How AI changed what I can build

Why it matters

A Verkada engineer traces a personal arc through consumer AI tools, from image generation with Midjourney and Stable Diffusion to using ChatGPT as a Google substitute, and finally to Claude, where Opus 4 and Claude Code enabled full-stack projects he had never managed before. The account matters because it grounds abstract claims about coding agents in concrete experience: he describes delegating work to an agent, evaluating its output, and catching cases where it solves the wrong problem. He argues this shifts engineers toward management-like skills even without direct reports, since they remain responsible for work they did not write by hand. He also predicts individual scope will expand, with support and operations staff able to build tools themselves rather than waiting in engineering queues. The piece closes on personal software, noting that cheap first versions make narrowly tailored tools viable. For readers, it offers a practitioner's view of how agent-based coding changes what one person can build and what skills matter, without claiming certainty about broader employment effects.