Stage 1 · Lesson 6 of 6
The parts of an agent system
A working system has five parts. The model makes decisions. Tools let it act.
Memory and state carry what it knows. The is your code that runs the loop, executes tools, trims context and enforces permissions and limits. The environment is the world it acts on, such as a database, a codebase or a customer's account.
Frameworks and mostly provide the harness; you still own the tools, the state design, the limits and the evals. When an agent misbehaves, ask which part is at fault before changing the prompt: a vague tool, missing context or an absent limit is far more common than a bad model.
Common trapswapping models or rewriting prompts to fix what is really a tool or harness problem.
A question to check yourself, an AI practice session and 2 cheat sheet entries come with the full roadmap. Unlock
Resources for this stage
Start with· follow one of these
- CoursePaidGrokking Modern AI Fundamentals
Design Gurus · Beginner · $72 one-time
Chapter 8, Agentic AI, covers reasoning, planning and memory in agents and ends with a small agent project
- CourseFreeHugging Face AI Agents Course
Hugging Face · Beginner
Free: Unit 1 explains the thought, action, observation cycle and builds a first agent step by step