user complains, we give it to an agent. If something crashes, we give it to an agent. If a PM writes a ticket, we give it to an agent. If a CEO writes an obnoxious essay about what we should be building in Slack, we give it to an agent and then you stop reading the code because that's going to produce way too much. No one can read it. And the PR reviews

become the bottlenecks, so you replace that with agent testing and agent code review. But none of these things have intuition for software architecture because we haven't trained it in yet. And so you're going to wake up one day and you're going to have an issue with this happening to us. And like, we got through it at the time, we're still worth it. It was like, spent three weeks onboarding back into the code base that we had to stop reading three months ago because no matter how much sophisticated expert prompting, we could not get open, it was 4.1 at the time, we could not get open 4.1 to actually find the root cause. We had to go spend several days digging through the code

and figuring out, like, oh, there's just actually a primary key that's being routed through this whole thing that needs to be changed to a different type of object that needs to don't take. So this happened to you. This happened to us. Yeah. And when it happened, I was like, you know what, that sucked. That was terrible. But we did it. We solved it. And it's still worth it's still worth

not reading the code for most of the time at the cost of every once in a while, I'm going to have to spend two weeks fixing an issue by hand. And I don't believe that anymore, because I think the amount of code we're able to write now is actually like 10x or 100x. And I think the problem is just getting worse. So let's talk about software factories. Yeah. In your mind, because I feel it's an over

little bit, but what do you think of a software factory before AI? And now post AI. You know the first definition of software factory the first time it was used? No. It was a NATO conference in 1968.

Oh, great. Which would know about this. Yeah, exactly. Yeah. Great. You should ask great about it.

They talked about the idea of like, okay, you actually need to build a system of steps and like just like a factory floor, you have like the coding part and the testing part and the validation part and the integration part we had no CICD. We barely had version control like, but you needed a factory. And then it was adopted by like Toshiba and a bunch of companies. And then the next moment was like DevOps and you have like this idea of like, okay, we're going to do CICD. We're going to automate when you chef and answer, but whatever all these technologies, like instead of having dudes running around data centers, like resizing disks and stuff or clicking around the ABS console.

Yeah, exactly. It was like, cool. We build loops. The server hits 90% disk space. That sends alert to Naggio's. Naggio's triggers. Chef runs chef makes the disk, the disk maker feedback loops,

right? This has been around for a while. And in 2018, I want to say this guy, Nick Chalane, who was a he was like a CTO or chief software officer of the Air Force. He wrote this 100 page essay of, hey, the DoD needs a software factory. The Department of Defense, the Department of Defense and Air Force and they called it DevSecOps factory. And he said, we need all the things

that all of the good enterprises are using. We need Jenkins. We need code quality scanning. We need security scanning. We need CICD. We need to be able to ship. We're shipping once every

three months or once a year. We need to be able to ship every day, like all these other companies. And the way we do that is we actually embrace all these automations and technologies that engineers are 90% of the issues are caught by automations instead of people actually like manually checking it or manually reading the code or manually integrating modules together. Wow. So I'll talk about forward thinking in the government.

I know. No, I was surprised. Oh, nice. I mean, that was part of it. It's like, hey, look, we're falling behind in like, you know, I don't know exactly all the reason. But I imagine also about

like attracting really good talent is like, hey, look, if we have like the modern software stack, we're building things fast and we care about efficiency and we care about people using people's time well, we care about them spending time on the hard parts of the job, not manually looking for SQL injection. It's like, you can automate that. So this was software factories pre AI. Pre AI. Now I've heard the term a lot more because of AI. Yeah. Is it the same as a different?

So this is really hard to say without a drawing, but I'll try to draw it out. At the core of a software factory, you have like a source of work. Most you can imagine a linear, a Jira, a source of truth, your object, whether it's a spreadsheet or whatever, you're like, what stages is the work in? Yeah. And pre AI, you would take, you know, you would maybe do some architecture review planning, you maybe do some sprint planning, and then people would take tickets off the queue and they would go build them. And then you would make a pull request and people would review it and you would run CI checks,

and then you would send it to prod and then it would make contact with your users and your users would complain about stuff and that would go to your support team and back into your work tracker. And it would crash and you have issues now going to your monitoring stack and that would go into your tracker. And that was your loop. And then people would take stuff off the tracker based on priorities, product managers, engineering managers, engineers, prioritizing work, and then we'd go and do that. And the first change is like this long, long, long, long, lots of phase. And this is also why when

like a developer shifts a bug, but by the time it comes back to you, it might be two or three months, or even longer. And by the time it gets fixed, it might be a year or two. And you know, this is why when you're using a piece of software, it's like that annoying bug and you talk with customer support, but it's just a very like long lead and sees that each part of the factory, if you will. Yeah. And the step where someone pulls a work item off the queue and starts working on it is, you know, a couple hours to a couple days before it actually gets integrated and everything else and touches user and that's in a great world, right? Sometimes you go build it and then you merge it

and then it actually gets released three months later. But we're going to assume we're in a fairly modern, like we're somewhere like the Netflix or a meta where engineers are capable of shipping 100 times a day or 1000 times a day, but it still takes two, three hours to do the work. And now with an identity factory, what you do is you take out that person building the thing and you replace it with an agent building the thing. And so you have orchestration to trigger things, you have a sandbox, you have an LOM, you have an inner harness, you have an outer harness, which is like the dev environment you build for the agent and maybe you go to browser, you get the video recorder, if you use like things like cursor background agents that kind of build this outer harness around the inner harness that is the coding agent. And then you make PRs with that problem. There's that like, okay, now it takes

