A Tech Lead has developed a structured AI-assisted development workflow that emphasizes thinking before coding. The approach addresses a common problem: while AI can accelerate implementation, it often produces code that works technically but lacks clarity and maintainability. The solution focuses on forcing deliberate thinking before any code is written.

The workflow begins with a free-form plan where the developer describes the problem, solution, constraints, and uncertainties in plain language. This plan then feeds into a structured Product Requirements Document (PRD) process that interviews the developer about every aspect of the design, catching bad ideas by forcing specific answers to questions about edge cases and dependencies. The PRD produces user stories that become the backbone for breaking down work into issues.

Issues are created using vertical slices that cut through every integration layer end-to-end rather than horizontal slices of a single layer. Each issue is classified as either AFK (AI can implement without human interaction) or HITL (human decision required). Tasks are then broken down into concrete, ordered steps with explicit outputs and dependency order. The workflow includes structured code review with six passes and a final cross-cutting audit to catch systemic issues across the full implementation.