Rasmic rejects AI plan mode in favor of a structured pre-build planning process he calls Agentic Engineering, arguing that letting the AI plan for you produces worse features with less predictable output.
The core method centers on the developer doing the architectural thinking before the agent touches code: defining scope, specifying constraints, and breaking work into discrete, reviewable chunks. The reasoning behind each step is what makes this worth reading in full, not just the final workflow checklist.
If AI-assisted development is part of your stack, this reframes where your cognitive effort should actually go. The shift is from prompting to planning, and the distinction has real consequences for code quality and iteration speed.
[WATCH ON YOUTUBE →]