Principles of Building AI Agents

Author: Sam Bhagwat Genre: Artificial Intelligence

About This Book

Rapid advances in large language models (LLMs) have made new kinds of AI applications—agents—possible. This book focuses on substance over hype, walking through the essential ingredients of reliable agent systems and how to put them together in practice.

Key Insights

• Core building blocks: Providers, models, prompts, tools, and memory are the primitives; design them explicitly rather than burying logic inside ad‑hoc prompts.
• Agentic workflows: Break complex tasks into plans and sub‑tasks; use planners, critics, and executors to improve reliability and clarity.
• Knowledge with RAG: Connect agents to your knowledge bases using retrieval‑augmented generation; treat chunking, indexing, and retrieval quality as first‑class.
• Observability and quality: Instrument agents with tracing and implement lightweight evals to compare configurations, catch regressions, and continually improve.
• Operational discipline: Version prompts and tools, capture inputs/outputs, and prefer deterministic tool contracts to keep behaviour testable.

Why I Recommend It

If you’re building LLM‑powered products, this is a practical guide to moving from prompts to production‑grade agent systems. It emphasises the patterns—tooling, memory, workflows, RAG, tracing, and evals—that shorten the path from prototype to dependable automation.