Agentic AI: Building AI Agents

Design agents that use tools, plan multi-step work and stay under control.

Comes after: Generative AI and LLMs

1. What Is an AI Agent?Understand what separates an agent from a chatbot and from a fixed workflow, and when an agent is the wrong choice.
2. Tool Use and Function CallingGive a model the ability to act by defining tools, and build the request-and-result exchange that every agent is based on.
3. The Agent LoopBuild the core loop of every agent (reason, act, observe, repeat) and learn how to make it stop.
4. Workflow PatternsLearn five reliable ways to arrange model calls: chaining, routing, parallel work, orchestrator and workers, and evaluator loops.
5. Memory and Context ManagementKeep an agent effective on long tasks by controlling what is in its context window and what it stores outside it.
6. Planning and Multi-Agent SystemsHelp an agent tackle large tasks by planning first, and split work across several agents when one is not enough.
7. The Model Context Protocol (MCP)Connect agents to tools and data through a shared standard, so that one integration works with many applications.
8. Guardrails and Human OversightKeep agents safe by limiting what they can do, checking what they do and keeping people in control of actions that matter.
9. Evaluating and Deploying AgentsMeasure whether an agent completes its tasks, trace what it did when it fails, and run it reliably in production.
10. Project: Design and Build a Support AgentApply the whole course: scope an agent, define its tools, write its loop and instructions, add guardrails and evaluate it.