What we covered
Function calling is the point where a model stops producing text and starts doing things. The workshop starts from first principles and ends with schemas that hold up against a model that is trying its best to misread them.
- LLMs explained, at the depth you need to reason about what function calling is doing.
- The Gemini ecosystem, and getting started with the Gemini API.
- Function calling basics: the loop, and what the model is and isn't deciding.
- A run of worked examples, each one adding a case the previous one didn't handle.
- Schema design, which is where most function calling implementations quietly go wrong.
- Getting structured data out, and what to do when the shape comes back not quite right.
- Resources for continuing past the session.
Format
A hands-on workshop, built to be followed along with rather than watched. The example set is deliberately long — the concepts are quick, the edge cases are the session.
Who it's for
Developers writing their first tool-using LLM integration, and anyone whose function calling works in the demo and not in production.