The idea
One app, all workshop long, and you watch it grow up: one agent does the job → then a team does it better.
You start at "what even is an agent?" and finish with a working multi-agent system — built live, step by step, in the Google Agent Development Kit (ADK), the same framework Google uses to build real agents.
No prior ADK experience. Basic Python is enough.
What you'll learn
Agents — build your first one: a name, a model, an instruction.
Tools in Python — give your agent capabilities with plain functions, and let it choose which to call on its own.
MCP — plug in ready-made tools you never had to write.
Multi-agent systems — Sequential, Parallel, and a Writer + Critic loop.
How we work
Not three straight hours of lecture. We build, then you build.
Week 1 — 3 hours
| 1 hour | building together |
| 30 min | hands-on, on your own |
| 1 hour | levelling up to the next stage |
| 30 min | hands-on again |
Week 2 — Project day
Everyone builds their own project with what they learned, and we review the projects together.
This is for you if…
- You know basic Python and want a proper on-ramp into AI agents.
- You want to see every concept actually run, not just theory.
- You want to walk away with a real project in hand.
Before you join — setup
Nothing complicated. Have these ready before Week 1:
- A laptop (Windows, macOS, or Linux) with internet.
- Python 3.10+ installed.
- A free Google AI Studio API key — no billing, no credit card.
- A code editor (VS Code recommended).
- Basic Python — you can read and write simple functions. No agent or AI experience needed.
A short setup guide is sent after you register, so everyone starts on the same page.
You'll get
- The full code — every agent we build, in your account library
- A reference PDF
- The recording of both sessions, available for 12 months
- Telegram support through Tue 29 September, one week after we finish