Welcome to AI Agents 101

Everyone talks about AI agents. Almost nobody can define one precisely. Here's the precise version. A prompt is one shot: ask, get answer, done. An agent is a loop: the model thinks, decides to use a tool, calls the tool, reads the result, decides what to do next, and keeps going until the task is complete. The agent has memory. It has tools. It makes decisions about what to do next. That loop is what makes agents powerful. And it's also what makes them tricky to build well. Most agent demos work for the first few steps and then go off the rails. Production agents need real engineering: error handling, observability, cost control, and patterns to keep the loop on track. This course gets you from zero to a working agent. We'll cover: • The agent mental model: thought, action, observation, repeat • Agent vs chatbot vs LLM: the differences that actually matter • The 4 components every agent has: model, tools, memory, planning • ReAct: the reasoning + acting pattern that powers most production agents • Building your first agent from scratch (50 lines of Python) • Adding tools: web search, calculators, APIs • Using a framework: when to use LangGraph instead of writing from scratch • From demo to production: what changes (and why it's harder than it looks) Some Python helps for steps 7 to 10, but you can read along even without coding. The conceptual frame in steps 1 to 6 is the most important part. Time: ~3 hours across 10 articles.