AI Unified Process - AIUP (aiup.dev)

AI Unified Process (AIUP) is a proposed development methodology that flips the traditional, code-centric model: requirements remain the primary artifact and AI is used as a “consistency engine” to generate specifications, code, tests and documentation from them. Rather than expecting perfect, deterministic specifications for AI code generation, AIUP emphasizes short, cross-disciplinary iterations where requirements, code and tests evolve together. The approach promises to stop documentation drift, keep business intent aligned with implementation, and make modernization possible by keeping traceability from business requirement to code line. Technically, AIUP relies on test-driven consistency and iterative regeneration: automated tests assert desired behavior so code can be safely regenerated or refactored without changing outcomes, while living documentation and stakeholder reviews keep artifacts synchronized. Each phase runs many short cycles with all disciplines working in parallel (not sequentially), improving spec clarity and AI generation quality over time. The model adapts principles from the Rational Unified Process but modernizes them for AI tooling, claiming measurable gains in delivery, maintainability and business alignment by treating tests and requirements as the stabilizing contracts that enable safe AI-driven code evolution.