10 minutes to do a build instead of two hours or two days. And so now the bottleneck is code review. So okay, let's throw a bunch of AI agents a code review and let's do agent testing so that like, we can basically catch a lot of the easy stuff and humans are only focused on the most like important critical core parts of the code base. And then the next level of your agent factory is you do the top is like, okay, then it gets deployed, it goes to prod and a user compliance, you just hook your support to you right up to the agent, someone complains about something agent tries to fix it and instead of looking at a ticket and then saying, okay, go set, you just close that loop and instead every time something goes wrong, you just get a PR and then every time something crashes in century, your data dog or whatever, it goes into the tracker, it gets picked up by an agent and you get a PR. This is the ramp and spec thing. This is the only difference is like,

then you have so much code review and people say, well, let's try turning lights off. Let's just take all the human testing and review steps out and we'll say, okay, cool, if users complain that it's broken and if users don't complain, then it's working and we're not going to read the code. We're going to use the whole system is a black box. So you said you tried this out when it was like, oh, it was working when you built this off or factory is running beautifully until it just blew up on your faces. How do you think of this model? Because I can see an ideal world where it

works, but clearly we're not an ideal work. Like, where do you think we are like, and could some of this actually work at some point? Like, what progress are you seeing right now? And what is it today, the situation, like how much of this do you believe we can automate or should we automate? Yeah. So if you know me, you follow my stuff, you know, I stand for three things. Number one is like

cutting through the hype and the jargon and going, trying things and talking to people who are using things and figuring out which parts of this actually work at our valuable number two, we talked about words. I try to find and protect useful bits of language because I think it helps us all move forward and when you take a useful word, like agents or you take a useful word, like software factory, and then you semantically diffuse it as another Martin Fowler word, you make it mean everybody likes the word and it all becomes hype and everyone starts agents means nothing anymore. Agents could be a chatbot, it could be a slackbot, it could be a coding agent, it could be tools in a loop, whatever it is. So I like to protect important, useful words and help us all elevate the conversation out of that hype and jargon. And then I care a lot about going one level down beneath where I'm generally working.

I think there's always, this is something with context engineering, I was really actually going and building LLMs or understanding or training LLMs, but knowing how the training and transformers works informs how you build it one layer up. And for the software factory, my version of that is I spent the last couple of weeks going really deep on reinforcement learning with verifiable rewards, RLVR, which is like this very productionized, it's not like RLHF, RLHF is still fairly academic and pure, RLVR is this like, it's a machine in these labs of how we train these models. And I'm studying like the benchmarks for coding agents and the techniques for training them and how we like, give it a small problem, have it solve it, delete the test changes it made, revert them, apply a test patch, see if it passed. And then even the frontier this year, we have like, we can get into this later, but like frontier code and sweet marathons, new benchmarks that are supposed to be like better in evaluating models ability to maintain a code base over time and write maintainable code.

And they are better, but I don't think they're sufficient. It was basically this idea that like, the only thing that made cloud code good was reinforcement learning. And the dimension along which it got good was like, we made a model, we trained the model and the harness together. And so the model got really good at calling the specific tools in that harness, really good at reading files, writing files, searching for files, all this stuff through doing these problems. And that was what made it feel so much better than all the other CLI coding agents that came before it. And so people are like, okay, that was so much better. And they're just going to keep getting

better. But it's like, it got really good in one dimension. And the dimension that they're not getting better in, because it's hard, expensive, maybe we need to like get a lot more creative with how we design these verifiers of benchmarks is in, how do I make code that in three months is going to like improve the productivity of humans and agents, mostly agents, but humans and agents instead of making it worse over time. And so you think that part is just missing, we haven't seen too much improvement. I haven't seen, obviously, no one knows what the labs are doing internally, because it's all very secret. But I think if we looking at where the benchmarks tend to reflect where the labs are,

right, if there is no benchmark that can convey to me, did this model write code that is going to make my code base better or worse, the best we have is I think frontier code from the cognition team is really interesting. They have like, did the test pass, and then they'd have like two layers of model review. So they have a judge model that checks, okay, is the patch the model made similar to the patch that is like the golden answer set. So even if the model didn't write the exact code that the benchmark was expecting, was it functionally equivalent? And the next one is like a code quality review from another judge model. And like, that's better, but it's not sufficient. And this is why

