In the age of AI, developers are increasingly leveraging agentic AI experiences to build their products. The boundaries between software engineering disciplines have become thinner than ever, enabling teams to move from idea to production rapidly. With the release of Kiro, developers can now build and edit their code as a spec, interact with the in-IDE assistant, and delegate complex tasks with confidence. In this talk, we'll explore how to maximize AI-guided development with Kiro and transform your development workflow.

Devoxx

A conference talk by two AWS solution architects introducing Kiro, an AI coding IDE built around spec-driven development. The talk contrasts vibe coding (unstructured, conversational AI prompting) with spec-driven development, a more structured methodology that uses persistent documentation—requirements, design, and task specs—to guide AI agents. The presenters demo Kiro on an existing React/Next.js and Java Spring Boot marketplace app, showing how to generate user stories from a GitHub issue, create architecture design documents, integrate Figma designs via MCP, and break work into executable tasks. Key Kiro features highlighted include steering documents for project context, agent hooks for automated compliance checks, MCP server integrations (Figma, GitHub, AWS docs), and 'Powers' which bundle MCP configs and hooks into reusable packages. The session also covers collaborative workflows where team members share spec files in a repo so each developer's Kiro instance retains shared context.

Rethinking Full Stack Application Development with Kiro by Alessandra Pasini and Romain Jourdan