AI Agents Mastery

Learn what an AI agent is, how the agent loop works, and build your first scheduled agents.

1. What an AI Agent Actually Is
2. The Agent Loop: Reason, Act, Observe, Repeat
3. The Model Inside the Agent: Tokens, Context and Reasoning
4. Tools: How Agents Touch the World
5. Memory: Short-Term Context and Long-Term Knowledge
6. Instructions as the Agent's Job Description
7. Scheduled Tasks: The Simplest Agent You Can Build Today
8. ChatGPT Scheduled Tasks in Practice
9. Gemini Scheduled Actions and Gemini Spark
10. Claude Scheduled Tasks and Routines
11. Workflows Versus Agents: Knowing Which You Need
12. The Autonomy Spectrum: From Assistant to Autonomous Agent
13. Why Agents Fail: Hallucination, Drift and Compounding Errors
14. Data, Privacy and What Never to Delegate
15. Level 1 Capstone: Your First Three Scheduled Agents

Build custom assistants in ChatGPT, Gemini and Claude with instructions, knowledge and skills.

1. From Prompt to Assistant: What Changes When You Package Instructions
2. The Assistant Landscape in Late 2026: GPTs, Gems, Projects, Skills and Plugins
3. Writing Assistant Instructions That Hold Up
4. Knowledge Files: Grounding an Assistant in Your Material
5. Building a Custom GPT
6. Custom GPT Actions and Apps
7. Moving From Custom GPTs to ChatGPT Skills and Plugins
8. Building a Gemini Gem
9. Gemini Skills: The Successor to Gems
10. Claude Projects
11. Claude Skills and the Agent Skills Format
12. Testing an Assistant Before You Share It
13. Sharing, Permissions and Versioning
14. Choosing the Right Platform for Each Assistant
15. Level 2 Capstone: Build One Assistant on Three Platforms

Connect agents to tools and apps, and add the guardrails and approvals that keep them safe.

1. Function Calling: How a Model Asks for a Tool
2. Designing Good Tools: Names, Descriptions and Schemas
3. Model Context Protocol: The Universal Connector
4. Connecting MCP Servers to Your AI Tools
5. Connectors and Apps
6. Automation Platforms as Agent Infrastructure
7. Building an Agent in n8n
8. AI by Zapier and Zapier MCP
9. Make AI Agents in Scenarios
10. Triggers and Events
11. Guardrails
12. Human-in-the-Loop Approval Design
13. Prompt Injection and Untrusted Content
14. Logging, Errors and Recovery
15. Level 3 Capstone: A Tool-Using Agent With an Approval Gate

Design multi-agent systems: orchestrators, subagents, handoffs and agent teams.

1. When One Agent Is Not Enough
2. The Five Workflow Patterns
3. Orchestrator and Worker Architecture
4. Handoffs
5. Shared State, Memory and Artifacts
6. Agents Building Agents
7. Subagents and Context Isolation
8. A2A: Agents Talking to Agents Across Organizations
9. The OpenAI Agents SDK
10. The Claude Agent SDK and Claude Managed Agents
11. Google ADK 2.0
12. LangGraph, CrewAI and Microsoft Agent Framework
13. Evaluating Agents
14. Cost and Latency in Multi-Agent Systems
15. Level 4 Capstone: A Research Team of Agents

Plan, deploy, govern and measure your own production agentic system.

1. Architecture of an Agentic System
2. Choosing Your Stack: Hosted, Low-Code or Code
3. Environments and Sandboxes
4. Identity, Credentials and Least Privilege
5. The OWASP Top 10 for Agentic Applications
6. Observability: Seeing What Your Agents Do
7. Cost Modeling and Budget Controls
8. Reliability: Retries, Checkpoints and Idempotency
9. Governance: Policies, Risk Tiers and Regulation
10. Workspace Agents for Teams
11. Deploying and Versioning Agents
12. Maintaining Agents as Platforms Change
13. Measuring the Value of Your Agents
14. The Agentic Operating Model
15. Level 5 Capstone: Build Your Own Agentic System