AGENT / 04 / Course
AI agents: design, build, and evaluate
Learn when an agent is the right tool, how to give it bounded tools, and how to check that it does its job well.
Who is it for?
For people who already code and want to build agents that use real tools and data without losing control.
What you will practice
- Decide whether a task needs an agent or a fixed workflow
- Connect tools and context with narrow permissions
- Ask for human approval before sensitive actions
- Measure the agent's quality, cost, and failures
4 modules
Proposed curriculum
01What an agent is and when to use one
Compare agents with fixed workflows. Define the task, success, and what the agent must not do.
02Tools and context
Connect tools and data, control what the model sees, and limit its permissions.
03Supervised workflows
Design approval steps, safe retries, and logs that explain each decision.
04Evaluate and operate
Build test cases, measure results and cost, and prepare the agent for real use.
The proposed project
An agent that completes a specific task with tools, human approval on sensitive steps, and an evaluation set.
Before you begin
- Basic programming knowledge
- Some experience with an AI assistant
- A repetitive task you want to automate
Wasyra Learning / Next step
Your next project starts here.
Choose the program that interests you and leave your details. We will contact you with information about the next edition.
This inquiry does not reserve a place or require payment.