On How AI Agents Are Redefining My Role as a Software Engineer
My primary role is no longer writing code, but orchestrating context and acting as the Chief Editor for my AI agents.
- llms
- agents
- engineering
My day-to-day job as a software engineer is no longer about writing code, but it is still fundamentally about delivering high-quality code. Looking back at my workflow over the past few months, one thing is abundantly clear: hands-on coding has taken a backseat. In practice, my primary role is no longer writing code, but orchestrating context and acting as the Chief Editor for my AI agents.
AI agents have taken over the bulk generation of code—one of the fun parts of being a software engineer—and the truth is, they operate at a truly incredible speed, delivering code in no time flat. While this pulled me away from raw implementation, it brought me into a completely new workflow where most of my time is spent today. Within this role, there are, at a minimum, three core pillars:
- Context Orchestrator: Gathering the right information—design choices, architecture constraints, user requirements—so the model understands the problem and knows how to make decisions.
- Editor: Auditing the quality of the generated code, detecting logic flaws, and suggesting or directly implementing improvements.
- Decision Validator: Ensuring the solution is maintainable and genuinely correct before it hits production.
Given the speed at which AI Agents generate code, I’ve officially become the bottleneck in this new workflow. But for now, I find it absolutely necessary. Keeping the human in the loop is the only way to guarantee that our users get the right experience and actual solutions to their problems.