Introduction¶
Unsurprisingly, this book is about three things:
Development
Agentic AI
Opik
What is Agentic AI?¶
Agentic AI represents a paradigm shift in artificial intelligence, moving beyond simple prompt-response interactions to systems that can autonomously reason, plan, and execute complex tasks. Unlike traditional AI models that respond to individual queries, agentic systems maintain context, make decisions, and use tools to accomplish multi-step objectives.
An agentic AI system is characterized by its ability to:
Reason about problems and break them down into manageable steps
Plan sequences of actions to achieve goals
Execute actions using available tools and resources
Observe the results of its actions and adapt accordingly
Iterate until the objective is met or a stopping condition is reached
Why Opik?¶
Opik is a powerful framework designed specifically for developing, monitoring, and improving agentic AI systems. Built on the Model Context Protocol (MCP), Opik provides the infrastructure needed to:
Track every interaction, decision, and outcome in your agentic systems
Debug complex agent behaviors through detailed traces and spans
Evaluate agent performance across different scenarios and datasets
Optimize prompts, tools, and system configurations based on real-world usage
Experiment with different approaches and compare results systematically
Who This Book Is For¶
This book is designed for developers, researchers, and practitioners who want to:
Build production-ready agentic AI systems
Understand the core concepts and architecture of agentic systems
Learn how to use Opik for development, monitoring, and optimization
Implement best practices for agentic AI development
Debug and improve existing agentic systems
What You’ll Learn¶
In this book, we’ll cover:
Chapter 1: Core Concepts - We’ll explore the fundamental building blocks of agentic AI systems, including LLMs, traces, spans, threads, tools, metrics, datasets, experiments, and the various types of prompts that guide agent behavior.
Chapter 2: Agentic Systems - We’ll dive into building complete agentic systems from scratch, learning how to combine the core concepts into working solutions that can tackle real-world problems.
Reference - Practical guides for self-hosting Opik and running online evaluations.
Getting Started¶
To follow along with the examples in this book, you’ll need:
pip install opik ez-mcp-toolboxThroughout this book, we’ll use practical examples and code snippets to illustrate concepts. Each chapter builds upon the previous one, so we recommend reading sequentially if you’re new to agentic AI development.
Let’s begin our journey into developing agentic AI systems with Opik!