Like my previous blog post on AI and dog training, I wrote another LinkedIn post about Claude Code, and I keep thinking about it.
I am in the fortunate position of working for a large technology organization with many people across departments leveraging approved AI tools such as Gemini, NotebookLM, Cursor, and Claude Code to assist them with various work-related tasks, and sharing their experiences.
Several co-workers have been talking about Anthropic’s Claude Code.
Orange icon staring at me
Near the end of 2025, I installed and launched Claude Code via the command line interface (CLI). This orange, retro-style graphic stared at me, with some retro-style text. I wasn’t sure what to make of it, what to tell it to do, or generally how to proceed.
At the start of 2026, under a tight release deadline, I leveraged an internal AI tool that could provide an overview, target personas, and user journeys, and Google’s Gemini, to draft a set of docs for each new feature. I was surprised at how well it worked. Each doc set needed revisions, a human touch, and engineering review, all of which are standard processes. That I had drafts to work with in a couple of days instead of a few weeks, was a game changer.
I found myself copy-pasting the same content across tools, which led to a lot of context switching. The internal tool to Gemini, Gemini to VSCode, back to Gemini for style checks. VSCode for updates. The command line to commit. GitHub to create the PR. Back and forth. Back and forth.
I will note that work across 2 products, so my whole day is context switching.
Claude Code with context, and less context switching
I didn’t notice the additional context switching until I installed the Claude Code extension for VSCode.
Suddenly, I didn’t have to leave VSCode. Claude Code opened in a new pane, and delivered responses in the pane. I could ask Claude Code to create skills that would let it complete repetitive tasks more quickly, like style checks, and, if I were feeling adventurous, have it make updates to files for me.
It was a revelation.
Less context switching.
I went from VSCode with Claude Code to the terminal to commit to GitHub, then to GitHub to create the PR, and that was it.
The file structure in VSCode gave Claude Code context for me, and I suddenly saw a wealth of opportunities to enhance my day-to-day work since I basically live between VSCode, the terminal, and GitHub.
Some use Claude Code for everything: updates, commits, pushes, etc. I don’t feel comfortable doing that, but I see how it’d be beneficial for some roles.
What about Claude Code and my own writing?
Using Claude Code for work, seeing what, I imagine, is just a small bit of what it can do, made me wonder: can Claude Code help me with the next version of my manuscript?
My manuscript is a sprawling piece of work. It is structured like a symphony, with movements that take the reader through the concert of events I chose to focus on at the time I wrote it.
Can I build skills so it can review my manuscript as if it were a developmental editor? A line editor? An agent? A publisher? A submissions reviewer? Can I build skills so it can identify the major themes of my manuscript, and pinpoint where those themes are strongest and weakest? Can it tell me where there are plot holes, where there is confusion? Where my manuscript is really strong, and where it needs work?
If I give Claude Code access to my manuscript, how much better would my next version be?
My curiosity is tempered by caution. Using Claude Code at work, where it’s been through the rigors of corporate legal review and specific terms and conditions on the use of corporate data, is one thing. Using Claude Code for my personal use, without the rigors of corporate legal departments, is another.
I’m encouraged by Anthropic’s updates to its Consumer Terms and Conditions, but I remain hesitant.
My curiosity grows stronger each day, and even if I don’t feed it with my manuscript, I can see it as a useful writing aid to plot, track, and refine other stories.