I also think the code review is like, yes, it will catch things and it will raise your floor. But I don't believe like the model writing the code is the same model reading the code. And if you ask a model, hey, is this code good? It's going to be like, oh, yeah, it's great comprehensive. It's going to test. You've tried this. I'm sure. And you say, okay, review this PR that my coworker wrote

and tell me everything that's wrong with it. And I'm like, oh, it has this problem and this problem and this is sycophantic. And they want to tell you what you want to hear. And so like, it's really hard for me to trust the model to evaluate the quality of code that's written. And so I have some ideas on like, okay, can you build a benchmark where the model builds 20 features in a row and maintains the code base the whole time? And it doesn't know what features are coming. You treat like a real product

team where you don't know what you're going to build next week until you get there and you find out what's most important. And then can we try to evaluate like, can we build a problem like that that's hard enough that most frontier models fail by issue six or seven? It's fair to say that, you know, like we've had the software factory before AI, it was just like lots of loop, it was like the PM giving a ticket to dev, the dev building, deploying to production user customers using a customer support getting tickets and then creating PM triaging and it kind of goes around like in this loop. Is it fair to say that the software factory of how a company, a team builds and maintains software that is changing because now everyone's replacing some parts of it, maybe the least advanced teams will just be, devs are starting to use clock code or codecs to write faster, they're not spending as much time on there. Some others are also having the deployment, the feedback, some actually have the agents already one-shotting bucks. Like, is it fair to say that the software is just changing

everywhere, maybe at different speeds, but everyone, I think every team who is building production software, they're frantically experimenting, trying and everyone's at a different pace. You'll have the AI native starters where most of this will have agents in them and you'll have the lagers who are more cautious ones, they have agents in a few places, but not in the others. Well, and I think that's the key is like if you want to do loops engineering, you should build one loop at a time and you should keep them small and contained. I think everything except stop reading the code is really good advice. Take support tickets and turn them into tickets in your system and then maybe turn those into PRs. Great. The advice that I have in like what we kind of like

are chasing at human layers, like how can I add another checkpoint in that factory? Instead of having one human at a viewpoint where you're reviewing PRs and sometimes there are 100 lines and sometimes 1000 lines, but it's quite a lot of effort for especially if it's bad, especially if it needs rework, it's quite a lot of effort for a human to be like, okay, this is wrong, go change it in this way and then you loop back to the agent and then you come with another one and doing a lot of loops on there. Once the direction has been committed to, it's really hard to steer off, like you're better off just kind of restarting from scratch, how you build controls and mechanisms around that. And then my take is like, if you do a little bit of human agent planning and like discussion before you hand it to the implementer, whether it's planning and specs, whatever you want to call it again, this is spectrum development is another word that has become kind of very like model as far as what it means. But basically, how can we spend an hour before we start building so that the PR when we read it only takes 20 minutes because the code is perfect, instead of not touching it, just literally saying every user reported issue becomes a PR through the loop, and then we read that PR and it takes six hours because it's back and forth with the changes and things. It's all I'm all about, like, let's find leverage. And so you basically you have three

options in the software factory world. If you're going to go all in on the agent, software factories, you can turn the lights off and just let everything flow and pray that you don't create too much slop and pray that the next generation of models comes fast enough before you create a giant pile of ash, you can slow way down and read every PR and read every line of code. And then you're only going to really get modest benefits from AI because I think you should expect maybe 30 to 50% lift in productivity is kind of what I see when we go into teams. Or you can find the right leverage points where humans can actually, in our spent over here in planning, can save you four hours in implementation in terms of fixing and going back and then getting the design right. And that's what I call like seeking leverage. If you find the right leverage points for the agents

to guide the work, then you can actually move like two to three times faster while maintaining a 99% accuracy to if the humans were carefully right at this code by hand, how would it come out? Jumping a little bit back to ideas that will come back to this, this was earlier, maybe it was last year, but you had the research plan implement. Can we talk about the original research plan implement framework and then also what you've learned about this approach? What you got wrong about it? Yeah, sure. Yeah. So, I mean, the first time we talked about RPI was in August 2025,

and it was basically like the research was this thing like, Hey, before you go build anything, go read lots and lots of code use a bunch of subagents and parallel understand all the code. It was this technique that worked really well for hard problems and complex code bases. You just ask Claude to do a thing that that would read three files and make a change that we have no context. So you start the research, you don't even tell it what you're working on. You just tell it, Hey, can you tell me how this system works and this system and how they connect together?

And then you get a markdown doc out and this was the context engineering part is like that would take 100,000 tokens of context, but you would get a 10k token doc out of it that summarized it. Then you would start a new context when you would do planning. And the planning would be actually realized like the plans that we were building last summer were actually terrible, but it would basically be this long you'd say, Okay, now here's what we're building. Here's the research doc, build a plan to implement it. And in retrospect, now that we see like everyone is obsessed with how do I get agents to work for longer. I think the reason why in like May, June, July, August of 2025,

that a lot of people became really interested in planning was there's a very powerful lever to get agents to work for longer. If you said build me a B2B SaaS for a burrito delivery, you'd get like a home page and that's it. But if you said build me a plan, it would build out this big plan. And then in the next context, when you'd say, Hey, here's the plan, here's the plan, here's all the change we're going to make go implement it would actually keep going until the plan was done. The plan was a really good way to anchor an agent and remind it that like, Hey, you're not done until this is all finished. So that was the original RPI and the plan doc. What

was bad about it is it didn't give you leverage. The plan was every single line of code that was going to change like in diff blocks and like all the new stuff to write. And so like people would review these plans. We recommended this. We told people to read the plans. We read all our plans.

