• Architect and build agentic AI systems using Claude Code • Design the AI harness, including: - State management - Context management (preventing context rot) - Tool orchestration - Error recovery strategies - Verification and validation loops - Failure reflection and learning • Define SOPs and system logic before building agents • Apply context engineering (ACE) so AI receives only what it needs — no more, no less • Convert non-binary business problems into binary, measurable questions • Ensure continuous improvement, performance tracking, and cost optimization • Act as the bridge between product vision and technical execution