AI-First Delivery: A Practical Operating Model for AI-Assisted Teams
Year: 2026
Type: preprint
Abstract: AI-assisted development tools can accelerate planning, coding, testing, and review, but without structural discipline they also increase verification cost, review burden, security risk, and architectural drift. This paper presents AI-First Delivery (AIFD), a lightweight operating model for software teams that combines Kanban flow, Shape Up-style appetite constraints, Extreme Programming practices, and AI-assisted acceleration loops. AIFD introduces explicit gates, reusable artifacts, role stewardships, and a learning loop that turns recurring friction into system capability. The framework also identifies five common anti-patterns in AI adoption and proposes a phased roadmap for implementation. The core claim is simple: AI improves delivery only when embedded in a disciplined system; humans retain judgment, while AI accelerates execution, review, and learning.
Author: Buniotto, Matheus
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7162189999
Cites:
Cited by:
Related to: