Skip to content
← Knowledge library

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

0 of 4 completeSaved on this device
  1. What Is Agentic AI?

    Recognize the agent loop and the surrounding control system.

  2. Agents vs Workflows

    Choose the least autonomy needed for the task.

  3. Tools and Function Calling

    Understand how models request bounded actions.

  4. 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

0 of 5 completeSaved on this device
  1. Build Your First Tool-Using Agent

    Implement a small agent loop around one narrow goal.

  2. Designing Tools Agents Can Use

    Create narrow, typed, retry-safe tool contracts.

  3. Structured Outputs

    Separate probabilistic decisions from deterministic code.

  4. Human-in-the-Loop Patterns

    Place approvals where consequences justify them.

  5. 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

0 of 5 completeSaved on this device
  1. Agent Tracing and Observability

    Make each consequential decision inspectable.

  2. Prompt Injection and Agent Security

    Treat external content as untrusted input.

  3. Permissions and Sandboxing

    Limit what compromised behavior can affect.

  4. Agent Governance Framework

    Tie risk tiers to controls and review cadence.

  5. Production Agent Checklist

    Run a structured readiness review before launch.