Ryan demonstrates OpenCode, an open-source CLI coding tool that allows you to use multiple AI models (GPT-5, Gemini, etc.) as an alternative to Claude Code. The discussion covers multimodal capabilities, Language Server Providers (LSPs), session sharing, and enterprise features.

Tara Schofield returns to dive deeper into advanced Claude Code techniques, demonstrating slash commands, subagents for validation, and sophisticated workflows for content creation at Datadog.

Tara Schofield demonstrates how she uses Claude Code and systematic resource management to create security training courses at Datadog, showing how proper setup and context management leads to consistent, high-quality educational content.

Jason Hand explores the concept of technical debt in the context of generative AI development, presenting a simplified visual model to understand how technical debt arises from imbalances between speed, cost, and quality considerations.

Jason Hand introduces his automated retrospective system for capturing learnings during AI-assisted development sessions, demonstrating how to create real-time documentation and team knowledge bases for collaborative projects.

Ryan MacLean shares his experience with test-driven development using Claude Code, demonstrating how to build a Lovable clone with Kubernetes, monitoring, and the importance of PRDs for complex AI-assisted projects.

Jason Hand shares his approach to instrumenting AI-built applications with Datadog RUM, discussing the importance of monitoring fast-iterating projects and managing GitHub repositories for rapid development workflows.

James Eastham demonstrates how to build and deploy remote MCP servers using AWS Lambda, showing how serverless architecture provides the perfect environment for conversational AI interfaces with proper observability and security.

Ryan MacLean demonstrates his workflow using Gemini CLI with MCP servers, showing how to set up sandbox environments and integrate with various AI coding tools for enhanced development productivity.