AI Agents Foundations

What you'll learn

• Explain the agentic loop — call the model, read why it stopped (stop_reason), decide what happens next, repeat — and why control flow is driven by that structured signal rather than by string-matching the reply.
• Describe the role of a coordinator that owns the routing decision (run a tool, hand off to a subagent, or finish) and why centralising that decision is what keeps a growing agent maintainable.
• Judge when a fixed pipeline is the right shape and when a dynamic adaptive workflow earns its extra complexity.
• Design a tool — and an MCP server — with a name, description, and schema clear enough that the model picks it correctly and calls it with the right arguments.