Insights from My Experience with AI Coding Agents and Burnout

Rowan Ashford

ago 2 hours

3D printing often evokes a sense of wonder. Users experience joy as they create objects they could not have built manually. However, the results may not be refined for mass production. Similarly, today’s AI coding agents, while innovative, require more than just basic interactions.

Insights from My Experience with AI Coding Agents

Since November, I have engaged in extensive experiments with AI-assisted software development. Using Claude Code and Claude Opus 4.5 through my personal Claude Max account, I explored their capabilities. Additionally, I utilized OpenAI’s Codex, although less frequently. After fifty projects, the excitement I felt while programming has not diminished since my childhood. I first learned BASIC on my Apple II Plus at just nine years old.

Understanding My Coding Journey

My journey in programming began in 1990. I have experimented with several languages, including:

BASIC
C
Visual Basic
PHP
ASP
Perl
Python
Ruby

While I am not an expert, I’ve developed enough skill to create small tools and modify existing code for personal use. My hobby projects over the years include games built using BASIC, Torque Game Engine, and Godot. This background has enabled me to grasp modular programming architecture.

Creating with Claude Code

In December, I leveraged Claude Code to develop a multiplayer online game inspired by Katamari Damacy, which I named “Christmas Roll-Up.” This experience highlighted the impressive capabilities of AI coding agents.

Limitations of AI Coding Agents

AI tools like Claude Code, Codex, and Google’s Gemini CLI showcase remarkable potential. They can generate prototypes for applications, user interfaces, and even games. However, their output heavily relies on patterns found in training data. Much like 3D printing, the journey from prototype to production-level code demands significant effort and expertise.

The Need for Human Skills

Creating durable and complex software still requires a profound understanding of programming principles. Experienced developers bring patience and skill, essential for overcoming challenges that AI cannot address alone. While AI coding agents can assist, they are not replacements for human creativity and expertise.

In conclusion, my experiences with AI coding agents have been enlightening. They open new avenues for software development but highlight the importance of human skill and knowledge in creating lasting solutions. For those venturing into AI-assisted programming, a solid foundation in coding is invaluable.