Module 01: Agentic Foundations

Before you can build production AI agents, you need to understand what they actually are - not the marketing version, but the engineering reality. This module builds that foundation.

---

Module Map

---

Lessons

#	Title	What You'll Learn
01	What are AI Agents?	Precise definitions, the 5 key properties, taxonomy of agent types, why agents became possible with LLMs, real examples from production
02	The Observe-Think-Act Loop	The core execution cycle in depth - observation, reasoning, action, termination, error handling, token management
03	Tool Use and Function Calling	How function calling works under the hood, tool design principles, parallel calls, error handling patterns
04	The ReAct Pattern	The 2022 Princeton/Google paper that changed everything, interleaved reasoning and acting, full Python implementation
05	Agent vs Chatbot vs Workflow	Precise technical criteria for each, the spectrum, when to use each, cost/reliability tradeoffs
06	Agentic Design Patterns	The 5 core patterns from Anthropic's research: prompt chaining, routing, parallelization, orchestrator-subagents, evaluator-optimizer
07	When to Use Agents	The honest answer - 3 conditions that justify agents, 5 that don't, cost/reliability analysis, decision framework

---

Key Concepts

• Agent: a system that perceives its environment, reasons about it, and takes actions autonomously to achieve a goal
• OTA Loop: the Observe-Think-Act cycle that drives all agent execution
• Tool use: the mechanism by which agents interact with external systems (APIs, databases, code interpreters)
• ReAct: Reasoning + Acting interleaved - the pattern that prevents hallucination by grounding thought in real observations
• Trajectory: the full sequence of (observation, thought, action) triples across an agent's run
• Compound error: the reliability problem where each step's error probability multiplies across a long trajectory

---

Prerequisites

• You should be comfortable with Python (functions, classes, async/await)
• You should have used an LLM API before (Anthropic, OpenAI, or similar)
• You should understand what a REST API is and how JSON works
• No prior knowledge of agents, LangChain, or frameworks required

---

What You'll Build

By the end of this module you will have built - from scratch, with no frameworks:

1. A minimal AI agent using the Anthropic API with a full tool loop
2. A complete Observe-Think-Act implementation with error handling, backtracking, and max-iteration limits
3. A ReAct agent that grounds its reasoning in real tool observations
4. Code examples of all three architectures (chatbot, workflow, agent) for the same task
5. Implementations of all 5 Anthropic agentic design patterns

Everything runs with pip install anthropic. No LangChain, no LlamaIndex, no frameworks. You understand the fundamentals first.