Am I the baddie?

A software engineer facing a critical deadline crunch at a road construction software company built a revolutionary AI agent system using the latest Claude Opus/Sonnet 4.6 and GPT-5.4 models. Starting with simple worktree setups to enable parallel processing, he evolved the system into a comprehensive automation platform featuring a custom CLI, IDE extension for managing local dev environments, and a dashboard that could pull tickets, write code with proper context from documentation and communications, and even fix review issues automatically. The breakthrough came when he configured the agent to analyze ALL tickets for a major feature, extract design requirements from documentation, and create an implementation plan—transforming what seemed like an impossible deadline into manageable work.

The system proved astonishingly effective during live QA testing, with tickets passing one after another, moving the team from "not going to finish on time" to "mostly done" with only minor bugs. The engineer's workflow became dramatically simplified: push a button for code generation, review, then potentially push another button for fixes. Company directors were "blown away" by demonstrations, leading to discussions about implementing the system company-wide and automating the entire two-button process. While celebrating a likely significant raise and recognition, the engineer confronted ethical dilemmas about profiting from systems that could lead to over-staffing and human displacement, questioning whether accelerating AI automation makes him "the baddie" in a broader societal context.