And then eventually I found myself like, I just kind of skim the plans. And so you're not really using it as a way to restore the agent. It's just kind of there. And then you go write the code and there's a crap. Some people would review the plans and the code and it's like, okay, well, the plan was Tokyo 20 minutes to read. And then the pull request takes you 20 minutes really and they're

different. And you actually doubled the amount of time you're spending reading code instead of like doing less of it, anti leverage. And hang on, what's the different development not related to this? The one that Amazon Kiro for example, and GitHub workflows a year ago did, which was it also it first generated a plan and it had the human review it and then it started to and you could edit it as well. And then it went off and input this part and it looked beautifully on the surface. It should have worked great. But it's also the garbage outside of some maintenance projects.

I think it just didn't work. Like I'm all the feedback I got people just stopped using it because it just didn't really work that well. It just rhymes to the RPI framework a little bit the original one, right? Well, so our thing too, like the biggest difference between RPI and spectrum and development and some people prefer to RPI as spectrum and dev because for some people SDD, all it means is I use a bunch of more down files while I'm coding and forget what's in them. I just spectrum and dev. Those are my specs and I'm using them to drive development. There's this opening eye researcher

who talked about spectrum and dev and like, hey, stop reading the code, just write the specs and treat like the coding part is compiling specs into code. That part never really materialized. Maybe with GPT seven, you know, but the channel I'm on a GitHub issue in spec it that has been open for a year and every couple of weeks, like there's a new email on the thread of people complaining about this problem like, okay, I had my specs and then I added the code and the code drifts and the specs. How do I keep the specs up to date as the code is changing? And it's basically like, you now have two sources of truth and it's it stops being useful. And so like that's why when RPI, the idea of

the docs is they're all for a while we kept them around. But after two or three months, we're like, Oh, these are actually like tactical execution docs. I do the research, I do the plan, I do the implementation, I throw the docs out. And the next time I need research, I just do it from scratch, because tokens are cheap and my time is expensive. And the amount of time I might waste if I reuse a research that is no longer in sync with the real estate of the code base. So we just create it live

every time. This is why it's a context engineering still matters, creating artifacts that compress the state of the code base and compress the intent of the builder into small things that can be reused in the future for the scope of a task is like a very powerful like tactical approach. But it's not a thing like I have very few opinions on like what sorts of docs that you should leave lying around your code base that are like evergreen. I've seen people try to maintain parity between documentation or specs and the code itself. And I don't think anyone actually like found it very useful. Like you can do it and it works, but it's like the ratio of the

effort it takes to keep them up to date. And that's a trivial you could do this with AI probably, but I've never known anyone who was like, Yeah, this is great. And we're glad we have it. Like you could do it and it might help. But I don't think anyone found it useful enough to like maintain a system to keep the specs in the code and sync versus just using the code as the source of truth always. Now you mentioned something interesting, which is with context engineering, you need to

sometimes compact and you've previously talked about intentional compaction that when context is noisy, the little bit compressed, the useful parts into a clear like market artifact verified and then start a fresh conversation. Can we talk about this kind of compaction and why it's important and and it sounds like it's going to be building block rate or is for context engineering, right? Yeah, no frequent intentional compaction is the building block. It is it is completely comes from context engineering is context engineering is like, how do we get the most out of today's models? How do we change what we're putting into the model into the context window into the agentic chat? How do we control that in such a way that we get the best results possible,

which means doing as much work as possible in the smart zone, the first hundred thousand tokens of the context window? And this intentional frequent intentional compaction is basically like, okay, the research step, we're going to go read a bunch of code and turn it into a doc. That's our compaction. We take that forward in the next session. We're going to read, we're going to read the ticket and the intent and turn that into a design document that we call it. Like, okay, here's the high level

spec of what we want to do. Here's a high level like current state desired end state and then a bunch of design questions the model has kind of like a very thorough, maybe even over engineered like plan mode and then you take the research and the design and you do a new session, new context when you're like, cool, you've compressed the intent and you compress the state of the code base so that you can then do your planning of like, okay, we know what the end state looks like, we know where we're going. Now let's break down how we're going to get there. All of these different steps of the research is pretty hands off. I don't read the research docs. It's just like, go read a bunch of

code and then like make a doc out of it. Models are pretty damn good at that. If you ask it to find a bug and have opinions about the code base, that's different. But if you just ask it, what is the intent and how this stuff fit together? That's usually pretty straightforward. But designing the

end state of the software, the architecture and the program design models are not great at. They make a lot of like, they make decisions and sometimes they're right and sometimes they're wrong. So we have a human in the loop there. And then the steps to get there, I talked about this before, but models love making what I call like horizontal plans. If you ask a model like, build a plan of steps to go build this app, is that cool for the database and then we're going to do the API and then we're going to do the front end. I'm like, well, that actually kind of sucks because we're going to be on the other side

