Introducing TAIPO, an AI-based product owner assistant for vibe coding

Abstract

. Vibe coding promises rapid progress by letting developers rely heavily on generative AI during implementation. In collaborative and educational settings, however, this speed also creates pressure on requirement traceability, prioritization, and change management. We introduce TAIPO (The AI-based Product Owner), an AI-supported product owner assistant that embeds generative AI into a Kanban workflow instead of treating it as a separate chat tool. TAIPO stores AI-generated clarifications directly on cards, generates prioritized backlog items from requirement text, decomposes stories into implementable tasks, supports refinement and acceptance-related feedback, and can proactively simulate product-owner interventions through contextual comments and unexpected change requests. The prototype is implemented as a web-based system with a PHP backend, a Vue 3 frontend, and Gemini-based AI services. It is grounded in a cleaned version of the TAWOS issue-tracking dataset in order to support more realistic project dynamics. We position TAIPO as an environment for structured vibe coding rather than as a pure code generator