Building Blocks of Agentic AI

A hands-on workshop for engineering leaders

This workshop grew out of work building composable agent systems. The core insight: every agentic AI framework — no matter how polished — relies on the same four primitives: a model, a system prompt, tool definitions, and structured outputs.

The workshop makes those primitives tangible. Using the open-source Composable Agent Playground, participants start with a bare model and progressively layer on capabilities, observing how each building block changes behavior in real time. 80 minutes, interactive and hands-on.

The Four Building Blocks

Models & providers — what models do (and don’t), and why provider choice matters less than you’d expect
System prompts — small instruction changes, dramatically different outcomes
Tool/function calling — the mechanism that turns a chatbot into an agent
Structured outputs — schemas that make responses predictable and machine-readable

Who It’s For

Engineering leaders, tech leads, staff+ engineers, and decision-makers evaluating AI adoption. No ML background needed.

Learn More

For the full workshop description, exercises, and philosophy: george.mand.is/workshops/agentic-ai

Get in Touch

Interested in bringing this workshop to your team or event? Reach out at [email protected] or schedule a call.