of 2000 lines of code. And let's imagine this is an existing code base, we're going to change all these different parts of the system. I can't test it till the end. And so what I would do is like, okay, how would I have built this if I were building my hand? Well, okay, I would probably create a mock API endpoint with fake data. And then I would go kind of get the front end kind of how I

wanted to look. And then I would actually go like build a services layer and actually wire the data through. And then I would make a database migration to make my new table. And then I would actually add a lot of business logic. And then I would add a bunch of error handling. And it's completely worth

talking about how models like models will write the database layer and all the error handling without anyone's ever touched or seen the code or whatever it is. And so this is another place where we like to have humans involved because humans have really good taste and judgment. And like, I would rather read five separate little mini diffs of like things that I can manually verify and explore than read 2000 lines of code. And well, it's not working. I don't know where you don't know where because you wrote the code you were supposed to get it right. We talk about compaction content and you're like,

how can you stay in the smart zone of the context window, which is, again, the dumb zone, I will say disclaimer, it's really good training wheels if you don't have intuition about this. So let's just define these things. What is the smart zone and what is the dumb zone? So it's, it's a little bit blurrier than like, I would like, I would like it to be. I think in November, we talk about the first 40% of the context window, but then we had million smart zone. Yeah, then we had million

total context windows and I changed it to like the first 100,000 tokens. If it's really like 4.8, I usually will go up to like 200k. But basically the thing Jeff only had in Raoul, the less context window you use, the better outcomes you'll get. And basically the smarts, the smarts will mean, meaning if you have context in that first part, it should work a lot better. And then like the dumb zone is like once you have stuff there is kind of forget about it, like it'll be confused. It's not

going to do much like it'll degrade. Yeah. And there are times and this is an intuition thing like, I will often go up to 3, 400k token for is rare, but I will go up to 250, 300k tokens for certain types of work where my intuition tells me that I can keep working without, without degrading the performance. But if you don't have good LM intuition, like 100k for smaller models, 200k for these like really beefy like codex and opus 4.8 models is usually a good like training wheel guideline of like, if you pass their your quality results may be degrading. The biggest tell I see for this is often the models trying to get the test pass and you're 200k talking about, well, let me try this. Okay, let me try that. And it's like trying a bunch of stuff and it's getting more

and more extreme and it's like, oh, let me delete your dot end file and try again, like, this is where things get really, really weird. And so it's like, if you start to see certain types, if I'm like, oh, we're at 300k tokens and I need to like fix the unit test, I'm like, cool, write everything we did to a file or even I'll just do like a built in compaction depending on the model. And then I'm starting a new session at 30k or 50k tokens. I'm like, cool, we're gonna do a hard thing, which is, you're gonna get this frickin test to pass. And you're not gonna be stupid about it. But anyway,

one thing that you said about the model being dumb is you said that if the model ever tells you are absolutely right, you should start over. And we've all had that when it tells me like, oh, you know, you're absolutely right. And I'm like, we just get annoyed. But why should we start over? What's happening there in your observations?

Yeah, it's great. Yeah. And the new you're absolutely right. I think is you're right to push back on that, right? Yes. That's open, right? Yeah, open. It's like, you didn't run the test, did you?

You're right to push back on that. I totally did it. But no, for me, you're absolutely right. It was always what the model would respond. If you were like, that's totally wrong. You did it. Like,

if you if you said something where you were angry or frustrated or just wanted to point out that it's done something wrong, it would respond with your absolutely right. And most of us have had the experience of it says that and then it continues to do the wrong thing. So it's like, once it starts doing dumb things, because there's there's four things in your context, one of the matter, there's the size of it, how many tokens? There's like the quality of the information is like, is there any incorrect information? Like if the model had some thinking trace where decided the wrong thing was true, is there missing information? Is there like, have context missing that it should

have? And then there's the trajectory. And the trajectory is very subtle, but you may have had sessions. That's your lecture, meaning you're prompting the actual history of every thing. I call trajectory is like the actual history of like what the agent has done in the past. And so if I say,

hey, make this change and the agent makes a change and then it runs the test and then they're broken and then it fixes the test, I have very high confidence the next change I asked it to make, it's going to follow that path again, because it's like, okay, here's a conversation. And the last time the user asked me to do a thing, I made a change or end the test, broken fix the test and then I called the user. But if I say make a change and it makes a change and it doesn't run the tests, then I'm on a different trajectory. And if I say, okay, make another change, it's like, basically they're autoregressive. So they're predicting the next, what's the next message in this conversation? And so the example

we talked about in Noviops is of course they're like, hey, the model makes a mistake and then you yell at it. And then I made another mistake and then you yelled at it. And then it's like, cool, what's the next message in this conversation? Well, look, if I read the history, I should probably make another mistake. So the human can yell, okay, that's a great example of like time start over.

