Part XIII — Expert Mode: Systems, Agents, and Automation

Welcome to Expert Mode

If you have built a few vibe coding projects, you might feel the ceiling. You want the AI to do 10 things in a row, not just one. You want it to follow a 500-line style guide perfectly. You want it to know your company's internal jargon without you pasting it every time.

This section is about moving from "Chat" to "Systems." It's about building pipelines where the AI is a reliable component, not a creative partner.

Advanced Topics

We will cover:

• Agents: How to make the model plan, execute, and correct itself (the loop).
• Expert Prompting: Techniques like Chain-of-Thought, Tree-of-Thought, and recursive prompting.
• Advanced Schemas: Handling recursive JSON, polymorphism, and schema evolution.
• Advanced RAG: Hybrid search, reranking, and "provenance" (knowing exactly where an answer came from).
• Fine-Tuning: When to stop prompting and start training (and why you probably shouldn't).

Where to go next

• 37. Agentic Workflows
• 38. Building a Code-Change Agent
• 39. Prompt Engineering for Experts
• 40. Advanced Structured Output
• 41. Advanced RAG
• 42. Fine-Tuning Decision Framework