Lesson 9: The Whole Thing — A Tour of Agents

Build a complete AI agent in ~60 lines of Python: tools, memory, guardrails, self-scheduling. Everything LangChain, CrewAI, and AutoGen do — no framework needed.

The Whole Thing Nine lessons. Each concept maps to something you've used in ChatGPT or Claude: | Lesson | Concept | You've seen it as... | |--------|---------|---------------------| | 1 | Agent function | Hitting Enter in any chat UI | | 2 | Tools | "Used browser", "Ran code", plugin icons | | 3 | The Loop | Multi-step tool use (search → read → search again) | | 4 | Conversation | Chat history within a session | | 5 | State | "Analyzed 5 files", progress indicators | | 6 | Memory | ChatGPT Memory, Claude Projects | | 7 | Policy | Content refusals, safety filters | | 8 | Self-scheduling | Deep research mode, autonomous sub-tasks | Now they compose into a complete agent framework. ~60 lines. No imports beyond json and pyfetch. > This is the same architecture as LangChain's AgentExecutor + memory + guardrails + task queue. The difference: you can read every line.

Concepts covered

complete agent
integration
no framework

Framework comparison: LangChain, CrewAI, AutoGen: thousands of lines. You need 60.

All lessons in this course

9 / 99. The Whole Thing