AWS re:Invent 2025 - Epic journey: Navigating the AI-powered SDLC from backlog to production(DVT208)

Summary of AWS re:Invent 2025 Presentation: "Epic journey: Navigating the AI-powered SDLC from backlog to production"

The Software Development Challenge

• Software development teams often struggle with more backlog than they can deliver, leading to painful prioritization decisions
• Traditional solutions like increasing headcount or ruthless prioritization have not solved the underlying problem

The Shift to AI-Powered Software Development

• The presentation outlines the evolution from human-centric to AI-native software development:
  1. Human-Centric: Developers as the center, with AI as an assistant tool
  2. Bolt-On AI: AI plugged into various development tasks to boost productivity
  3. AI-Native: AI as a central collaborator, orchestrating the development process

Maturity Model for AI-Driven Development

• The presenters introduce a maturity model for organizations to assess their AI-driven development capabilities:
  1. Software Development with AI: Basic access to AI tools, used on a best-effort basis
  2. AI-Assisted Development: Providing templates, access to MCP servers, and some dynamic agents
  3. AI-Driven Development: Integrating the "system knowledge plane" and AI agents, transforming development processes
  4. AI-Managed Development: AI orchestrating the entire development lifecycle, with humans providing oversight and validation

Ericsson's Journey to AI-Native Software Development

• Ericsson has been on a 12-month journey to implement AI-native software development
• Key elements:
  • Leveraging the Kiro CLI tool, which provides AI capabilities to assist developers
  • Integrating a "system knowledge plane" to provide contextual information to the AI agents
  • Implementing a maturity model and governance approach to drive organization-wide change
• Benefits seen so far:
  • Reducing weeks-long tasks (e.g., software specifications, testing, troubleshooting) to hours or days
  • 1,500+ engineers using the AI-native approach

The Future of AI-Driven Development with Kiro

• Kiro CLI provides flexibility in how developers can work:
  • Structured, spec-based development with AI guidance
  • Conversational "vibe coding" with AI collaboration
  • Automated workflows through "agent hooks"
• Kiro adapts to the user's preferred workflows and development practices
• Custom agent profiles allow teams to configure Kiro to their specific needs
• Key principles for effective AI-driven development:
  1. Provide context upfront to improve accuracy
  2. Start with specifications for production code
  3. Iterate and verify the AI-generated solutions
  4. Use Kiro for the right tasks, maintaining human judgment

Business Impact and Future Outlook

• AI-driven development is transforming the software development lifecycle, moving from human-centric to AI-native approaches
• The goal is to enable 10x thinking and creativity, not just incremental performance improvements
• Kiro provides a technology foundation to enable this paradigm shift, with the flexibility to adapt to unique organizational needs
• The future vision is for AI-managed development, where AI proactively orchestrates the entire lifecycle based on performance and user data