Historically, software developers showed up to work and used whatever tools their employers provided. The IDE, the frameworks, the deployment pipeline - it was all handed to you. You brought your brain, your experience, and your problem-solving approaches, but the tools themselves were part of the job.
But something’s shifting. In the age of AI, I feel like we’re getting closer to the craftsmen we see in other industries - the ones who show up with their own hammers, wrenches, and specialized tools that they’ve carefully selected and mastered over years.
The Remote Work Foundation
Remote work already started this transition. If I’m working from home, I’m responsible for my own setup - my camera, my microphone, making sure I have high-fidelity communication with my teammates. Maybe the laptop is provided by work, but the quality of my remote presence? That’s on me.
Still, the main tool I brought was my brain. All my prior experiences, the shortcuts and mnemonics, my approaches to solving problems, ideating, collaborating, and executing. That was the toolkit.
The Agentic Shift
But as we move into this agentic world - managing multiple agents together, Ralph loops, critique loops, distilling approaches - something fundamental is changing. Knowledge and problem-solving approaches are becoming programmable and repeatable.
My toolkit is expanding beyond just thoughts in my head. It now includes the approaches and skills I’ve defined or acquired and know how to master. When someone hires me, they don’t just get me and my brain - they get me and my agentic swarm, my approaches to solving problems.
My toolkit exists beyond just thoughts. It’s becoming more tangible, getting closer to being akin to a craftsman’s tools.
It’s Not Just About Having Tools
But here’s the thing: it’s not as simple as someone copying my processes and being as good as I am. Having the tools isn’t enough - it’s about wielding them effectively.
It’s about knowing which tool meets which job. It’s about understanding how to parallelize work and execute effectively in this much more complex world. It’s about the judgment that comes from experience - knowing when to use a Ralph loop versus a critique loop, when to parallelize agents versus running them sequentially, when to trust the output versus when to add another layer of validation.
The craftsman doesn’t just own a hammer… they know when to use it, how hard to swing, and what angle to approach from. That’s what we’re becoming.