Ships 100x Faster with Agentic Engineering

• Channel: David Ondrej
• Video: Watch on YouTube
• Tags: dev ships 100x faster agentic engineering

Overview

David Ondrej interviews Rod Mickey — a senior dev who uses AI for 95% of his code — about his exact agentic engineering stack, tools, models, and loops.

Key Takeaways

• Agentic engineering as the future. in 2026, the people who ship 100x faster aren’t typing prompts into a chatbot. they’re running multiple agent harnesses in parallel. so i had to make you on the podcast, a senior developer who has
• Quite changed since the last time we talked. it’s no longer vibes. like we got to be serious with this stuff. and at the same time, i’ll be honest, like i think i would say in the last three months,
• Code for fun. i’ll do it every now and then in the weekends. but like, i think it could be foolish to not see where we’re going. yeah, the models are not perfect. they’re at a point where like, there
• I use in my workflow. and i think anybody else can do this. it’s super replicatable. i’m actually new to the exact tools. and it’s also going to feel like her path is like auto research loop. so i think
• Funny. i’m using opus 4.7 max fast instead of cursor. but also instead of cursor asia. here’s actually a big difference. a lot of people don’t know. there have been benchmarks, right? and like these are like benchmarks done by
• Still a good model. especially when it comes to ui changes. so i will preface there whenever i do anything ui or front end, uh, why’s i will use opus 4.7. but i’m using it max. i have no time
• Their new agentic view is pretty pretty nice. so i guess let’s touch on the hardest, right? because a lot of people say word that’s becoming more popular now in in the last two three months. obviously it’s been around
• Or not, the model itself can’t do anything. the model is just our predictor of next text. technically the model doesn’t even think. whatever english you give it, it converts it to tokens. those tokens are mapped on a graph.
• The model doesn’t do the model just predicts the next text. what a harness does, you can think of a harness as a wrapper of a model. and this wrapper is api tools, a specific system prompt, markdown files like
• The video. and i will email you all the skills, all the repels, all the tools mentioned in this podcast. again, it’s completely for free. so go below the video and grab it now. now the main important thing with
• This file, it searched this thing, right? those are tool calls that it’s making. again, the agent doesn’t have the ability to do so, but the harness gives it the tool so that the agent can do this. the models
• Long as people use the best model available. and that’s the thing, right? the model does matter. i will gbt5.5 extra high, especially i work mainly on large code bases. so i may be in that biostate, it’s just really
• Source. now it sounds like the term open source, but it’s actually a repo. if you go on google, it’s actually by versil shout out to versil for open source. this basically what open source does is it fetches the
• Matters is i have this great up folder here that says open source. a folder called repo is if i click on that, there’s a folder called github. where am i? you see a bunch of popular projects or companies
• Agent the best source of truth, which is the code. because the code is the single best source of truth. what i’ve done is in an agent.md. i’m not a big fan of agent.md files. unless it’s something that the