Let's talk about some observations on how software engineering is changing. One thing you talked about recently on the evolution of the coding meta is going from token harder to token smarter. Can we talk about what you mean by token harder and token smarter? Yeah, so token harder is, I mean, I'm in a group chat called Hyper Engineering and it's all like people trying to max out their clubs. Oh, wow. Okay. That sounds like a fun place. It's a fun place, but it's like all token

harder. It's like, look at all the side projects I built. It's look at everything that I've gotten my cloud token. I've got six cloud code accounts. I've gotten all of them maxed out every five hour period. I've timed it out. So I always use all the tokens and it starts up immediately when the limit

resets. And so it's like getting into Eli Goldrad and the goal is like optimizing for utilization and efficiency of one node in your factory rather than the end and end goal of like, how do we shift value and things that people like that are stable and like will last a long time? But that's my idea of token harder. And it's the same thing with the dark factory thing is like, hey, if you if you're removed humans from code review, you can push more tokens through the system. So we talk about software factories, but what is the dark factory? So the dark factory is this comes from this idea of

like, there are factories where everything is automated by robotics. So you can imagine like a car factory where it's all robots building the cars and they don't have lights because there's no humans. Oh, so that's where it comes from dark factory. Yeah, you walk in there, there's no lights, there's not even light switches. So it will be the fully all-made software factory where it will be like no human input, basically. No human input raw materials go in, cars come out. And I think in

a micro like you can have many loops that are dark in your in your thing. Like, hey, if the code review agent comes back with a problem, you loop that back to the builder agent, it fixes it and comes back. And that's dark. You don't need to keep in the loop for that. But the full dark factory, we don't read any code. Yeah, it's a good way to maximize your token utilization. It's like if your belief is

like, my job is to extract as much intelligence out of the machine God as I can, because that's how I get the most value and the most leverage on my time, then token harder. And my take is basically what we talked about before, token smarter is like, okay, how do I move faster? How do I get as much value out of as AI as I can without having to turn the lights off while still maintaining control and taste and judgment and understanding the system architecture and having a lot of like applying my hard one opinions through 10 years of software engineering to the design of the program so that I can feel confident that the code's going to get better and more maintainable over time. The same thing of like you look at like the SRE team inside Google, they brought out this book, SRE site reliability engineering. And the whole tech was like, hey, we're going to go from one data center to five data centers. And we need the same six person team to be able to manage five data centers.

And we need the same six person team to be able to manage 50 data centers next year. And it's basically how do we apply software to this problem? So then instead of scaling linearly, like, okay, every data center needs five DevOps people, so we need to scale the people with the things. How do we continually automate the parts that we don't need? So a little bit orthogonal and maybe even like contradictory to what I just said, but this idea of like, how do you find leverage and the way the way what I think what you were saying there is like, from Google that never seek to remove those SREs from the process at all, they just said like, look, can we think ahead and scale yourself and they actually grew the team. It wasn't actually six people. It was more like, I think Google specifically

said, okay, we have five data centers next year, we have 50, you're six of you, we don't want to have 60 people, we don't want and then management leader and all that is like, how can we do it with like 12 or like, or like 10? And then when we'll have 500, and now actually their SRE has grown, but of course, yeah, but but they never, you know, I think as engineers, like we feel pretty threatened when someone says like, all right, we just want to have zero engineers, like, I mean, that's not a fun place to work out. But what sounds like, it's not possible place to work out if they're zero engineers, neither of us can work there, right? But the notion of token smarter is like, let's keep humans in the loop, let's keep adding value and figure out what are the parts which are not as relevant, boring, where we don't need it. And so like one developer can probably do more than before, but you are built to like be part of this whole thing and the lights are on in the factory.

Yeah, and it's like basically, I think what I'm trying to do is like the connection here is like SRE builds a thing where like headcount scales at like a square root function or logarithmic function, whereas their output scales like literally. And you want to say that the way you do that is with good architecture and good program design. And so in order to like avoid this problem where you have to throw more people or more tokens at the problem, if you design good software in such a way that it gets more maintainable and more scalable over time. And like just today, it doesn't feel like like basically you need humans in the loop to be able to do that. Let's talk about AI swap. At one point you wrote, yeah, I can write your code, but it also

writes respects and PRDs. But the same result we swap in, swap out. If you also are thinking, you're going to get garbage. Yep. So yeah, that's basically the idea is like the way we think about like getting high quality outputs is like, yeah, you can write the code by hand, or you could sit with a model and work back and forth and go maybe a little bit faster. And you have control and every

time it makes a change, you go read the change and with bad, you tell it, nope, we want it like this. You kind of incrementally, slowly, this is like kind of the stage two or stage three version of working with agents where like the agents writing all your code, but you're kind of very much in the loop. And this will make you go faster, but it won't make you go that much faster. It won't make you go anywhere near there's like that level. And then there's like the maximum speed you can go while still caring about the code. And then there's like the maximum speed you can go if you turn the lights

off. And so we always think about it as like in terms of leverages like, okay, let me take everything starts with like a sentence or a voice note ramble. Like I want to build this thing as it can work like this, whatever it is to let's say like average, like two sentences, I got to fix this thing, or there's a support ticket, I got to fix this thing. If you can turn that with AI into a one-pager and then turn that one page and make sure that's correct and then turn that one-pager into a three-pager, make sure that's correct. And then turn that three-pager into a 10-pager like detailed outline.

