Resources
What we learn building AI systems in production
Architecture, reliability, guardrails — notes grounded in systems that actually run, not in theory.
- Multi-Agent Orchestration with LangGraph: What Production Taught Us Running 7 specialized agents in a loop, with retries and strategic reflection, surfaces problems a demo never shows.
- RAG in Production: Beyond the Prototype A RAG that works in a demo and a RAG that holds up in production often share nothing but the name.
- Structured Outputs and Reflection Loops: Making an LLM Pipeline Reliable How to force a model to produce exactly what the system needs — and what to do when it fails anyway.
- Guardrails, Human Validation and Observability: Securing an AI System in Production An agent that writes SQL queries or publishes on a brand's behalf can't rely on a single point of control.