Skip to content

My First Week with Claude Code🔗

This past week I shifted my focus toward exploring the capabilities of Claude Code. If you are trying to keep up with the breakneck speed of AI releases, you have likely seen the buzz. It is constant, with new services like Claude Cowork and Google’s Antigravity popping up alongside the standard web tools.

Initially, I was hesitant. I had become incredibly comfortable with the Claude web app. I spent several months refining an augmented workflow that allowed me to churn out documentation and release notes faster than I ever thought possible. I had a groove.

But then I watched Alex McFarland’s masterclass on building an AI co-writing system, and everything clicked. He was not just using Claude Code to build apps. He was using it to build a writing system. Seeing how he translated developer tools into an engine for creative content made me realize I could do the exact same for technical documentation.

We have to be honest: AI is slamming industries that revolve around the written word. You can feel it in the technical writing community where there is a palpable sense of fear. But it is not just us. Software engineers and creators across the board are feeling the impact.

If you are looking for a reality check that will chill you to the bone, read Matt Shumer’s latest post, "Something Big Is Happening". He moves past the polite version of AI talk and gets honest. We are in the "this seems overblown" phase of something much bigger than we are prepared for. His cautionary tone is clear: we either learn to engage with curiosity and urgency, or we get left behind. It sounds apocalyptic, but in this industry, it rings true.

Building Brick by Brick🔗

Instead of staying in that fear, I decided to dive in. Over the last few days, I have been adapting the architecture Alex highlighted to fit my specific needs. Rather than following a rigid template, I customized the flow to focus on my own technical writing history.

What has been truly transformative is realizing that the prompts I built in the Claude web app were not wasted. They were just the raw materials. In Claude Code, I am able to use the collection of knowledge base articles I have authored over the past few years as a persistent context. This means the tool actually understands the specific product logic and style I have already established. While you truly have to build this system brick by brick, the trade-off in efficiency is massive.

The Return of the Scientist🔗

This shift has forced me into a systems thinking mindset. I am no longer just a solo writer. I am a system architect. I am thinking about how to create pipelines, employ agents, and operate as a coordinated team of one.

There is not a standard Technical Writing AI course for this yet. We are all just figuring it out in real-time. This experimental phase actually feels familiar. It brings me back to my Biotech days. I feel like a scientist in a lab again, setting up experiments, documenting variables, analyzing results, and optimizing for the next run.

Some parts of my system are working perfectly. Others are "good enough" but will need a rewrite later. But that is the point. I am not just using a tool anymore. I am building a partner that I hope to evolve into something more agentic. The goal is to create a system that can eventually manage the more routine documentation tasks on its own. By automating the mundane, I can keep the human in the middle and focus my energy on the high-value, complex projects that actually move the needle. It is a massive shift in how I work, and I am excited to see where it goes.