Then you can write 100 pages worth of code. And it's maybe not perfect. You shouldn't like sweat over these documents and make sure they're perfect. But you're increasing the chance that like you're decreasing the uncertainty of the outputs. It's like you can think of like a line of like where it's going and then you have like the probabilities of where like it might go in that range. If you are kind of reviewing along the way as you get more and more detailed into how

what you're building and how you want it to be built, you kind of collapse the uncertainty and the set of end states that you could land in. That's me doing the physics thing. Like you got to super impose all these probabilities. And like I just think that like I think people who really like playing real-time strategy games are probably going to be really good with AI because you kind of have to like I don't know, Matt Polkock was just talking about Bog of War and like things that are at the frontier of like there's stuff we don't know about this problem yet. How can we find that out and how can I make the best decision now knowing what I have seen? I've seen a couple pieces of information and so

there's a 30% chance it's this and there's a 40% chance it's this. How can I get more information? So my head I can like recalculate those probabilities and decide what's the most likely path that's going to lead us to success. Speaking of the most likely path that leads you to success, let's talk about your company. That's you've just come out of stealth, human layer. What is human layer and what is the

probability that you're setting up for success? That's a good question. 100% probability. Maybe 100% but no. So human layer is an AI IDE. It's a collaboration platform and it is building blocks for

your software factory and the basic pitch is like engineers solving hard problems in complex code bases. Basically there's two categories of builders. There's like five coders building side projects and then there's people building production software where the stakes are high and if something breaks where you get fined millions of dollars or you know we're going to lose millions of dollars of money for the company and there's a whole spectrum in between there. But it's like if you were kind of in the left half of that spectrum you're building software that matters and it has to last and be around for a while then we are helping people like that solve problems two or three times faster without descending into slop. It's like how do you maintain that near human level of quality and move two to three times faster. And what were the ideas that you built and that you came with?

One idea that we're really excited about right now, it all comes from this RPI and it's like using specs to like I mean I've kind of been hinting at it this whole time right of like okay cool like start really high level and zoom in layer by layer and re-sear and like find that leverage that helps you move faster and increase the chance that your agent's going to build exactly what you want or something that's really high quality. The other thing that's really interesting that where I just posted yesterday I said hey chat should we kill the pull request and that's something I can't talk too much about but basically the idea is like the idea of the future needs to be rethought from the ground up for agents and it might not even be a like I don't know a lot of editors kind of started with the text field and bolted on an agent's tab and then eventually you seem like cursor three I can't even find a text editor. I know it exists people have told me you can get to a text view of files but it's also very agent-first and so we started from the ground up of like what is an IDE designed for helping a developer interact with and manage the work of agents and then we zoomed out and said how do we make this collaborative and build in a sync engine and durable streams and all of these like pieces of tech that enable me to get human input and feedback on what I'm doing with agents in real time rather than waiting for the pull request time and great engineering

teams have been doing this for decades of like hey we're gonna have a design review we're gonna talk about how we're gonna build the thing as like a two-page google doc or whatever 10-page whatever VRD or the architecture requirements document and then you go to sprint planning and you break it down into little tickets and you decide who's gonna do what it's like AI can help with all of this you should if you're just using AI to write the code you're missing out on a lot of the benefits that AI can bring your SDLC and a lot of people say like well we don't need any of those meetings anymore because we have the loop with the dark factory things just fly around the loop but it's like okay but if you want to actually move faster maintain quality then like you should have these checkpoints before you go to actually write the code and you should use AI to help with that so we build this like cloud platform that's kind of has like a google doc style component where you can comment and the agent can surface like mock-ups and mermaid diagrams and HTML and all these things so basically how do we make agents like big must style everything everything's in the cloud everything's collaborative I see all my co-workers sessions they see all of mine it's almost like the benefit that slack had over email was that you didn't have to be in every conversation to know what was happening you could maintain you could see all these channels light up you could check on them you okay I don't care about any of that but if you saw a conversation that you cared about that and it's like how do we do that for engineering work versus like we really had these like very

strict even when we called it agile it's very water-folly like PRD ARD tickets everyone goes and builds for a day and then you get the PR back and then one person reviews it how do you create this more just like soup and like what is the data model for that world where you have like agent traces you have documents you have tasks and projects that group these things you have actual get diffs being streamed everywhere where it's like why would I review all the code at once when I can just always everybody's work lives in a shared environment that anyone can go interact I mean what it reminds me is like what you know github that's a software team before github and its competitors you might have a tracker somewhere but most teams were just kind of like inside the company you didn't know what one team was I remember pre-github like you know the individual teams they some of them had like a board with stickers but no one else in the company knew what they were doing they were all working isolation and now when you have github or even the internal version of github inside the company you can always see when you go to a team you see the plot was flying you can join in you have history it's all it is all kind of connected and it came together and now it's like you know for a very long time it was like you're done you're gonna use github or people will copy it so do I sense that you're trying to build something like this

