# ai

I just refactored my whole website into a completely new design without writing any code myself. Looks pretty good, right? Would you believe me if I told you I one-shot this whole thing in one day using my AI orchestrated workflow? No? Good, you shouldn't because I didn't. It's hard to believe given all the AI hype and hyperbole spreading on social media that entire applications are being one-shot by agents that are left to run on their own for hours or days. Let me show you how the process goes in reality for any non-trivial project.

If you've been paying attention to the agentic coding space lately, you've probably come across the concept of a "ralph loop." The idea, popularized by Matt Pocock and others, is straightforward: put your AI coding agent in a bash loop and let it churn through work autonomously. You write a prompt file that tells the agent how to read its state from disk (a task list, a progress file, specs), pick the next piece of work, implement it, commit, and exit. Then the loop restarts, the agent reads the updated state, and does it again.

Before AI came along, my coding workflow was something like this: open up IntelliJ, checkout a branch, start coding. Sometimes I would switch between branches to review someone else's code, or maybe just to tackle a different problem for a bit. On rare occasions I would use git worktrees to checkout multiple versions of the code, usually to solve a regression. Occasionally if I needed help with something I would look things up on Google and Stack Overflow. Maybe I would pair-code with a coworker. Today, I'm working in a secure container that has several worktrees checked out at any given moment, each with their own associated terminals, IDEs, and AI agents.

If you're using agentic AI today, there's a good chance you're using Claude Code; and if you're using Claude Code, there's a good chance you've heard of, or maybe even used, subagents. If you're like me though, when you first heard of subagents you may have been confused about how they differ from the custom slash commands you may already be using, or if they're meant for something entirely different, like managing concurrent workloads. So let's talk a bit about what they are and what problem they are solving.