Lost in AI Coding

I’m down the rabbit hole again.

Around December of last year, something shifted. A new wave of agent-based coding assistants started landing, and they were a fundamentally different beast from what came before. For most of 2025, the buzz had been around “Vibe Coding” — using AI to develop a project, often starting from scratch each time and lamenting the low quality of the code and difficulty fixing and maintaining what was created. It was useful, sure, but it still felt like a souped-up version of the tools we already had.

These new agentic tools are something else entirely. I’ll save the technical deep dive for another post, but the short version is their quality has arrived at the point that they are capable of doing all the low-to-mid complexity tasks I tend to work on. My oversight and prompting still plays a role, but my industry is changing at a rate faster than anything I’ve ever seen. This has triggered a race to keep up. Conference keynotes and Youtube influencers are starting to use “don’t get left behind” rhetoric with every incremental advancement. and it is definitely affecting me.

I’ve been spending my weekends getting lost in it. For the last month, I’ve shipped a near-complete, functional project every single weekend. That’s not a pace I’ve hit since… honestly, maybe ever. Old skills are coming back to life, new patterns are clicking into place, and the way I tinker with side projects has fundamentally changed. The friction that used to kill weekend projects has melted away and I’m in an intense cycle of learning and building skills.

The tool that kicked this off for me was Claude Code. It’s Anthropic’s command-line coding agent, and it’s the one that finally made agentic development feel like a process that would help me both be more effective and improve myself at the same time. The biggest risk I see from using these tools, aside from losing my weekend and enacting an almost complete social isolation, is the risk of outsourcing my understanding. Taking things step by step and understanding these new tools by breaking open black boxes and examining how they work and how I can iterate my use of them has been essential.

Impact of AI Coding Tools visualization

Just recently, the OpenClaw era of this industry shift has arrived. Personal autonomous agents are maturing rapidly and I expect we haven’t seen the final iteration of what all of this is going to turn into. It’s equal parts exciting and disorienting. I’m re-learning how to function in this industry, and I’m not entirely sure where it’s all heading. But I’m having a blast finding out.