workflow for like when you have the software factories which are like dark factories and loops at a bunch of places how can we have this new way of working which will feel natural but like coming up with it like it's hard work it's counterintuitive how can we do something that accomplishes what github did but like 10x better like more can specifically like more continuous and more real-time and more collaborative than like these discrete units of work that is like the pull request well I know I'm starting to understand why you're saying maybe you should kill the pull request because pull request was invented by github right like it's it's not part of git but they did it as a way for you to do a code review merge before it goes in and be able to modify it or like just reject it etc and it's probably a lot better than whatever we had before which I guess was like emailing your git patch to linus and asking them to merge it into the kernel or whatever they still do it it works for them that's the point but it only works for them yeah I don't know anybody else who does that I mean I'm sure even before get up for you you guys said what like cvs or cvs or tfs so if you have a lot of money for Microsoft they may just use subversion at in undergrad because the guy who invented subversion was a uChicago guy the year after I graduated this wish everybody to get I was like damn I learned a useless thing just for somebody's

ego specifically for AI startups or startups with building on top of AI or building AI products how important do you think location and network is especially you are based in the the valley we see research that AI startups are more frequently funded from here than normal startups as well do you see this advantage also do you see some disadvantages of being a specific made that be silicon value elsewhere I don't have really strong opinions on this actually like Paul Graham gave a talk in Sweden about why SF is cool rather than just regurgitate that I will I will forward people onto that one we can put it in those or whatever but he talks about all of the dynamics of silicon valley and the paid forward culture and the like people take you way more seriously just I live in Chicago for a long time I have a lot of really good friends from high school from college going up in LA and never before have I felt like so locked in with like my people more never have I felt more seen more connected like there's just so many people here again talking about the people who care deeply who are incredibly competent who like we'll have all the same types of problems we love all the same types of things like I don't do land parties we play video games but all my buddies will come over and we'll sit in the office till 11 we'll just do co-working and like hack on cool fun fun projects and stuff and like you can't do that anywhere else there's not enough

like a critical mass for that to just happen organically everywhere you go and I absolutely love it I yeah then the critical mass is nails it on the head what it comes to hiring what types of folks are you hiring for specifically because I'm interested in how hiring changes and what what a standard engineer means for you and how you are trying to you know confirm that those traits exist in general we we're looking for people who are really strong software fundamentals so understand distributed systems understand like the core fundamentals of CS and operating systems and these kind of things I mean you don't have to be a PhD in frickin kernel design or whatever but it's a lot easier we can we can teach we can teach somebody I think to be a really good AI developer a few months you can build enough intuition where you are you know accelerated off the ground and you can go like keep growing there it's really hard to teach someone a CS undergrad program and free muss and what's the problem space that you're excited about and in self-wrenching or the product engineering building products that you think in the next few years it's going to be one of the interesting things that you're going to be attacking my co-founder could talk more about this but like there's a lot of interesting things happening in in real time and cloud and sandboxes and sync and kind of like using these new building blocks that have gotten really solid in the last couple

years we're big fans of the electric sequel team we're users of durable streams and like how can you build systems that kind of are a lot more spread out and distributed and almost like decentralized this is really interesting because you want to be able to run coding engines anywhere you want to be able to run them for a short time for a long time on demand on a schedule all these things and have them all be part of this kind of like brain so I don't know parts of what we're doing really boring like all our data is in postgres and then parts of what we're doing is really interesting but there's a lot of distributed systems problems a lot of infrastructure problems like we are building tools for AI but there's a lot of problems in building collaboration platforms that are really really hard and there's a lot of new tech that makes it easier and more interesting but it's still by far from an easy problem it sounds like what you're saying it's like in for layers it's some extent a new for layers being built and it'll take some time and but it'll be like just new blocks and it will eventually become the primitives like for cloud we have from it already but it took for decades to get those together or more yeah AWS in what like 2006 yeah and then you got Kubernetes a decade later and as closing what's the book or reading that you would recommend something that you personally enjoyed nowadays we talk a lot about refactoring by mornfeller

classic because we spent a lot of time improving the design of existing code and trying to figure out how to get models to build code that is easy to maintain and like easy to read and easy to understand and easy to build on I feel like I probably have a better answer than that but that's what's top of mind these days we're doing a lot of like classics of software engineering refactoring clean code the pragmatic programmer like all that stuff is I think it's more relevant now than it has ever been love it well thanks thanks so much this was fun this was a blast dude thanks for having me on this was great I love it I don't know about you but I really enjoyed this conversation it's actually such a believer in gender coding yet he's the one warning us that if you stop reading the code you have about three to six months before your code base becomes easier to rewrite than to fix and this comes from first-house experience his team built a lights off software factory ran it and then had to shut it down I also like the idea of the slow loop loop engineering feels like a somewhat meaningless term to me what Texas team does is actually pretty boring a cronged job runs every night fixes one issue or one anti-pattern and opens one small pull request the team wakes up to code base that's a little bit better every morning and deaf still needs to review and prove it this is a practice that honestly any injury team purchased adopt today finally I really enjoyed the history

lesson the term software factory comes from natal conference in 1968 the idea of software used to build software with analogy to factory is more than six years old and every generation of our industry has tried to automate more of the loop of building software AI agents are just yet one more attempt although probably the most successful one do check out show notes below for the related the private engineering deep dives that go even deeper into AI engineering and other related topics if you enjoy this podcast please do subscribe on your favorite podcast platform and on youtube special thank you if you also leave a rating on the short thanks and see you in the next one