Course

Agentic AI

How autonomous, tool-using systems actually work, built directly on Prompt Engineering's context-window mechanics: turning next-token prediction into tool calls, reasoning-action loops, self-directed planning, memory across steps, and the guardrails that keep an agent loop from running away from you.

Lesson 1

What Makes Something "Agentic"

An agent is still just next-token prediction, the only difference is that some of those tokens now trigger real actions instead of getting read by a person.

Start

Lesson 2

Tool Use & Function Calling

The structured-output lesson from Prompt Engineering, except now the JSON the model writes actually gets executed and fed back in.

Start

Lesson 3

ReAct: Reasoning and Acting in a Loop

Chain-of-thought's reasoning trace, except every few steps it stops to check a real result before continuing.

Start

Lesson 4

Task Planning & Decomposition

Prompt chaining had a human decide the steps ahead of time, here the model decides its own steps as it goes.

Start

Lesson 5

Agent Memory Across Steps

The same context window problem from multi-turn conversations, except now the agent is talking to itself and its own tools instead of a person.

Start

Lesson 6

Retrieval as an Agentic Capability

Grounding with retrieved context, but now the agent decides when to search, and whether the first search was actually good enough.

Start

Lesson 7

Guardrails & Failure Modes

What happens when an agent loop goes wrong, and the same tool-permission thinking from AI Security applied on purpose instead of after an incident.

Start