I enjoyed this article for two different reasons:
- It emphasized the ownership we need to have over the results of our code, whether or not it's written by an agent.
- The focus on working code as the deliverable reflects my changing view on what I'm paid to do: I loved hand-written, artisanal code and TDD to get there, but if I can get to working code in fewer iterations and then focus on refactoring confirmed-working code, then I can hopefully deliver that value more quickly. Getting the code refactored to be readable, clean, and complexity-avoiding is still the current friction I'm facing — but reducing this friction ensures I'm truly evaluating the code and working to make sure it communicates the context and the why behind it.
This focus on delivering value rather than perfect, uniquely-me code started before I began coding with agents. It started when I heard an interview about viewing code reviews not as criticism of your solution, but as opportunities to get more data on what the right solution actually is.