Transcript

agentic engineering as the future. In 2026, the people who ship 100x faster aren’t typing prompts into a chatbot. They’re running multiple agent harnesses in parallel. So I had to make you on the podcast, a senior developer who has AI right 95% of his code to walk through his exact AI stack. The tools, the models, the loops. This is the David Andre podcast. Enjoy. Rod Mickey, how do you think about building with AI in 2026? Yeah, it’s quite changed since the last time we talked. It’s no longer vibes. Like we got to be serious with this stuff. And at the same time, I’ll be honest, like I think I would say in the last three months, all the code that I’ve created, I would say 95% is generated by AI. And you’re an actual developer, right? So like, like, I’m an actual engineer and developer, like, and like, there’s a part of me that misses writing code for fun. I’ll do it every now and then in the weekends. But like, I think it could be foolish to not see where we’re going. Yeah, the models are not perfect. They’re at a point where like, there are productivity gains, especially if you understand the vertical that you’re in. 90% of the applications people are building. A little bit of brains and AI will take you a long way. So I’m going to share like the tools I use in my workflow. And I think anybody else can do this. It’s super replicatable. I’m actually new to the exact tools. And it’s also going to feel like her path is like auto research loop. So I think it’ll be fun. Awesome. Let’s jump into it. First is harness. I chase the model, right? I think the model is the more important thing but I use cursor and right now I’m using GPT 5.5. Interesting. Okay. So that’s funny. I’m using Opus 4.7 Max fast instead of cursor. But also instead of cursor Asia. Here’s actually a big difference. A lot of people don’t know. There have been benchmarks, right? And like these are like benchmarks done by like actual developers where cursor performs both cloud code and codex for their models, right? Yeah. A lot of people have been dunking on Opus 4.5 singing Opus 4.5 is terrible. Some people on turn call it slow piss. It’s still a good model. Especially when it comes to UI changes. So I will preface there whenever I do anything UI or front end, uh, why’s I will use Opus 4.7. But I’m using it Max. I have no time for any other variant. But the harness matters. Now a lot of people are priced out of cursor, cursor doesn’t subsidize the way it codex and cloud does. cursor, my opinion, is the best harness. I can switch between models. Their new agentic view is pretty pretty nice. So I guess let’s touch on the hardest, right? Because a lot of people say word that’s becoming more popular now in in the last two three months. Obviously it’s been around for a long group of a lot of people. It’s their first time hearing this really and they don’t really understand it fully. So you have the model. What is the hardest, right? It’s everything around it. So believe it or not, the model itself can’t do anything. The model is just our predictor of next text. Technically the model doesn’t even think. Whatever English you give it, it converts it to tokens. Those tokens are mapped on a graph. And then what it does is it looks at that graph and it predicts the next token. Some cool mathematics happens there. And then it returns a token to you, which is converted to English. So the model doesn’t think the model doesn’t do the model just predicts the next text. What a harness does, you can think of a harness as a wrapper of a model. And this wrapper is API tools, a specific system prompt, markdown files like agent md files. These are all things that wrap the model that guide it to perform or specialize in a specific action. Real quick, if you want everything mentioned in this video, completely for free, click the first link below the video. And I will email you all the skills, all the repels, all the tools mentioned in this podcast. Again, it’s completely for free. So go below the video and grab it now. Now the main important thing with the curses and the claw core in the cortex is the tools it gives it. Whenever you use cursor or any sort of agent you notice in the trace and the response it gives you it says, oh, it read this file, it searched this thing, right? Those are tool calls that it’s making. Again, the agent doesn’t have the ability to do so, but the harness gives it the tool so that the agent can do this. The models have got to a point where an investment in a really good hardness will maximize the output you get from the model. And we know this because people’s experience with claw code and cursor is not the same. Yeah, as long as people use the best model available. And that’s the thing, right? The model does matter. I will GBT5.5 extra high, especially I work mainly on large code bases. So I may be in that biostate, it’s just really good at understanding large code bases, complex code bases. And it’s been two weeks and other than UI changes, this is the only model I’ve been using. There are some tools that I highly recommend. The first one is open source. Now it sounds like the term open source, but it’s actually a repo. If you go on Google, it’s actually by versil shout out to versil for open source. This basically what open source does is it fetches the source code of whatever package you’re working on and dumps it into the code base. And I’ll give a specific example. This is a pretty large code base of an app on working on. The structure doesn’t really matter. What matters is I have this great up folder here that says open source. A folder called repo is if I click on that, there’s a folder called GitHub. Where am I? You see a bunch of popular projects or companies you made of herd of browser use, Composio, Daytona, OpenClaw. A lot of the technology is actually open source, which is fantastic for agents, right? Instead of dumping in your docks or whatever that are manmade, you can give the agent the best source of truth, which is the code. Because the code is the single best source of truth. What I’ve done is in an agent.md. I’m not a big fan of agent.md files. Unless it’s something that the agent would have known by had, for example, like people will put, oh, this is a React code base in the agent.md file. And I believe there’s sort of been a good idea if we were using Sonic, fourth, you know, fourth, four, four, five, or like opus, four, oh, but again, the models have gone so good that it will just read your code base and know exactly what the text that is and stuff like. So if you notice, the harnesses are sort of getting lighter and lighter and thinner and thinner. This is why pie has been winning so much. You actually don’t need much, like the models are really, really good. So you just want to tell it to stuff that’s not obvious. But I like how this project will be used was the vision behind it. Exactly. If there’s a certain structure that I have and stuff like that. Now, in this case, I tell it that it can fetch for additional source code. And by the way, all this is an agent, right? I don’t handwrite this, right? So if anyone reads this, it’s like, oh, so smart. I wish, AI generate this. It’s like, I think, and I tell the AI what to do. And literally, there’s a small block that says to fetch source code for a package or repo, you need to understand and run. So I tell it basically how to add packages. And the best way to add package, I can like maybe show an example. Actually, we’ll just do open source itself. So let’s say I’m building a product that uses this package or this open source or you’re using something like browser use. You find the repo and then you can literally go to terminal and then I can do MPX open source and then just paste the repo and then watch this. Okay, this field, why did this field usage? Oh, because I’m on a Linux machine. Okay. Don’t mind that. But literally all I have to do is just run this command and what’s going to happen and then it will pop up here. Now, here’s the cool part. David, whenever I want to build a feature that uses set technology, I will tell it in the prompt reference the code base, right? So I’ll go back to my drawing board here. Every time I prompt, let’s, like, for this project I’m using Daytona a lot, which is a really powerful sandboxing platform. There’s a lot of cool stuff you can do. It’s very, very technical. It’s very, very deep. I can go in and spend some time, but I want to build fast. I want to ship fast, right? My competitors are shipping at light speed. So in my prompt, I will tag the folder and I will say reference the code base. Is this like almost a death of documentation? Basically, right? Because code is the best source of truth. Now, there are people that might be like, oh, but isn’t this going to inflate the context window? You see, in the old days, the harnesses used to index the entire code base and they would use vector and ragon all that stuff. Now, the models are so good at search. All the needs is a search tool. So you just need to guide it where to search to, right? And it’s going to reference the code base. It’s going to find the exact function. No guessing. It’s going to find the exact function. And I can’t even tell you, like, eight out of 10 times, like using this, like, I’m getting it spawn on. Yeah, I think the point is that, like, if you do your job properly in the context engineering, do you actually save so much time on testing and debugging? Because then you don’t have any errors. Yeah, context engineering is, like, it is, it’s so important, right? Like, if this is your 227k context window, your agent is smart, probably up to like this level. Maybe I’m being a little mean. Maybe you could go a little further. But like, the more you blow it, the number it’s going to get. So you want to stay in a sweet spot like this, right? So being able to give it the exact snippet it needs or explaining exactly what you need. And this is why a genetic engineering will yield better result than vibe coding, because in vibe coding, you’re offshoring the thinking to the agent, right? Yeah. In agent to engineer, you’re doing the thinking and you’re just letting your minions do the work. You’re letting a bunch of junior grads. We’re very cracked, but need a lot of guidance to do the work. So if you have like a 272k context model, I love codecs because I’ll tell you right here. Like, at 77%, unless I’m done here, this is like, I just need to start a new thread. Like, this is too much information for the

(Transcript truncated — full length available on YouTube)