Stage 01 · Understand
Learn in the order that decisions happen.
Three focused paths turn the library into a practical curriculum. Mark progress as you go; completion stays in your browser.
Path 01
Agent systems foundations
Build a precise mental model before choosing frameworks or architecture.
Best for: Product leaders, designers, and new builders
- What Is Agentic AI?
Recognize the agent loop and the surrounding control system.
- Agents vs Workflows
Choose the least autonomy needed for the task.
- Tools and Function Calling
Understand how models request bounded actions.
- Levels of Agent Autonomy
Set an appropriate supervision boundary.
Path 02
Build a reliable first agent
Move from one verifiable job to a controlled tool-using system.
Best for: Engineers and technical product teams
- Build Your First Tool-Using Agent
Implement a small agent loop around one narrow goal.
- Designing Tools Agents Can Use
Create narrow, typed, retry-safe tool contracts.
- Structured Outputs
Separate probabilistic decisions from deterministic code.
- Human-in-the-Loop Patterns
Place approvals where consequences justify them.
- How to Evaluate AI Agents
Define evidence that the system succeeds safely.
Path 03
Production operations and governance
Establish evidence, visibility, security, and accountable ownership.
Best for: Platform, security, operations, and risk teams
- Agent Tracing and Observability
Make each consequential decision inspectable.
- Prompt Injection and Agent Security
Treat external content as untrusted input.
- Permissions and Sandboxing
Limit what compromised behavior can affect.
- Agent Governance Framework
Tie risk tiers to controls and review cadence.
- Production Agent Checklist
Run a structured readiness review before launch.