Hans Reinl ▁

I map my AI-in-production workflow to the software development lifecycle, from discovery to operations, with clear handoffs between AI agents and human owners. The core idea: agentic engineering is not a coding trick but a delivery system for reliable software.

I want to share my workflow for moving from vibe coding to agentic engineering. I will walk you through the process of defining a feature, exploring the UI in v0, integrating the code in Cursor, reviewing and refactoring the code, and testing, merging, and shipping the feature.

With the release of Claude Opus 4.5 and the hype surrounding "engineering-grade" models, I moved beyond frontend generation to test their capabilities as full-stack engineers. I took the three current heavyweights—GPT-5.1-Codex-Max, Gemini 3 Pro, and Claude Opus 4.5—and ran them through a rigorous MVP development cycle to build 'Speakit', a text-to-speech application, to see if benchmark numbers translate to shipping products.