Sign InTry a questionToday's repAccept invitations

Sign In Try a question

← All coursesWorking With AI: Context, Tools, Agents and Evals6 levels · 24 lessons

Each request re-sends the whole conversation~9 min
Say the thing you want~8 min
Thinking is a budget you set~8 min
What You Send to the Model · Review~5 min

The context window is billed on every request~9 min
What a JSON schema enforces and what it does not~8 min
System prompts and instruction files: what the model can still ignore~8 min
The Context You Build · Review~5 min

A tool's description is all the model receives about it~9 min
What MCP standardises and what it does not~8 min
An agent is a model calling tools in a loop~9 min
Tool Use: Letting a Model Call Functions · Review~5 min

Whether a local model fits in your memory~9 min
Splitting one job across two models~8 min
Your own files as context~8 min
Running a Model on Your Own Machine · Review~5 min

Giving a model your screen: screenshots or page structure~9 min
Everything an agent reads can act as instructions~8 min
Designing an agent's permission boundary~8 min
Letting a Model Act on Your Computer · Review~5 min

Build an eval set instead of checking by eye~8 min
A model's confidence says nothing about correctness~8 min
Explaining every line of the system you built~8 min
Measuring Whether Your AI System Works · Review~5 min

Course

Working With AI: Context, Tools, Agents and Evals

Model output depends on the context you send, not on how you word the prompt.

Start free: Each request re-sends the whole conversation

The levels

Level I · 4 lessonsWhat You Send to the Model